Midwestern Urban Geography and High-Speed Rail

I’ve been uploading videos about high-speed rail lately, of which the most recent, from a week ago, is a return to my attempt at producing a high-speed rail map proposal for the eastern half of the United States. I streamed and then blogged a map here with followup here, but having looked at the model more, I’d like to do a refinement – both to introduce a slightly bigger map and explain why it is so, and talk about the issue of connecting low-speed lines. Along the way, I feel like I must talk about an issue mentioned in comments occasionally about the politics of only connecting major metropolitan areas, especially since this map still has fewer lines than various state wishlists stapled by Amtrak; this is especially important because one of the motivations for this post is a criticism of current plans by Matt Yglesias.

The map

In this map, like some old maps but unlike in last year’s map, red denotes high-speed lines and blue denotes low-speed ones, some running through to the high-speed network and some not

A full-size (6 MB) version of the map can be found here. This is not intended as a comprehensive map of all desirable low-speed connections – I made no effort to include the Northeastern ones, which I wrote about in the context of New England and Upstate New York, and which Ben She has done good work on in the context of eastern Pennsylvania and the mid-Atlantic. Rather, I want to focus on the Midwest.

But first, to explain a little more about why this map includes more red (high-speed) lines than previously, the reason has to do with my spreadsheet for computing ridership density based on Metcalfe’s law. The original posts computed everything by hand, which meant that some low-ridership city pairs I just rounded to zero; the spreadsheet does include them, making every line look much stronger. This, in particular, makes St. Louis-Kansas City and Atlanta-Birmingham, omitted last year, and Nashville-Memphis, suggested last year as a maybe, solid propositions.

A note of caution is still advised. Those weak city pairs that aggregate to sufficient ridership for significant return on investment are often at long distance, such as Kansas City-New York. The ridership model is trained on Shinkansen data out of Tokyo and sanity-checked with some French, German, and Spanish data, but the same model overpredicts Shinkansen ridership on inter-island trips for which planes are a convenient alternative, like Tokyo-Fukuoka or Tokyo-Hakodate. This makes me reluctant to add a Kansas City-Dallas connection, which the spreadsheet thinks generates a bit more than $1 million in annual operating profit per km of new construction: the extra ridership out of Kansas City-Dallas includes some very long-distance trips like Dallas-Detroit, for which the model is likely an overprediction.

The truth is likely between the spreadsheet and the handmade version of the model; while the Shinkansen is not competitive with planes when trains take five hours, European high-speed trains are, for example Paris-Nice. This leads to the inclusion of the new sections, but the exclusion of Kansas City-Dallas. Note also that I did look at Birmingham-New Orleans and Memphis-Little Rock, and both were weak even in the spreadsheet (though I did not attempt Birmingham-New Orleans-Houston) – the Deep South is too low-density and rural to support as expansive a system.

But the topic of this post is not the South, but the Midwest.

Midwestern urban geography

The United States is usually a country of fewer, bigger metropolitan areas, like rich Asia and unlike Europe; unlike both Europe and Asia, American cities are very decentralized, and the exceptions are in the Northeast and West, not the Midwest. In particular, naive comparisons of Midwestern to French high-speed rail corridors are unwarranted: while Chicago is of the same approximate size as Paris, and secondary Midwestern metropolitan areas like Detroit and St. Louis are substantially larger than French ones like Lyon and Marseille, Lyon and Marseille are ringed by many small metropolitan areas with their own TGV service, whereas at the same radius, St. Louis has only its suburbs.

However, this phenomenon of fewer, bigger metro areas has exceptions. Michigan, in particular, has a slightly more European geography. Using the smaller numbers produced by the metropolitan statistical area (MSA) calculation rather than the broader combined statistical area (CSA), Metro Detroit has 43% of Michigan’s population as of the 2020 census. The median Michigander lives in the Grand Rapids MSA, with 1.1 million people, fewer than the US-wide median of 1.6-1.7 million. Michigan is a fairly urban state, and below Grand Rapids is a succession of six-figure metropolitan areas: Ann Arbor, Lansing, Flint, Kalamazoo, Battle Creek, Saginaw.

Ohio is similar to Michigan. Its three main metro areas, excluding Cincinnati’s out-of-state suburbs, have just a hair less than half the state’s population; the median Buckeye therefore lives in Dayton, MSA population 800,000. Moreover, the southern half of Michigan has fairly high population density, as does Ohio – nothing as dense as the Northeast or Germany, but they’re comparable to France.

This geography lends itself to an expansive intercity rail network: the cities are relatively close to one another, and there are many of them meriting a connection. In Ohio, this happens to take the form of an entirely high-speed network, since Cleveland, Columbus, Dayton, and Cincinnati all lie on one line, and then the most natural east-west route between the Northeast and the Midwest passes through Cleveland and Toledo. Ohio, in this case, is a state with fairly good geography for low-speed intercity rail that just happens to also have good geography for high-speed rail due to its location. Michigan, in contrast, is not on the way between much, and thus should get a low-speed rail network, including both connections to Chicago (such as to Grand Rapids) and an intra-state network.

Wisconsin has many, smaller cities as well: the median resident is in an MSA of around 200,000 people, currently Racine. Fortunately, many of those cities lie on just one line between Chicago and Minneapolis, plus a low-speed branch up to Green Bay. Unfortunately, coverage is lacking by the standards of Ohio, Michigan, or much more big city-dominated Illinois and Minnesota.

Getting low-speed rail right

I am happy to report that in Michigan and Ohio at least, good projects for low-speed rail are pursued. When I streamed my video, I was told in Twitch chat that Michigan is looking into funding a Detroit-Lansing-Grand Rapids intercity train. Ohio likewise has long had ideas for a statewide network, beginning with the Cleveland-Cincinnati spine.

It is unfortunate that these projects are not planned well. In a future post, I should write more about the concept of the wrong project versus the right project done poorly; I obliquely pointed this out when writing about leakage in the context of urban transit, where some American cities have poor project prioritization (such as Los Angeles) whereas others choose more or less the right projects but execute them poorly (such as New York and San Francisco). In this schema, the current plans for low-speed rail are often the right project, done wrong.

What I mean by this is that there’s a set of best industry practices for getting low-speed (that is, legacy) rail right, emanating out of Germany and surrounding countries, especially Switzerland. These include,

  • Integration of timetable planning and infrastructure, to minimize construction costs – if higher costs are acceptable, just build high-speed rail.
  • A clockface all-day schedule with a minimum of one train every two hours, and ideally a train every hour if the distances are shorter or the cities are bigger.
  • Timed connections at major nodes to both other intercity trains within the same network and regional mass transit (such as regional trains or connecting buses).
  • Reliability-centric planning, in which sources of delays are to be proactively eliminated – on a system this tightly coordinated, delays on one line propagate across the entire system.
  • An average speed of around 100-130 km/h – the higher numbers are more appropriate in flat terrain.

Marco Chitti has an excellent post that I must revisit in the future giving more detailed guidelines, mostly at regional level but also with an eye toward national intercity rail planning.

The upshot is that trying to incrementally build up ridership for a few trains per day does not work. The US has a few trains per day on a few corridors now, such as Chicago-Detroit, and daily trains on most others, and this hasn’t built up ridership. The low-speed, low-frequency intercity trains Europe had before the introduction of the TGV in France and the high-frequency, tightly-linked InterCity network in Germany were rapidly losing market share to cars and planes. To build such a network now would be like to build infrastructures wired telephones in a developing country rather than just skipping straight to cellphones as developing countries have.

Politics and Matt Yglesias’s post

Matt likes pointing out that current transportation plans in the United States are deficient, and to link to my posts as a better alternative. There was a lot of dunking on Matt’s post about this (as there is on anything that Matt writes) by left-identified people who, in effect, think daily or twice-daily trains between small cities are a great investment. This dunking usually takes the form of “how dare Matt, a lifelong East Coaster, tell people in [insert Midwestern town] that they don’t deserve trains?”. In a less abrasive form, some people in comments here, like Pharisee, have made the point that these expansive maps proposing daily trains to many places have good geographic coverage whereas what I propose does not. Let me explain why this line of thinking is wrong.

The issue is that the United States is, again, a country that for the most part has fewer, larger metropolitan areas than Europe. The map I made above hits MSAs with a large majority of the country’s population. Of the top 50 as of the 2020 census, the only misses are Denver, Oklahoma City, Salt Lake City, and New Orleans. Denver and Salt Lake City are far from everything else, and the other two are in theory on routes from Texas to the rest of the country but there’s too little population on the way for a connection in the geography of 2022.

Moreover, within the Midwest, coverage is ample. The Midwest is a highly (sub-)urbanized region, much of which has fairly high population density, and the biggest exception to the high density, Minnesota, has a large enough city to justify a line to Chicago by itself (and Milwaukee is on the way, too). The areas that are usually most moralized as Real America – Michigan, Ohio, Pennsylvania – are on the way. This shouldn’t be too surprising: the Real America moralization centers areas with a past industrial history, evoking feelings of nostalgia for midcentury American industrial dominance, and those areas remain major cities today, just relatively poorer than they were in the 1960s. This way, in the Midwest, every state has a large majority of its population in an MSA with a high- or low-speed train station on my map, except Iowa, which is unusually rural.

This is not even out of any consciously political desire to serve these areas. I draw maps out of a ridership model. It just so happens that metropolitan areas of 4 million people situated in dense geographies scream “build high-speed trains to me,” and those include Detroit.

The problem – the reason Matt is so negative on current plans – is that current plans are bad. They are low- and not high-speed rail, which by itself is not horrible, but they’re also bad low-speed rail. Daily trains are just not good. But this does not mean the Midwest can’t or shouldn’t get a good intercity rail network: it should, combining high- and low-speed trains as appropriate.


  1. Alon Levy

    By the way: the map doesn’t make it clear, but the reason for the double stations in Lafayette and Bloomington is that they should probably not get city center stations for reasons of noise, level crossings, and urban slowdowns; instead they should get stations tangent to the built-up area, and then the orbital line can do double-duty as connecting transit to their centers (in addition to its primary duty of connecting Peoria and Urbana-Champaign to the high-speed rail network).

    • Phake Nick

      Would it be possible to extend the Lafayette-Bloomington line to Madison on the west and South Bend on the east to form a complete grand rail ring around Chicago, in order to activate transit usage across the entire region? Even tho the nature of speed difference probably mean stations on the ring can still get to each other faster through Chicago than to travel on such ring.

    • adirondacker12800

      You are mindlessly connecting dots again. There aren’t enough people in metro Bloomington to have two stations. People in Peoria get Japanese levels of demand to use high speed rail you are talking about a 50 passenger bus load once an hour during the day. They all own cars and will just drive to the HSR station.

      • Alon Levy

        On an essentially regional train, in a city where little through-traffic is expected (people from both Peoria and Champaign are probably transferring to HSR there rather than traveling Peoria-Champaign), it’s fine to have multiple stops, one for the HSR transfer and one for the historic town center.

        And at Japanese demand, Peoria-Chicago is 900,000 passengers a year, so 1,250 a day per direction; Champaign-Chicago is another 570,000, so a bit under 800 a day per direction. (This is first-order and I think Champaign generates more travel relative to population because of UIUC.) It’s a two-car EMU every hour, maybe every half hour if there’s significant commuter travel, which the model doesn’t take into account and therefore underpredicts short-distance ridership. And this does take the “everyone owns cars” line into account – in secondary Japanese cities like Sendai or Morioka car ownership is high; I worry about the model’s applicability to connections between two no-transit cities, like Indianapolis-Nashville or Pittsburgh-Detroit, but the numbers from secondary Midwestern cities to Chicago should be taken as solid.

        • Phake Nick

          On a low speed regional line there are no need to limit the number of stations to only what can be drawn on a continential map. It is possible to have countless number of simple haltepunkt serving even small villages or city districts along the track for non-express type service.


          Per Japanese practive, a place with demand of about 1000 passenger per day mean 1 train every 2 or 3 or 4 hours and future of rail service to the place will be questioned, and are also area where companies like JR East would be looking for devestment like single tracking and deelectrification. Izumo area maintain roughly 1 intercity express every hour and 1 local train every hour off-peak, but that’s with for example 5000 people using Izumo-shi station every day and there are still other stations in the area generating ridership for the same trains. Shinjo station on Yamagata Shinkansen which have a daily combined ridership of 2000, get an intercity express train every 2 hours and a local train every ~1.5-2.5 hours. [Although on the other hand the city of Shinjo is only a city of 33k people.]

          In these places where the society is completely motorized, people do not really use train for their daily trips. Most destinations are outside the reach of train stations too. Majority of demand are during commuting time for school-going students and workers who happen to have their trip aligned with rail service. In this sense the North American sense of Commuter Railway can already cover most of the demand. But Japan have no NA level freight railway industry and thus passenger railway still have to own the track and pay for the full track maintenance cost even if the service have already been draw down to 3x daily, one at morning one at noon and one at night.

          Japanese towns and cities even if small still want to keep rail service in general, but the main purpose is to serve as a symbol and gateway of the town and to attract tourists. Most of such line already have such a low demand, coverage, frequency that honestly buses would better cover the need of people who use transit, both local and intercity so that doesn’t constitute a real reason to keep the rail service in these places. There are also cases where cities want to maintain a train simply because they don’t want to pay for operating cost of alternative bus service. But when they are being asked to pay for the train, even Hakodate a city with population of 260k now is now being forced into consideration of terminating its local rail service.

          • Matthew Hutton

            Interesting that Japan is so different.

            I would call a station with 400k passengers a year one with good ridership. It’s the ones with 40k passengers a year or fewer that I would question.

          • John D.

            “I would call a station with 400k passengers a year one with good ridership.”

            If I’m not mistaken, the level for closure/divestment that Phake Nick mentioned is a transport density (輸送密度) of 1,000/day (or rather, 1,000 passenger-km per route-km per day), which is not the same as total ridership or per-station traffic of 1,000/day.

            For example, the mountain section of the Senzan Line between Ayashi and Uzen-Chitose had a transport density of 1,917/day in fiscal 2020, but multiple stations with less than 100 entries per day.

      • adirondacker12800

        People who can afford to flit off to Chicago or St. Louis are rich enough to own cars. You are imagining things from your view across Ninth Ave. They can put the enormous parking lots out on I-55 which keeps the parking dessert and traffic out of downtown, such as it is.

  2. John Mellor

    Note correct spelling of geography. 





    div dir=”ltr”>Sent from my mobi

  3. Matthew Hutton

    I wonder if one of the reasons the model gets the passenger numbers for longer trips wrong is partly because the fares are wrong.

    For example if in France the costs are (say) 2¢ per passenger km incremental costs for an extra train, 3¢ per passenger km fixed costs and 2¢ per passenger km capital costs. That would allow you to presumably sensibly run a London-Barcelona train for something like 3-4¢ per passenger km – as it’d be an extra service on top of what the line was designed for. And you’d probably get more passengers on such a train if it was directly price competitive with flying which at that price it might be even given the longer route.

    Similarly if you were to run a train from (say) Paris to Chamonix it probably makes sense to price the high speed section of that at maybe 3-4¢ per passenger km and to hand most of the profits to the regional line at the end – without which the passenger would probably drive the whole way anyway due to convenience.

    Obviously that means the core trains – in this case the Paris Lyon high speed ones – need to be more expensive – but I think that’s ok.

    • Alon Levy

      Hmmm. Degressive fares should make ridership decay more slowly with distance, whereas Shinkansen data shows ridership decay faster past about four hours out of Tokyo. The issue here is that airfares are insensitive to distance at this scale, but that should already be priced into the model. It might just be a quirk of Japanese inter-island air travel, I’m not even sure…

      • Matthew Hutton

        Some of the Japan thing is probably that Kansai and Fukuoka airports have more flights so if you went just to Hiroshima you might as well get the train the whole way back to Tokyo.

        Also if we follow my pricing model and say that Osaka-Tokyo “should” be 10¢ per passenger km that would point at a price of ¥7250 vs the actual fare of ¥14700. Whereas based on my pricing “model” for Tokyo to Fukuoka the fare “should” be only ¥6000 vs the actual fare of ¥23390.

        In the real world if ¥14700 is reasonable for Tokyo to Osaka then the Tokyo to Fukuoka fare should be at worst be something like ¥16000.

        • Marc

          Shinkansen fares are costly, fairly rigid and seemingly target first class and business travel between city pairs up to 2 hours apart. If one can pre-plan for even a few days, discounted non-peak economy airline tickets can be less than half the cost of Shinkansen fare for Tokyo to Sapporo (still 8 hours via Shinkansen and express train), Fukuoka (5 hours via Shinkansen), Hiroshima (4 hours), or even Osaka (2.5 hours).

        • Phake Nick

          Tokaido Shinkansen “only” have ~50% profit margin. If you lower the fare to less than half then they would be losing money. And if the passenger take even longer trip, then JR Central will need to share the revenue to other JR companies based on the distance travelled by the passenger on the track of each different JR companies, making cost of transporting them unable to be recouped.

          16000 from Tokyo to Fukuoka would still mean less than 8000 for JR Central and they will be losing money from these passengers.

          It is not that they can attract any more passengers given the line is already at capacity.

          • Matthew Hutton

            To be fair though I doubt there’s loads of overlap between trains from Osaka and trains from Fukouka. The first train from Fukouka leaves at 6am and arrives in Tokyo at 11am. Even if there was an earlier one leaving at 5am it still wouldn’t get to Tokyo until 10am so wouldn’t clash with the morning peak.

            Departing in the evening the last train leaves at 7pm and arrives at midnight. But that is pretty late. I’d have thought most people would leave Tokyo at 4-5pm at the latest if they wanted to go to Fukuoka.

          • Phake Nick

            There are legal cap against Shinkansen operation hour between 0-6am due to old NIMBY noise complains, if they need to operate in these hours then each trains need to individually notify all local governments along the track and they must operate at lower speed limit.

            And trains on less desirable hours are already those that more likely to receive discount offer.

  4. Martin

    I am curious why the map does not have a true Texas triangle. It seems like it would probably outperform nearly everything on this map, and the geography also should mean lower construction costs.

    Given the low ridership on the lines (and consequently the opportunity to squeeze in some marginally slower high-speed trains, which 4-track at the stations, on top of the pure intercity trains), and the decentralized structure of US cities, I presume a lot of energy should go towards each city having 4-5 major suburban stops at most major metropolitan areas, for many (half?) of the trains to stop at. Such trips are a large share of for example the Taiwanese HSR, Chinese HSR (which has less major suburban stops, but tons of stops in between large cities), or the German HSR. I think HSR in the US would have to be combined with such regional traffic to be viable (and the demand would mostly exist in the dense larger urban conglomerates, not far away between the cities).

    • Alon Levy

      Ooh, good questions.

      Re the Texas Triangle, the issues there are,

      1. The cities are huge, but they only connect to one another (NOLA and OKC are too small and far away to be part of this network).

      2. The cities are incredibly decentralized, so I’m worried about underperforming the model due to the malus of point-to-point travel.

      3. The terrain is flat, but the Midwest is even flatter, and Texas Central made formal and informal political promises to farmers based on JR Central’s construction techniques that lock in high costs, with unnecessary viaducts rather than at-grade construction.

      Re suburban stops: the problem with American decentralization is that it’s rarely that polycentric. California HSR was planning on a bunch of stops like this around Los Angeles, like Burbank and Norwalk, but none of them is much of a sub-center – the region is far too diffuse.

      • adirondacker12800

        Suburban stations means the ginormous parking garage can be near the limited access highway. That keeps the ginormous parking garage and automobile traffic out of downtown…. People aren’t walking to BWI….

      • Martin

        Regarding 1 it is kind of true, with the (very large!) exception of Austin. Still the regions are very large.

        Regarding 2, and the broader point on suburban stations it seems as that in China and Taiwan, suburban or regional train stops a bit outside the cities support massive ridership. These are often outside urban areas for right of way reasons, but still seem to attract all ridership within very large uptake areas without local density. Taiwan HSR us basically just a dotted lines of such stations, supporting dense semi urban areas dominated by scooters and cars. Not everyone has to park, they can bus interchanges and many people are dropped of by other riders (Chinese and Taiwanese stops lack parking). Presumably self-driving cars will also interact very well with this model as well.

        Same model should work in US suburbia I think, in particular in 5 million plus metro areas. One will probably never attract someone living 20 km north of downtown Houston to a city station, but maybe a suburban stop.

        In the 25+ years longer perspective it will of course also shape the regional urban patterns (in particular in looser zooned parts such as Texas).

        • Eric2

          I wonder if that is due to false stereotypes of how Americans “have to” travel. Or because they think policymakers demand it. Because I think neither project would be less viable than it is if the parking were replaced by real developent.

          • Phake Nick

            JR Central is doing similar in Japanese project too, their Chuo Shinkansen non-urban stations are completely designed with only car user in mind. Opted against connecting with nearby railway, located specifically near key highway entrance, and let local government figure out maybe they would want to provide some sort of transit connection on their own. Specifically Yamanashi prefecture in Japan tried to ask JR Central re-evaluate their decision to not co-locate Chuo Shinkansen station with existing rail station, yet after redoing study of benefits it cannot turn JR Central’s mind. And now the local government there is weighing to build a BRT to the maglev station on themselves

            As for parking vs development, I recall at least for Texas Central project the station site selection also have development potential in mind, but really not sure how that would interact with the large amount of parking they plan to provide. Maybe it’s like stations like Gifu Hashima on Tokaido Shinkansen, where there are some hotels and car rental around but not much beyond that?

          • Matthew Hutton

            The Chiltern line which is one of Britain’s most successful railways in terms of increased passenger numbers has a whole series of parkway stations at Oxford, Warwick and Haddenham/Thame.

          • Frederick

            The overwhelming majority of Chuo Shinkansen passengers will be travelling between Tokyo and Nagoya. Every station in between is just an Haute-Picardie beet field.

      • Onux

        I fail to see how DFW (7.7M) to OKC (1.4M) at 207 mi (233mi through Ft. Worth proper) is “too small and far away” when you include Nashville (2M) to Memphis (1.3M) at 212 mi and STL (2.8M) to KC (2.2M) at 248 mi. Birmingham has 1.1M, Louisville has 1.3M, VA Beach is slightly larger at 1.8M but is 244mi from DC via Richmond.

        • Alon Levy

          Hampton Roads is the easiest. It’s not especially big, but it is within HSR distance of the entire Northeast; New York-Norfolk is around 710 km, so a bit more than three hours. The entire Southeast crescent from Washington to Atlanta is really strong, because Richmond is really strong since it’s 180 km of construction and 400 km of revenue to Philadelphia and 540 to New York, then Norfolk and Raleigh are really strong, then Charlotte, then Atlanta.

          The others have a weaker Metcalfe’s law accretion, but still something – Kansas City just to St. Louis and Chicago is too weak to put on the map, but when you count lines to Milwaukee, Detroit, Cleveland, etc., it starts looking more interesting. Memphis, same thing – it’s not a lot of construction to Nashville compared with the revenue you get to Atlanta and Chicago. OKC doesn’t have that – Texas is big and has a lot of people but not-Texas is bigger and has even more.

          • Phake Nick

            Beyond connection to faraway city is something I really question. Like I just checked the table of ridership prediction you uploaded in 2021tor try to validate against real Shinkansen ridership. I checked Hakodate station which have a daily ridership of 1300 in real life, or 0.47 million a year. The table however predict 1.82 million annual ridership from Hakodate to rest of the Tohoku Shinkansen, involving 1.08 million for Tokyo and 280k to Sendai and 460k to rest of the line. And it also predicted a 370k ridership to Nagoya and Osaka combined, and 270k to rest of Tokaido-Sanyo-Kyushu Shinkansen. Even if the forecast from Hakodate to rest of Tohoku Shinkansen can be said as within order of magnitude of original forecast and thus not completely wrong, I don’t think it is rational to believe 25% Shinkansen rider from Hakodate are riding pass Tokyo to destination beyond, even if that include Osaka and Nagoya.


            This data indicate breakdown of passenger volume by share between regions, as of FY2019. For data on JR trains, from Southern Hokkaido (which is essentially just Hakodate) to Tokyo+Chiba+Saitama+Kanagawa the number of riders are 183k. To Miyagi where Sendai locate it is 77k. Combined share to other prefectures are along Tohoku Shinkansen is 344k, in which 261k is Aomori prefecture and 83k is the rest. Note that this data include not just high speed rail but also other ways of using rail, for example leisure cruise train.

            From Southern Hokkaido to Aichi (Nagoya) the number of people is 2.8k. It is also 2.8k in total for Osaka+Kyoto+Kobe. For rest of Tokaido+Sanyo+Kyushu Shinkansen, there are also 3.2k for Shizuoka, 0.1k for Okayama and 0.2k for Hiroshima. All other prefectures along the line registered 0.0k including entire Kyushu, although there are also through passengers like 0.1k between Fukuoka and Central Hokkaido, which mean there are 100 people in the entire year taking the 14 hours train between the two cities with ~1-2million population, and 0.2k people between Kagoshima and Northern Hokkaido, which most likely mean there are 200 people in the year completed the challenge of moving across mainland Japan from tip to tip using mainly rail.


            In other words, the predicted vs actual ridership to Hakodate is roughly as follow:
            Tokyo: 1080k/183k
            Sendai: 280k/77k
            Aomori: 110k/261k
            Rest of Tohoku Shinkansen: 350k/83k
            Aichi (including Nagoya and Toyohashi): 180k/2.8k
            Osaka: 210k/2.8k
            Shizuoka: 90k/3.2k
            Okayama: 23k/0.1k
            Hiroshima (including Fukuyama): 37k/0.2k
            Rest of Tokaido-Sanyo-Kyushu Shinkansen: 95k/0.0k


            This mean the formula have order of magnitude error even when predicting ridership to large cities with 10+ million population. And that is with there is no road for car to drive to Hakodate.

          • Alon Levy

            Yeah, the inter-island numbers are very wrong, as are the through-Tokyo ones. I’m actually surprised at how much extra Hakodate-Aomori ridership there is, but I guess that at that range the main competition to the train is the car and that is not available inter-island.

          • Matthew Hutton

            I assume the Japanese ridership figures for long distance travel are low because flying is a lot cheaper.

            I mean you can do Paris-Barcelona for €39 each way on the train vs something like €150 each way in Japan and there’s few discounts.

            Of course in Europe the Eurostar is expensive and interchanges in France and London are inconvenient. But if you could avoid those issues I think you’d do OK for longer journeys.

          • Marc

            I mentioned this earlier, but perhaps not explicitly enough. Using the Shinkansen as a ridership model will be inherently skewed, as the fares are far too high for routine use by many Japanese residents. Planes, limited express trains (legacy speed), ferries, overnight buses are the less expensive options for longer distance inter-city travel. I’ve used the Shinkansen to travel from Hakodate to Tokyo and from there to Fukuoka, but only because I had a JR Railpass. That option isn’t available to those in Japan, the discounted holiday passes they can get don’t cover the Shinkansen, so almost no one except foreign tourists would make those kinds of trips.

          • Phake Nick

            @Marc I checked 2009 version of the same data source, aka before Shinkansen extend to Hakodate, and find the data to Southern Hokkaido via rail was 1.1k for Aichi and 2.5k for Kyoto and Osaka and Kobe. Which mean there are still more people attracted to travel via Shinkansen to Hakodate than the cheaper legacy night train. (Although the number of people from Kansai and Aichi who take the train to Central Hokkaido in 2009 data have virtually vanished in 2019 data – 20k in 2009 vs 2k in 2019 – compared to air travel to Central Hokkaido – 1600k in 2009 and 2100k in 2019)

          • Matthew Hutton

            I looked up the Paris-Barcelona prices for the Valentines weekend to leave Friday afternoon and come back Wednesday 15th.

            Difficult to do an exact comparison as the flights quoted directly on kayak don’t include checked bags. However the 5pm flight with checked bag and seat choice is €420 return with Vueling for a couple vs €360 return on the train at 3pm.

          • Phake Nick

            @Alon it’s having pretty wrong number even for city pairs that aren’t crossing Tokyo and aren’t inter-island. Like from Hiroshima (including Fukuyama) to Kansai(Osaka+Kyoto+Kobe) or Aichi (Nagoya+Toyohashi), the model predicted 8.5M for Hiroshima-Kansai and 5.3M for Hiroshima-Aichi, however 2019 data is 3.3M for Hiroshima-Kansai and 0.5M for Hiroshima-Aichi.

          • Matthew Hutton

            Even Eurostar to Amsterdam from London for the same dates and times is £435 vs £350 to fly with easyJet from Luton plus the train return to central London.

          • Phake Nick

            Data of 2019 JR trip between prefectures are as follow:
            Hiroshima: Kobe 751.2k Osaka 2121.7k Kyoto 447.9k Shiga 53.0k Gifu 43.5k Aichi 522.1k Shizuoka 134.7k Kanagawa 368.0k Tokyo 1405.7k Saitama 37.5k Tochigi 22.4k Gunma 11.3k
            Aichi: Okayama 338.0k Hiroshima 503.5k Yamaguchi 128.8k Fukuoka 585.3k Saga 19.8k Kumamoto 39.2k Kagoshima 20.7k
            I cannot come to a conclusion just by these data, but my feeling is that it have to do with regional center. Like people usually move long distance to look for oppoturnities, no matter it is for business, leisure, making friends, hobby, or such. And Osaka being a large city and acting as a regional center provide more oppoturnity in all these aspects than all cities beyond it before Tokyo. Thus people have little need to travel from Hiroshima to destination beyond Osaka when their needs and wants can be satisifed within Osaka region which is shorter distance away from Hiroshima. For example if you are from Hiroshima and you want to attend university, you could search in neighboring prefectures, you could search in Kansai area, but chances are few that you would go all the way to Nagoya for universities where there aren’t much particular attractive that you can’t find among the choices in Kansai.

            This could also explain why inter-island travel demand are less than anticipation, as regional centers are Fukuoka and Sapporo on the two islands respectively, and it can also explain why forecast of data across Tokyo are wrong, because Tokyo have most everything one could want or need and can access inside Japan that one have few reason to go past Tokyo for a place even further away.

            If this also hold true for the US then that mean forecasted demand for inter-region travel between Mid-West, Northeast, and Southeast might be less than what the formula say. It can also be checked by using Northeast Corridor data and see the amount of passengers crossing NYC. In Europe it could mean that even if France connect its TGV lines through Paris, improve service across national borders, and lower the price of trains across Eurotunnel, they might still earn less ridership than what one might otherwise expect.

          • Onux

            Your explanation doesn’t hold up empirically. Setting aside Hampton Rhodes it is simply not true that places like Memphis and KC can access more people in the Midwest/South than OKC could in Texas, give the same distance.

            OKC to Houston is 445 mi, to SA is 467. In this distance an OKC resident can reach 21.7M other people. From KC, using you map, just getting to Chi is 510 mi, and only accesses 13.3M. Adding Milwaukee and Indy gets to 17.8M, but Mil is 600 mi from KC and Indy 700 – ridership at this distance should be dismal compared to Dallas 233 mi from OKC.

            Memphis is worse. Within 452 mi (distance to ATL) you only reach metros totaling 11.5M (inc. Nash, Chattanooga, Louisville. At 497 mi (stilll farther than Hou or SA) you can get that to 16.3M with Indy and Cincy, far less than OKC to TX.

            Detroit? Cleveland? You can’t possibly think that cities 800+ mi from KC or Memphis are contributing meaningful HSR ridership.

            If your model says they do, but doesn’t say the same for OKC, then together with Phake Nick’s Japan analysis I would say your model needs some careful study and revision.

            “not-Texas is bigger and has even more.”
            This is perhaps part of your issue, because it isn’t really true, and certainly not for an equal area analysis. TX is closer in size to California than it is to NY, and yet no one seems to say either of those states is too small to anchor HSR networks beyond their borders. Sure, you can aggregate states to get a higher pop, but the TX Triangle has about the same population as Ohio+Michigan, or more than Illinois + Indiana, but the Triangle doesn’t have that population spread over as large and area as those states.

          • Alon Levy

            New York is adjacent to way bigger and denser things than Oklahoma, is the difference.

            As for Memphis: it’s around 4-4.5 hours from Chicago, which is within the model’s validity range albeit at its upper end.

          • adirondacker12800

            Kansas City to Indianapolis is, according to Google maps, 482 miles, on I-70.
            It makes more sense to build St. Louis to Indianapolis than it does to build St. Louis to Kansas City. Anything east of Indianapolis is “free” and serves more people than going to Oklahoma City. More people than going to all of Oklahoma.

          • Onux

            “New York is adjacent to way bigger and denser things than Oklahoma”
            The issue isn’t will the NEC generate more ridership than TX (it, will, bigger and big cities closer together) but is TX big enough to anchor lines to other metros, such as OKC. TX is.
            Also, as I noted above, KC and Memphis are nowhere near close to anything the size of NY, yet you have lines of equal or longer length to them but not OKC. If your model says this is because KC-Cleve or Mem-Cincy ridership will be huge but OKC-Dallas not, then your model needs some investigation.

            “As for Memphis: it’s around 4-4.5 hours from Chicago”
            So Memphis is now 4.5 hr from about 26.3M, while at those speeds OKC is 2.5-3 hr from 21.7M, and just 90 min from Dallas, which is quick enough to take a huge bite from the drive mode share not just air. Empirically, at 4+ hr air is dominant to HSR, but at 3 hr or fewer HSR dominates. Would you rather have 10-30% mode of the 10M people in Chi or 100% mode of the 15.5M people in Dall/Hou? Don’t forget that your rail route Mem-Chi is not direct, it is a shorter straightline distance to fly or drive, hurting mode share more. And OKC is slightly larger than Memphis to being with!

          • Alon Levy

            At four hours, the majority of people on Paris-Toulon are using the TGV; Tokyo-Hiroshima has a majority for the Shinkansen as well. The three-hour rule is just SNCF airline brain bullshit that isn’t even supported by SNCF’s own ridership numbers.

            And the bigger issue is the piggyback effect. Memphis generates all this revenue out of just one extension out of Nashville, creating a lot of p-km of revenue relative to new route-length. OKC awkwardly generates less revenue because it’s too close to Dallas, is the frustrating part of this.

          • AJ

            I don’t understand that last part … Memphis is an extension from Nashville, wouldn’t OKC be an extension from DFW? Dallas may have the bigger downtown, but DFW is one big blob so it shouldn’t matter if the train is piggybacking from Dallas or Forth Worth? (and either way there is likely to be a station in Denton before the train hits cruising speed outside the Metroplex). Plus, if you’ve got regular service coming from SA/Austin & Houston, one service can go to FW and the other to OKC, for an easily understood takt pattern.

            You have the Metroplex as the end of a Texas network, but would it be better to view it as a center node of a network that extends beyond Texas? I’d run HSR north-south and hit DFW (for suburban access) and Denton in addition to downtown Dallas, probably a station somewhere south of Dallas for suburban access (Waxahachie?) and let Forth Worth be connected to the regional HSR by simply upgrading the TRE with double-tracking and electrification and run 15 minute all day frequency. Downtown Fort Worth is an important node in the context of an transit within the Metroplex, but I’m not convinced it’s distinguishable enough from the broader Metroplex to serve it’s it’s not ‘on the way’ of the north/south HSR axis.

          • adirondacker12800

            Getting from Saint Louis to Indianapolis through Chicago is like getting from Albany to Boston through New York. Which is what Amtrak’s trip planner will do because the Late For Sure Limited is so execrably slow between Albany and Boston that it can be faster. In round numbers Saint Louis is as far from Kansas City as it is from Indianapolis. From Kansas City you can get approximately nowhere. From Indianapolis you can get to most of Ohio on tracks that will be there for other reasons.

          • Onux

            I agree with Adirondacker than StL to Indy might be better, since ridership from the 6.3M people in Indy/Louis/Cincy(plus Columbus?) would probably exceed what you get from the 2.5M in KC, even including KC-Chi traffic. Maybe not. StL-Chi-Indy is of course not practical.

        • Onux

          I’ve seen that Tok-Hiro is 40-60% rail at 4 hr, Tok-Okayama the same at 3.5 hr, Paris-Marseille 66% rail at 3.5 hr. Chicago is 4.5 from Memphis and ***is not the shortest path so will underperform compared to other 4.5 hr routes that are the same distance as the drive/fly option.*** If you can do the 686 mi from Mem-Chi in 4.5 hr you can do OKC-Hou/SA in 3 hr, with Dallas closer, at at <3hr HSR all over the world dominates with 80% plus of the ridership. 66% of 10M is 6.6M (Chi – best case), 80% of 9.9M is 7.92M (Hou + SA) so TX wins over some long ride from Mem to Chi and Dallas + Austin certain draws more riders than Mem to places like Atl, or Cincy, which are already farther away than Hou/SA are.

          Also you started out with OKC being “too small and far away” but now it is too close? OKC-Dall is 233 mi through Ft. Worth, DC is 226 mi from NYC, is DC too close for HSR service? OKC is the same size as Memphis (technically 100k larger) and within optimal HSR distance (2-4 hr) of all of the Texas Triangle, but Memphis does better because it has destinations outside of optimal range where flying is more competitive? This makes no sense.

          I suddenly realize the importance of your "too close" comment – your model has an issue with how it treats shorter direct travel vs longer travel with possible connections. This came up last time when your first map made Atl to Jacksonville look like a great line, but Portland-Sea-Vancouver didn't at the same distance. Now Memphis is somehow better than OKC despite similar (or better for OKC) conditions – except for VERY long distance connections. Phake Nick's analysis of the Japanese numbers vs your model confirms it, where the only travel your model underestimates is the closest, overestimation for other pairs gets higher as distance increases, and overestimation for very long connections can exceed two orders of magnitude. These very long connections are exactly what is making things like KC and Memphis look good, and so how the model treats them should be reconsidered.

          • Phake Nick

            Tokyo to Hiroshima is 4 hours by rail and ~50% modal share, yet on other Shinkansen trips with 3-4 hours by rail, Shinkansen market share are: (only count against planes, other modes not included)
            Tokyo-Hakodate (4 hours) 25.7%
            Nagoya-Sendai (3.5 hours) 56.6%
            Nagoya-Fukuoka (3.5 hours) 52.9%
            Nagoya-Kumamoto (4 hours) 27.8%
            Osaka-Kagoshima (4 hours) 25.6%
            Osaka-Sendai (4.5 hours) 14.4%

          • Frederick

            Osaka Itami Airport is just 15 km away from Osaka-Umeda while Chubu Airport is much farther away from Nagoya. That’s why Nagoya people are less likely to fly.

            The lesson here is that modal share is not only determined by how good your high speed rail is, but also by how good your air and car infrastructure is.

            TGV has high modal share not only because TGV is good, but because France has anti-car and anti-air policies.

          • John D.

            @Phake Nick

            Those statistics make sense. Tokyo-Hiroshima is a one-seat trip with frequent departures, and competes with an inconvenient airport at the Hiroshima end. Of the listed comparisons:

            – Nagoya-Fukuoka has direct and frequent trains, but suffers from the inter-island travel effect that Alon has mentioned elsewhere, as well as the prime location of Fukuoka Airport.

            – Osaka-Kagoshima and Tokyo-Hakodate are in the same inter-island boat, with the added challenges of their direct trains being infrequent (Mizuho), seemingly slow (Sakura), or both (Hokkaido-bound Hayabusa).

            – The other city pairs are intra-Honshu but require a transfer, which likely weighs on passenger decision-making. I suppose Nagoya-Sendai stands out because it’s a somewhat shorter journey than the rest.

            If we’re going to apply this data to make forecasts about the US, all these ancillary factors are worth considering.

          • John D.


            “Osaka Itami Airport is just 15 km away from Osaka-Umeda while Chubu Airport is much farther away from Nagoya. That’s why Nagoya people are less likely to fly.”

            I don’t think airport access accounts for the difference. Umeda to Itami (by Hankyu and Monorail) takes about half an hour. Nagoya to Chubu Centrair (by Meitetsu) is also around 30 minutes.

          • Phake Nick

            @John D.
            But what is this “inter island travel effect”.
            Note that we are now talking about market share instead of absolute quantity of passenger volume, hence it is not a demand problem of cultural/metro region or something like that.
            Also Fukuoka station on Sanyo Shinkansen is still operated by JR West, so it isn’t an inter-company fare scale problem either. And thus it is also not a problem of whether direct trains exists
            And Honshu to Fukuoka is not exactly too wide, the Kitakyushu urban area extend across two sides of it with boats and bridges and even walking tunnels connecting them together

          • John D.

            @Phake Nick

            “what is this ‘inter island travel effect’. Note that we are now talking about market share [versus planes]”

            I think part of it might be psychological. Getting from Honshu to Kyushu or Hokkaido is imagined as ‘longer’ than the objective distance/travel time by Shinkansen due to the strait-crossing element, which increases the perceived necessity of flying.

            “And Honshu to Fukuoka is not exactly too wide, the Kitakyushu urban area extend across two sides of it with boats and bridges and even walking tunnels connecting them together”

            Sticking to the theme of psychology, how a Shimonoseki resident perceives distance to Kitakyushu (visible across the water with all those heavily used short-distance links) would naturally differ from how an Aichi resident gauges their distance to Fukuoka (which relies on a mental picture).

            “And thus it is also not a problem of whether direct trains exists”

            A one-seat journey is always going to be more appealing than one involving transfers, regardless of how transfer-tolerant your passenger base might be or how well you arrange connections. If memory serves, a Nagasaki newspaper poll some time ago showed lower enthusiasm for the Shinkansen-Relay service compared to the existing one-seat limited express, in spite of the travel time reduction, with many respondents indicating they might switch to the direct bus instead. It’s not a stretch to see a similar penalty applying for trips through Hakata or Tokyo, especially when those are often more demanding transfers than Takeo-Onsen’s timed cross-platform setup.

          • Frederick


            “What kind of anti-car and anti-air policies are you thinking of? The Riviera is very auto-friendly.”

            Green tax, fuel tax, road tolls etc.


            “I don’t think airport access accounts for the difference. Umeda to Itami (by Hankyu and Monorail) takes about half an hour. Nagoya to Chubu Centrair (by Meitetsu) is also around 30 minutes.”

            I should have phrased it differently. What I mean is, for nearly all of Aichi, Chubu Airport is 30 minutes farther away than JR-Nagoya, because of the remoteness of Chita Peninsula.

            Meanwhile, a bus from Namba or Tennoji only needs 30 minutes to Itami, while taking the train to Shin-Osaka will take 25 minutes, just 5 minutes faster. For many places in northern Osaka, it is also as easy to reach Itami as Shin-Osaka.

          • Onux


            “ “What kind of anti-car and anti-air policies are you thinking of? The Riviera is very auto-friendly.”

            Green tax, fuel tax, road tolls etc. “

            Japan has the same anti-car policies and some even stricter than Europe (must prove off street parking before buying a car, etc.) and yet as shown again and again on this thread doesn’t get the ridership at distance that France seems to. Something else must be at work.

      • AJ

        RE #2, I think you make a mistake in viewing the land use along I35 and I45(ish) as interchangeable. I35 follows the Balcones Escarpment, which is the dividing line between the Great Plains and the Coastal Plains and means the cities along this corridor are much older than elsewhere in Texas. So yes, the vast majority of the population of San Antonio, Austin, etc. are suburban/exurban, cities like New Braunfels, San Marcos, Temple, and Waco all have sizable, pre-automotive urban centers with legacy rail alignments serving the downtowns directly. These cities would all be well served by a HSR line with urban stations (even if the express service was only SA-Austin-Dallas). While this alignment probably couldn’t get the raw speed of a greenfield alignment, building an alignment that would also support a strong regional rail service would be better than an HSR-only line that skipped those cities? Or is your exercise here solely focus on plausible greenfield HSR?

        Between Houston & Dallas, OTOH, there truly is nothing aside from College Station and any intermediate station would been in a proverbial Texas beet field.

        • Onux

          It wouldn’t even be a beet field station. The TGV Haute-Picardie station was built in a beet field to compromise serving Amiens and Saint Quentin (~200k between them) without the expense of engineering an urban alignment to keep speed up between Paris and Brussels/London. North of Conroe there is nothing that big along I-45.

    • adirondacker12800

      I am curious why the map does not have a true Texas triangle.
      Because railfans are deeply fascinated with college towns? Which might have high ridership ten days of the year and are filled with poor college students the rest of it?

      • AJ

        Yeah it seems like Waco is a much more sensible junction than College Station, not because Waco is any more important (both MSAs are bit under 300K) but because missing the half million in Killeen–Temple–Fort Hood seems like an oversight.

        If Alon wants to do SA-Houston via Austin, which I think is clever as it boosts the frequency on SA-Austin, then why not a Texas “A” rather than a Texas triangle?
        1. SA-Austin-DFW
        2. SA-Austin-CS-Houston
        3. DFW-CS-Houston

        This allows for SA-Houston to follow the I35 arc of the SA-Austin conurbation, and also allows for Dalls-SA and Dallas-Houston to share track as far as possible before needing to split (probably Waco but not necessarily). I think Alon is overemphasizing the need to maximize the straightness of the Dallas-Houston line (the largest endpoints) over service along I35 (much more stuff along the way).

        • Martin

          Such an A limits capacity into Houston and SA, while constructing basically the same amount of track. It would also slow down trafik between SA and Houston quite a bit. The right of ways may be easier though.

        • Phake Nick

          I thought the point is minimal extra construction needed to connect Austin and San Antonio to both Dallas and Houston on Texas Cemtral line, which already have defined alignment and station position

          • AJ

            Oh OK, so Alon is assuming the Dallas-Houston line is built & is a sunk cost?

          • Onux

            Alon cannot be assuming the Texas Central Line because it does not go through College Station, it goes Hou-Dall direct. Alon built the T-Bone but incorrectly flipped the legs by not having a continuous line along the I-35 population corridor.

            @AJ – Your Texas ‘A’ could be an evolution of my T-Bone routing as ridership grows. The Dall-Hou link is the most important so giving it a shorter route direct from Coll Sta when demand justifies it is possible. It might not be necessary though, the route through Coll Sta and Temple should be fast enough to meet Dall-Hou travel needs. In that case the extra track to build would be a direct Hou-SA link; those two cities are closer than either to Dallas, so the T-Bone routing makes travel time not as competitive by going farther out of the way.

          • Alon Levy

            The Texas Central line is supposed to be passing fairly close to College Station, in at least one version. And even the most direct version can still have the San Antonio leg go via College Station. The reason for this rather than the T-bone is that Dallas-Houston is the strongest city pair by far and should get a direct line, without detouring through Killeen.

          • Onux

            The selected route for TCR bypasses Coll Sta, it places a “Brazos Valley” station in the middle of nowhere 25 mi from Coll Sta (official website description is “60-acre plot of land on Highway 30, just west of Highway 90”, not even a named city, closest community is Roans Prarie, pop. 56).

            The Dallas station is urban, about 1 mi from the current Union Station, but the Houston site is very suburban, over 6 mi from downtown.

            The problem with privileging the Dall-Hou pair vs the T-Bone is that the pair is strongest not dominant. SA-Aus-Temple-Waco combine for 5.6M vs 7.3M in Hou. And all 5.6 are in a straight line, with as @AJ noted pre-automobile town centers built originally around a railroad alignment, all features for strong demand.. Cutting 1.3M out of that service, and making the rest of the service detour, isn’t optimal.

            In comparison, TCR is planning on doing 240mi in 90 min. A 300 mi T-Bone route Dall-Hou could be done in less than 2h, plenty fast to capture the entire market, so there is no real downside to detouring for Dall-Hou (especially since most service will be express, as there are effectively no stops on the Houston leg either way.

            SA-Hou is the loser in any T-Bone scheme or similar like yours, it is a 197mi direct shot between them but 319mi via Temple/CollSta. Luckily all population on that leg isn’t in SA, so the Austin-Hou times remain competitive.

          • Phake Nick

            > TCR is planning on doing 240mi in 90 min. A 300 mi T-Bone route Dall-Hou could be done in less than 2h, plenty fast to capture the entire market, so there is no real downside to detouring for Dall-Hou (especially since most service will be express, as there are effectively no stops on the Houston leg either way.

            Problem is, door to door time competition with cars, and also longer route mean cost more to construct and operate and also passenger have to pay more in fare to make up for that, which are all negative to market share of HSR. Which is also why JR Central was hellbent on not detouring the Chuo Shinkansen via Suwa area of Nagano despite being both a popular tourist place and can also be expected to attract connecting demand from mid-sized city Matsumoto nearby.

          • AJ

            If Dallas-Houston is the strongest leg, then that’s an argument for that segment to be built first, but not an argument against serving Kileen/Temple & Waco? Seems plausible for Alon’s alignment to be an intermediate service pattern, if Austin-Temple-Waco-Dallas hasn’t yet been built, but not anticipating an A-T-W-D service seems like a major oversight.

            I’m much less concerned about the lack of direct SA-Houston service … SA-Austin is a far more important trip pair, to the extent that SA & Austin are growing into a single conurbation. A SA-Houston direct line could be built later as a relief line if SA-Austin gets too crowded, if Texas somehow manages to maintain it’s level of growth for a few more decades.

          • Phake Nick

            @AJ If Austin-CollegeCity-Dallas line already exists, then building A-T-W-D would only have two effects, 1.) adding new service to TempleCity and Waco, 2.) make Austin and SA arrive Dallas faster than via CollegeCity.

            Given that Temple City and Waco only have about 200k people in total when added together, it’s almost certain it won’t in its own justify a high speed rail, at least not before many other potential routes in the US. As for the demand of faster trip from Austin/SA to Dallas, with Dallas-Houston expected to be far from packed and adding SA/Austin would probably be less than double the traffic, it’s almost certain there will not be capacity issue that would have make building a parallel high speed line being reasonable.

            As for SA-Austin…. High speed rail are inherently more costly both to operate and to run, and they also have few stations inside city due to high speed rail cannot stop too often. If the goal is to provide service for people to use daily, it might be more suitable to build lines and trains that are up to say for example 79mph speed, which will also save a lot of money to construct due to less demanding on the graident and curvature and clearance of the track and such, and also saving money from maintenance and operation.

          • AJ

            City populations are usually irrelevant for American stations. The Temple station is to also serve Killeen & Fort Hood, which has nearly a half million people combined; the station is suggested to be in Temple because that is a pre-automotive city with a good core, but the station could be like I35 sit in between Temple & Killeen. McLennan County has >250K people, which is more relevant than the Waco city population of ~140K.

            For SA-Austin, I think you make a great point. There was a effort recently (Lone Start Rail District) that failed to get political tractions but seems to have the fundamentals right. There is an existing UP rail line that serves all the pre-automotive urban centers and could support 79mphs speeds; it does seem that creating that service would be better than going straight for HSR.

          • Alon Levy

            Don’t add up the entire population of the cities from Waco to San Antonio. Austin and San Antonio would be getting service either way, just with the same half hour detour that’s supposed to be nbd for Houston. Waco and Temple would be bypassed, but not enough people live there compared with the others.

          • Martin

            Regarding stop frequencies it is entirely possible to run high-speed rail with quite frequent stops. Taiwan HSR use Shinkansen stock and has 12 stations on 350 km, where 4 stations are within the Taipei subway system. It is one of the more successful systems in the world. Of course all trains do not stop at all stations.

          • Onux

            “Don’t add up the entire population of the cities from Waco to San Antonio”

            Why not? Do these people not exist? Total pop of Waco, Killeen/Temple and the northern suburbs of Austin is 1.3M. If you plan a whole HSR line (not just a stop along the way) for Birmingham or Memphis, why not route a line to serve this population instead of bypassing it.

            The half hour detour is NBD for Houston, and also NBD for SA/Austin for a CS-Dall route. In all cases the primary nodes are connected in less than two hours, so the concern is not travel time, but serving population along a linear corridor. Which leads to…

            “Regarding stop frequencies it is entirely possible to run high-speed rail with quite frequent stops.”

            Exactly. Japan does this with Nozomi/Hikari/Kodama, Amtrak does on the NEC with Acela/Regional (and everyone agrees with true HSR there would be an even more express stopping pattern). Texas should be the same way. Just because the fast train goes SA-Aus-Temple(transfer)-Dall doesn’t mean there cannot also be a SA-New Braunfels-Aus-Temple-Waco-Dall service, and a regional service also stopping in San Marcos, Round Rock, etc.

            “If Dallas-Houston is the strongest leg, then that’s an argument for that segment to be built first, but not an argument against serving Kileen/Temple & Waco?”

            Excellent way to phrase it, hits the critical issue exactly.

            “If Austin-CollegeCity-Dallas line already exists”

            It doesn’t, hence the reason to plan the better Austin-Temple-Waco-Dallas line.

          • Phake Nick

            I am not sure why you would want to emphasize “a whole HSR line, not just a stop along the way”, as the nature of HSR dictate it is not going to have a lot of stations for everything along it, and are instead connecting cities to cities.

            If your goal involve anticipating dedicated service for individual area like a specific part of suburb, then it would not be wise to consider HSR as your solution

            Japan have fast and slow HSR trains, but even in such segregation, the slower trains only stop once every ~30km. And in this case the slower train mainly serve local population who want to head to main cities much further away, rarely would that be used for short distance commute due to nature of accessibility and frequency and fare of such local service.

            >>“If Austin-CollegeCity-Dallas line already exists”
            >It doesn’t, hence the reason to plan the better Austin-Temple-Waco-Dallas line.
            We can assume CollegeCity-Dallas line already existed due to Houston-Dallas rail plan. Austin-CollegeCity would then be a giveaway due to Austin-Houston travel demand. Hence these two legs doesn’t need to be individually justified, unlike the proposed A-T-W-D line.

          • Matthew Hutton

            Strongly agree that the places along the way all need a station. Political realities require them – or they require huge unnecessary expense like HS2. And frankly when almost no line has 15tph or whatever the maximum is you can have a train that stops every 30km that has a lower fare than the other trains.

            Don’t forget the part of the Chou Shinkansen that is being held up is the only province that doesn’t have a station – and it’d be difficult to add one as there looks to be no road access to the location it’s that remote.

          • Phake Nick

            @Matthew this sort of mentality is the reason that Hokuriku Shinkansen now need to a long tunnel deep under national park in order to avoid flat land in another prefecture and increasing the cost to a level that almost make the line become financially unviable. It is not realistic to expect all locations along the line to be “served” by a high speed rail or else it won’t be high speed. Even if you make trains bypass the stops the stations will still cost extra to construct and the station itself will also affect schedule of other trains.

            Chio Shinkansen being held up is because their governor is ideolgically oppose urbanization and thus he oppose the Maglev project which is expected to foster the growth of main urban area of Japan. He instead believe Japan should be deurbanized to allow more balanced development across rural area of the country. Even if Chuo Shinkansen instead follow the route of Tomei Highway sand have multiple stations instead the prefecture I do not believe he won’t oppose it.

            The “political realities requie them” indicate there are problem with political realitiy. There have been people in Japan pointing out the problem of how lines that merely pass through or connect to a prefecture still require their funding and require them make sacrifice on conventional rail service to facilitate high speed rail construction being unjust, thus there are calls to reform the political condition being set out against high speed rail construction.

          • Onux


            I emphasize stops along the way versus a whole line because Alon’s map has lines of 150-210 mi just to access cities in the 1.1-1.3M range. That’s all the ridership they can serve. Well guess what, there is 1.3M worth of people in central Texas that require only 195mi to access it, and you get Dallas-Aus/SA along with it.

            Good HSR does in fact have stations along the way, not just in the main origin and destination cities (a “flight-level-zero airline” model). The 30 km Kodama Shinkansen spacing is pretty close to Northeast Regional spacing from DC-Boston (24.5km) and no one has ever suggested that service to New Rochelle or Bridgeport should stop if Acela is ever replaced with real HSR. That 25-30km spacing is also just about what I would expect for the 440km from SA-Dall (New Braunfels, San Marcos, S. Austin, Austin, Round Rock, Georgetown, Temple, etc.) Connecting the 475k people in Kileen/Temple (with a single station) to San Antonio 153mi away IS an intercity journey, not some service for a specific part of a suburb as you suggest.

            I am sympathetic to the sunk infrastructure costs if the TCR line does get built, but it isn’t yet. Alon’s map does not use the TCR alignment because TCR does not go to College Station, as I noted. The latest news I’ve seen from last year (2022) was that TCR was stalled, funding is not materializing, land acquisition is slowing, and newspaper editorials are saying the project should come clean if it is not going to happen. In that environment we can discuss and plan the best possible system for TX, rather than shackling ourselves to legacy alignments that are not legacy yet because they don’t exist.

          • Phake Nick

            Can you show the math of there exists 1 million people between Dallas and Austin, not counting the metro area of either of them?

          • Onux

            Waco is 280k metro area
            Kileen-Temple is 475k
            Williamson County is 609k

            That totals to 1.364M

            Wait, Williamson County is part of the Austin MSA you say? Yes, but Round Rock is 18 mi north of Austin and Georgetown is 25 mi north. Under Alon’s plan the tracks do no go through there after Austin and the county gets no stations. Metro areas in the US are very arbitrary along county lines. For instance, Wilmington and Trenton are both about 25 mi from Philadelphia (same as Georgetown from Austin) yet Wilmington is in the Phila. MSA while Trenton is its own MSA that then joins the NYC CSA. Even if you cut out places closer to Austin that are basically contiguous to it like Round Rock and Cedar Park you still get more than the 250k beyond 20 or mi to bring the population between Dallas and Austin over a million (and even then Round Rock is large enough to justify a station to serve N. Austin suburbs, just like Metropark outside of DC or Rt 123 south of Boston).

            Riders do not care about what metro area a statistician assigns them, they care about access and distance. Putting one station in the “Austin Metro Area” doesn’t mean you automatically serve everyone living there just because the county line happens to include them. Wilmington and Trenton are the 9th and 14th busiest stations on the NEC, do you think you could shut those stations down and retain all of their 1M yearly riders, because they will all drive or catch a commuter train 25 mi to Phila. (perhaps starting their journey going the wrong direction, as in Trenton to Bos)? Obviously not.

            So yes, there are 1M+ people between Dallas and Austin that can be served by an HSR line following I-35 (and soon to be more, Waco and Temple both grew at >10% the last Census. Some north Austin areas like Georgetown have seen census-to-census growth of 40/50/60% in recent decades.)

          • adirondacker12800

            It doesn’t matter how you slice and dice the population along 1-35. There are more people along I-45.

          • Phake Nick

            Well if you want a Northern Austin station, that’s easy, just make the track from Austin to Houston pass through Northern Austin (Maybe even less if your station is closer to Hutto/Taylor), it would only add ~20km detour instead of needing 200km new track, making it much cheaper while covering more than half of your target passengers.

          • Onux

            There are more people at the end if I-45 than I-35 (Hou vs SAntonio) but there are more people *along* I-35 in that the population living on the direct line between Hou-Dall is effectively zero (I-45 doesn’t even go through CollSta). No matter Dall-Hou direct, or Dall-CollSta-Hou or Dall-Temple-CollSta-Hou the “along I-45” population is served when you get to Hou (assuming good station locations) so why build a line where there are no people in between when you can build one where there are people?

            An I-35 alignment doesn’t require 200km “extra” track, the T-bone layout has the same amount of track no matter which route is the long leg and which is the dog-leg.

            Hou-Dall through Temple is fast enough to gain 100% mode share from flights and easily out compete car travel even when speeding with no traffic. There is no need to go faster than this because the ridership gains at that point are minimal. The ridership gains from serving the population along the northern half of I-35 are not.

          • Phake Nick

            Longer track for most people = More travel time needed, more manpower hours needed to operate the trains, higher operation cost, higher fare.

            Specifically when it come to Dallas – Houston, the extra time and fare could reduce the rail’s attractiveness to car drivers, reducing its effect of attracting driver to change modes, and this shortcoming could very well negate extra ridership that might potentially be brought onto the line if there are stations serving them, due to sheer difference in the base population number, and also due to how end to end traffic will be significantly dominating majority of traffic that even just a few percents of change in expected ridership there is probably already cancel out potential riders that these lower usage stations can achieve.

          • adirondacker12800

            There being more people along the way doesn’t change that there are more people.

          • Onux

            “Longer track for most people = More travel time needed, more manpower hours needed to operate the trains, higher operation cost, higher fare.”

            There are 7.3M people in Houston, 5.6M along I-35. It’s not a huge difference in people, as if Houston was multiple times the size of Aus/SA. As a result there will should be roughly the same level of service for both areas (an express train every 30 min, a local train every hour, etc.); the Houston trains will just be more full. If the service is equal on each branch then the time saved by one train Dall-Hou is lost by a Dall-SA train and so your total operation cost for the system stays the same. My T-Bone is the same track and layout as Alon’s just flipped so the through line is on I-35. So my T-Bone has the same total operating cost but serves more people; more riders will mean lower fares.

            Note also that the Texas megalopolis is not a perfect triangle. Due to actual geography my T-Bone has the advantage of relatively even legs; 275mi shortest trip Dall-SA, 318mi longest Hou-SA. At those distances it is feasible to make each express trip 2hr for a takt, with Temple being close to 1hr from each vertex. Alon’s layout makes the legs more unequal: Dall-Hou at 244mi is much shorter than Dall-SA at 344mi. You can get Dall-Hou down to 1h40m or so, which fits no takt, while Dall-SA becomes difficult to do in 2hr, which means that on-the-hour departures require extra equipment sets. College Station is poorly sited as an interchange, given that it is twice as far from SA as it is from Hou.

            “Specifically when it come to Dallas – Houston, the extra time and fare could reduce the rail’s attractiveness to car drivers”

            A 2h journey will not be less attractive to drivers than a 1h40m journey. Both are much faster than the 3h30m drive without traffic, and with traffic it can be much longer. Texas Central cites a report that travel time is expected to average 6h30m by 2050 and while it is likely that Texas would build additional freeways to keep it from getting this long before then, it will not be hard to make HSR a better deal than driving using either route. All around the world HSR mode share reaches its max at journeys of around 2hr and does not rise much for shorter journeys. So no, you won’t get much more ridership from a 20 min shorter journey at that distance, but you will certainly gain riders SA-Temple, Dall-Round Rock, Hou-Waco, etc. if you place the through leg and stations along I-35.

          • Phake Nick

            @Onus Expected pattern of service with Alon’s T-bone could be, Dallas-Houston 1/hour, SA/Austin-Dallas/Houston 1/hour, with the train splitting to different direction at the station near College Station. A total of 2 trains per direction per hour. It can even be timed to coordinate schedule, so that the trains from/to SA/Austin join with the Houston train at the midway station in both directions, to further save the cost.
            If you build a high speed rail line from SA/Austin to Dallas and then expect service from Houston to join the line, it depends on where you connect the line to. If the line split at Waco then the distance won’t be too different, however the detour is too much for Houston-SA/Austin passengers, which the system would either fail to capture those passengers or need to spend extra track to dedicatedly serve these passengers (Even if the travel time is still fast enough, the fare would cost so much that it won’t capture many travellers. At least it would be more expensive than Dallas-Houston with such layout.). On the other hand, if merged near Temple, that’s almost 100 km extra on the Houston to Dallas line, and passengers there would need to pay ~25% higher fare for this. Texas Central’s fare now is expected to cost between driving cost and flight ticket cost, and such proposed route would mean 25% higher fare than these compared transportation means to passengers.

          • Phake Nick

            And 1h40m vs 2h is not no different to travellers. Even if the portion of traveller shifted to train remain unchange, it will still affect induced demand. Not to say, currently Texas Central expect a trip time of /less than/ 90 minutes, which mean even taking into account the turnaround time, only 3 trains is needed for active service between Houston and Dallas off-peak, but if it’s 2h, then 5 trains will be needed.

          • adirondacker12800

            More than half the population of Texas lives in metro Houston and Dallas. Detouring anywhere makes the trip longer for more people.
            Phate: A two hour trip means the train can make 5 trips in ten hours. 1:40 means it can make six. Use 20 percent more train, the fare has to be 20 percent higher.

          • Onux

            SA and Aus combine for a population of about Seattle. Pre-Covid, Amtrak was running really slow service at 6 tpd (about 1 every other hour) between Seattle and Portland. You want about the run the same level of service from SA/Aus to Dall and Hou (1 tph then splitting at CollSta)?!?!?! Spending tens of billions on infrastructure and then saying routing doesn’t matter because you are only going to run 4 tph total on the system to serve an area of 21+M isn’t a great argument.

            The Texas Triangle is big enough to justify at least 2 tph (one express one local) each direction from each vertex to each other one (2 tph Dall-Hou, 2 tph Dall-SA, etc.) Dall-Hou might get more express (biggest pair, basically nothing in between) while Dall-SA might get more local/limited (linear corridor, serve SA-Temple or Aus-Waco, etc.) So no free lunches. What you gain on Dall-Hou being shorter you lose on Dall-SA being longer, or vice versa.

            Talking about starting the Hou leg at Waco does raise the question of whether a mini-triangle of Waco-Aus-CollSta with spurs to the big 3 (the “Texas pinwheel”) is the best choice. Does add extra track but makes things marginally better for Dall-Hou and probably much better for SA-Hou.

            “passengers there would need to pay ~25% higher fare for this”
            According to Alon’s suggested HSR price ($0.135/km) the Dall-Hou ticket goes from $53 to $65 dollars using an I-35 T-bone. 25% increase on a reasonable cost is still a reasonable cost.

            If 100km extra path is terrible for HSR ridership, then that means we need to drop HSR service to Tampa (more than 100km extra to go via Orlando then direct via I-75) and upstate NY (again, more than 100km more to get to Buffalo via the Hudson River Valley then drive direct via Pennsylvania). 55 extra mi is a big deal on a trip of 55mi to start, its double. 55 extra mi might be a big deal on a trip that is already long and at the wrong end of the mode share curve, i.e. where every mi counts (CAHSR and the Palmdale detour fits into this category). 244 vs 299 mi? Both within great HSR performance range.

            With a 2 hr nominal journey you need 4 trains not 5, the 10:00 departure from Hou becomes the 12:00 departure from Dall and vice versa (if TCR can do 244 mi in 90 min including turnaround than 299 mi in 2 hr with turnaround is equally easy). But even then, you are worried about 2 extra trainsets!?!? Amtrak has at least two trainsets working routes like Springfield to Greenfield in MA, we’re talking Dall-Hou, they have 15.5M people between them! If you are only running hourly service during part of the day then you are doing something wrong and and having 1 extra train on the tracks isn’t it.

          • Alon Levy

            To be very clear, the pricing model is at best first-order. In sufficiently circuitous cases I had to put my finger on the scale – the original Metcalfe’s law post explicitly prices DC-Pittsburgh based on the direct length and not the route length in order to be competitive, which means that in practice it’s a breakeven price. So you should assume that fare receipts are not increased if routes get longer, but only operating costs are.

          • Onux

            Alon, thank you for the helpful clarification. Re-reading your post I see that Dall-Temp-Hou would have 244mi of revenue ($53) and 299mi of cost ($33). So $20 profit per ticket, when optimal is ~$25. Much better than a breakeven price (the detour is far less than DC-Pitt via your plan) and the profit given up per ticket has to be considered against revenue given up on the Dall-SA leg (fewer tickets but greater missed profit per ticket, ~$8 per rather than $5) and the total lost revenue of riders to/from N Austin, Killeen/Temple, and Waco.

          • adirondacker12800

            It depends on how you cook the books. If the tracks between Pittsburgh and Harrisburg are there for markets that have more demand, having more trains use them doesn’t cost much. Cook the books so that each passenger is paying per mile/kilometer the fares are lower or the profits higher. The no-build option is to have Pittsburgh-DC trains go through Philadephia. That would add around an hour to the trip. Renting the seat and staffing the crew costs money. Which raise the fare.
            The farther west you move the Philadelphia bypass the more track you have to build. Lancaster PA to Perryville MD might be a compromise. People in 2113 can hash it out. So that something gets built by 2150.

          • Phake Nick

            @Onux Dallas-Houston 1tph is Texas Central’s proposal. Based on this it make sense that the SA/Austin branch is also 1tph.

            It might actually be able to support 2 or even 4 tph, especially if shorter trains are used. But at this range of number they will simply be the same situation as my reply except the time span of 1 hour lower to 30 or 15 minutes.

    • Onux

      A Texas Triangle would be nice, but it is almost always better to connect three locations via a point in the middle (ideally the Fermat point – where the sum of lines to the center is minimized, i.e. least track length).
      For cost effectiveness you connect all three locations with less track, usually less track than to connect just two points via the sides, which means less money spent and also starting service quicker.
      Operationally, a triangle splits frequency of trains leaving ‘A’ – half to ‘B’ and half to ‘C’. With a center point all traffic leaves on a single line, which means leaving ‘A’ you get on any train; if you leave on a train to ‘C’ and want to go to ‘B’ you transfer at the center (Alon of course knows that these transfers should be timed on a takt with a A-C train meeting a C-B and so forth, the transfer then has little to no time penalty and can be done cross platform).
      The disadvantage to center point routing is that it is slower because you are not taking the most direct path. However, as a practical matter trains never take a absolute straight path, so the additional distance by going through the center usually isn’t too significant – in Texas the Fermat point would only add 10-15 minutes travel time at HSR speeds, and each major city pair could easily be connected in under 2 hr.
      But Texas is complicated by that fact that the triangle is not three isolated metros, but many metros in a line (Dallas-Austin-S Antonio) with an outlier (Houston). Thus a statewide system plans should be a ‘T-Bone’ layout, with a line along I-35 from Dall-SA and then a dog-leg from Temple to Houston via College Station. Alon’s map, driven I assume by the Texas Central Railroad plan for a line direct from Dall-Hou, inverts this, with the dog-leg from College Station to Austin then SA.
      Alon’s plan is inferior to the I-35 primary, because is cuts service to a decent population in a linear corridor. Waco and Temple/Killeen don’t get HSR service, although they combine to 750k people. The northern part of Austin (Round Rock/Georgetown) also lose stations directly serving them, and Williamson County is another 600k. Lest anyone object that 1.3M isn’t significant (especially broken up) I note that Alon includes lines to Memphis and Birmingham (1.3 and 1.1M respectively) plus stops in Dayton (813k), Erie (269k) and Mansfield (125k) – given that the mid-I-35 cities I mention are anchored by Dallas and Austin/SA, which are far larger than cities anchoring the examples above, it is reasonable to assume stops (on a limited type service, not the express) would do well.
      All T-Bone plans have the disadvantage of making travel to the dog-leg longer that the minimal impact center point layout. The problem is acute for Hou-SA, where the dog-leg distance can be over 60% greater, with 45-60 minutes longer travel compared to a direct route, and it becomes difficult to keep travel under two hours. On the other hand, mid triangle side cities, in this case Austin to Waco, do much better with a T-Bone vs a triangle, with service that is faster than having to go to another vertex city first.
      Instead of why not a Texas Triangle, the better question is why does Alon have a triangle for Indy-Cincinnati-Louisville, instead of three lines meeting around Osgood?
      This three-body-problem has relevance to HSR issues outside of Texas, particularly California, where the Fermat point of LA-SF-Sacramento is approximately Tracy, which should drive an Altamont pass routing (as well as Tejon pass, to minimize LA-NorCal track distance), while the Fermat point of LA-San Diego-Las Vegas is approx. San Bernardino, driving a Cajon pass route. Conversely, the geography – both physical (swamp) and population – of Florida argue against a center connection for Miami-Orlando-Tampa service, instead placing the line along the densely populated coast and all service to Tampa going through Orlando. Alon gets these cases correct, but misses on Texas.

      • Eric2

        “the Fermat point of LA-San Diego-Las Vegas is approx. San Bernardino, driving a Cajon pass route”

        So would you recommend SD-LA to go through Temecula like in the CAHSR plan? The terrain is difficult in parts of the route.

        • Onux

          Yes, I didn’t look at the map closely and didn’t see that Alon had it half right. There is not-flat places between SD and SB but I wouldn’t call it “difficult” like mountains; I-15 has neither tunnels nor a “pass”. But the terrain along the coast is not-flat in places as well; at one point the coast line is a single track hanging between the beach and the bluffs. It adds only ~45mi to go SD-LA via SB but it saves ~65mi from SD to LV or PHX. At 120mi LA-SD doesn’t scream for HSR even though the markets are huge. Flights LAX-SD do exist, but the real competition is driving, and it is easy to beat that with either HSR via SB or upgrading the Surfliner along the coast to one of Alon’s blue lines. Even if they end up the same travel time (90min?; why not have both to improve SD-Orange county travel despite routing HSR inland?) you gain a lot cutting SD-LV/PHX time by a half hour or more. Cutting SD-LA from 90-70 min is meh.

          If push comes to shove I will grant Alon’s plan with SD-LA-Cajon-beyond is still workable compared to SD-SB-LA-Tehachapi-LV, which is terrible for SD and still bad for LA.

      • Martin

        I think the T-bone you suggest is an improvement on the original suggestion, but it also has a major disadvantage in that it is rather poor for Houston. Arguably the most important city.

        I think it is quite clear from the overall map that a true Texas triangle is rather marginal more extra track, compared to very large stretches of the map, where we talk trains every 120 min (or more) and not every 20 min.

        The triangle also has large (basically double) capacity advantages also (in particular if you add suburban stops), which is why it is not very popular in countries with more rail traffic.

        • Onux

          Yes, the T-Bone slightly diminishes Hou-Dallas by making it not as straight, but it shouldn’t be significant. Hou-Dall through Coll Sta and Temple is maybe 300 mi, vs 240 on the straight route. 300mi can be easily done in 2 hr with HSR, or even which means it should capture mode share completely. In exchange you give direct service to 1.3M people in north Austin and Killeen/Temple and Waco. That is an easy tradeoff.

          Looking at the map, I see Alon isn’t even doing the Texas Central direct route, he goes through college station. Deviating the line for a 120k metro but ignoring 10 times the people on I-35 doesn’t make sense.

          • Martin

            There is also the capacity issue. Any non triangular structure would put all trains in any end node from all the 3 other major destinations on a single track. That is like demand from 15+ million people regions entering a 7 million people region. I think that is really an unfortunate constraint in rush hour, and would probably make any semi-local service infeasible (which would mostly have demand in rush hour).

          • Alon Levy

            I’m not worried about trains in Texas filling beyond capacity, I’m worried about underfilling. The Deutschlandtakt has the busiest link in Germany, either Frankfurt-Cologne or Frankfurt-Mannheim, at 6 trains per hour; this assumes pretty slow trains and they should be planning around high-speed rail with more traffic, but also Texas is a lot smaller than Germany.

          • Martin

            I think if one indeed assumes tracks with decent right of way between SA and Dallas, I think it is likely it would have enough commuter demand in rush hour to create very real capacity constraints.

            SA + Austin in itself should generate a lot of commuter traffic given the distance, and all other mid-sized towns between SA to Dallas as well. Austin would also likely overperform on commuter traffic as it is the state capital.

            If you add the rush-hour commuter traffic on interregional travel + intercity rush-hour trains at SA-Dallas and SA-Houston, I think (rush-hour) capacity will very much be an issue, nearly from the start. And that is before the existence of such high-quality rail, will induce demand for a more reasonable urban structure in a 25-year perspective (probably easier in Texas with looser zooming). Getting all of that traffic on a single-track SA-Austin section seems a bit painful, it is after all a 20 million people region.

            A more general point is that assuming that one could get right-of-way tracks l SA, Austin, Dallas, and Houston a reality, it would be unfortunate to not use the same right-of-way for inter-regional travel (e.g. 2-3 stops in each metropolitan area, and stations in mid-sized cities on the way, in particular in the section between SA and Dallas). Building and optimizing for pure inter-city traffic would be a waste of valuable right-of-way. This I think is a general principle for investment in HSR, in low-rail environments. Texas probably needs regional trains just as much, or even more than high-speed trains.

          • Matthew Hutton

            Isn’t the only high speed line in the world at capacity the Tokaido Shinkansen?

            Given the stopping patterns all vary it’s difficult to look at Paris-Lyon but it didn’t look to be close to capacity when I had a quick look. Maybe it is on Friday afternoons.

        • Frederick

          Nanjing to Shanghai has about 200 trains per day, but there are two lines (one is Beijing-Shanghai HSR and the other is Nanjing-Shanghai Intercity), so they are probably not at capacity except in holidays.

          The Cologne-Frankfurt line and LGV Nord all look like 6tph. LGV Est near Paris might be more frequent, but still far from Tokaido levels.

          • Martin

            I think there are plenty of HSR systems that are at capacity during popular times of the day and popular seasons.

            And a transit system that is at capacity for 12-14 hours a day, i would argue is rather a symptom of being under strain. För all transport modes traffic demand differs over the day.

            Of course, at capacity is also a function of price.

            The more commuter oriented your system becomes, the more it will have shifting demand.

          • Martin

            The Taiwanese HSR frequently sells out at popular times, even with high frequencies from very early morning to late night. So do many of the Chinese lines I think (getting tickets in the summer in the southwest was nearly impossible for me). The Japanese HSR seems clearly designed to push customers away from shorter trips towards alternative regional lines, which I think is quite common. Mixed HSR such as in northern Europe is nearly always in stiff competition with both commuter trains and regional trains for track space (during rush hour). E.g. in Stockholm even with 4 tracks, capacity is very constrained with HSR/regional/commuter traffic, which is far from unqiue.

            For more suburban stops it is also important to recognize that capacity is a matter of service and stop patterns. If you organize the HSR and pricing so it easily can accommodate rush-hour commutes within the <100 km range, demand for that very easily fills up. In fact, HSR often needs ticket schemes to discourage such use (and HSR nearly always exists in contexts where there exist reasonable regional trains, unlike in Texas).

            But in Texas, such regional traffic would be very desirable, and something that would be good to plan for, not to avoid. Thus, given that it would be the only rail corridor in a hugely populated area, I think planning for more regional trains on the corridor makes sense, which means that capacity constraints in rush hour times seem very likely.

      • IAN! Mitchell

        Texas Central is using (very roughly) the intersection of highways 90 and 30 as their Brazos Valley station location.


        If you wanted to include the northern portion of Austin/Williamson County (almost half of Austin’s total MSA) and any of the Killeen/Fort Hood area (itself another half million people), you’d put a station either in Belton (where there is a very walkable downtown and local transit connections currently exist to Killeen, Temple, Harker Heights, and even Waco), or somewhere in Williamson County that’s easier to drive to for the ~1 million people who will live there by the time any of this actually gets finished.

        As you can tell, I favor Belton.

        It’s ~130 miles from Austin (using Bergstrom) to 30/90. It’s ~50 miles of additional track to add a station for the Fort Hood/WilCo area.

        Austin to NW Mall (TC’s Houston station) directly would save ~50 miles over connecting at 30/90.

        So it’s basically a question of whether the ~million people in the fastest-growing part of Texas are worth bending the track for, and I don’t think it’s rational to say they aren’t but the ~200k people in the College Station area are. Both are detours.

        It’s ~250 miles to serve Houston, College Station, Fort Hood/Wilco, and Austin, or ~150 for Houston-Austin without detour. Does that 100 miles matter more for folks coming from San Antonio, or even Laredo? Maybe.

        Bullet Trains can detour quite a lot and still be faster than driving or (post-9/11) flying.

        • Matthew Hutton

          A fair chunk of the slowness at the airport is inherent to airports themselves.

          Even ignoring security to get through Heathrow you probably need 30 minutes from arrival by taxi even without contingency and without checking baggage. Beyond that I think it would be very foolish not to allow half an hour for traffic on the M25 and toilet stops in the airport etc.

          Whereas if you’re going from Brussels Midi, say, for the same reliability you probably need to plan to arrive 15 minutes before your train.

        • AJ

          Giving Austin two stations, one in Williamson County, I think would still fit Alon’s T-bone plan.

          But to go all the way to Belton, that may be only +50 miles to Houston vs an Austin-College Station ‘direct’ connection, but what does that do to Austin-Dallas if the train goes up to Belton, back to CS, and then up to Dallas? Even if it’s ‘only 50 miles,’ that just seems unintuitive.

          • IAN! Mitchell

            350 Miles via Belton and TCR’s brazos valley stop. 205 miles via I-35.

            It’s generally 4 hours to drive it, though it can take *much* longer, too.

            The average speed needed to beat driving and post-9/11 flying is pretty mediocre in high-speed rail terms.

            Belton is closer to Round Rock, Georgetown, and Leander in terms of realistic travel time than any Austin stop that doesn’t totally miss downtown (IKEA in round rock isn’t a good station location for Downtown Austin), and unlike anywhere you can viably build anything in Williamson County, it’s walkable and connected to the Killeen/Fort Hood area from Lampassas to Temple and Waco today, even without a train.

          • IAN! Mitchell

            Realistically, if trains are filling up between Austin and Brazos Valley, you build the Belton-Ft. Worth (and probably DFW) leg. The Loan Star Rail project is a good project with or without high-speed rail between Dallas and Houston.

          • adirondacker12800

            They aren’t going to be filling up between Belton and College Station because there aren’t many people out there. And it’s very unlikely the trains will be going through or to College Station.

  5. AJ

    Why 6 stops in metro Detroit but none at the airport? The Detroit airport is a major international gateway & national hub, and the Toledo-Detroit alignment only needs to curve slightly (and has freeway ROW to leverage) to serve the airport.

    • AJ

      Similarly, a Mansfield-Cleveland line could have a station at/near the Cleveland airport for suburban access before terminating in downtown Cleveland, but you only put one station in Cleveland metro so that’s less of a quibble.

      Do you consider airport-rail transfers in this exercise? They won’t be captured in your Metcalfe’s model spreadsheet, but are they are plausible adjustment when you are crayoning how alignments to enter & exit cities?

    • Alon Levy

      Ooh, good point, will revise for next time.

      And yeah, my first-order analysis didn’t look at air-rail links or anything that specific, hence the pattern of one stop per metro area except where there are multiple preexisting stops like Boston or Detroit.

      • Richard Gadsden

        I’m on my phone and can’t see the map in enough detail to check, so you might have done this already, but Atlanta would be better served from the airport than any possible city location

      • adirondacker12800

        Cleveland Airport already has rail, Almost no one uses it but it has it. It’s also on a likely route out of Cleveland to Columbus. They can hash it out in 2237 when Ohioans decide that trains are a good idea.

        • AJ

          Yeah I considered the possibility that the train from Columbus would just end at the airport and riders could take the subway into the city, but the Chicago-NYC rail line is highly likely to serve downtown Cleveland directly (given topography & existing rail alignments), so I’d expect any Cincy-Cleveland line would make it all the way to downtown Cleveland.

          Cleveland has a great legacy of streetcar suburbs and is one of the American cities that could support good transit ridership if/when car parking becomes less free & abundant in the urban neighborhoods. Cleveland’s heavy rail, light rail, and BRT lines are already in the ‘right’ alignments, IMO, which underscores the fact that good transit design is worthless without corresponding good land use & transit-oriented development.

          • adirondacker12800

            It could continue into downtown-ish and keep on going to Buffalo and beyond. People from Columbus get off and people from Cleveland get on…

  6. AJ

    Having driven between all these cities many many times, I wonder if the connection between the Tennessee and Ohio is better served going through Indianapolis rather than Cincinnati? I71 between Cincy and Louisville in particular is surprisingly rugged and windy, while 65 & 70 are both generally flat & straight. I would certainly still run the Cincy-Columbus-Cleveland line, but for a inter-regional network I wonder if it would be more cost effective for trains to go Louisville-Indy-Dayton*-(onwards to Cleveland), with a transfer to the Chicago-Cincy line in Indy, and not build the Louisville-Cincy segment.

    Kentucky may even prefer the Cincy line to serve Lexington rather than Louisville. Lexington is smaller but more urbanized (geographically, economically, & culturally) than Louisville.

    *It may even make sense to put the junction up along 70 (in Springfield?), with the Dayton proper only served by the Columbus-Cincy trains.

    • Phake Nick

      The SDF-CVG line can provide onward through service from like Nashville to NYC even if the time is long, and also to various other closer cities in that direction.

      • AJ

        Sure, but my point was a SDF-CVG HSR line would be more expensive because of topography. To run even at 110mph would require moving a lot of dirt; unlike I70 & 65, I71 does not provide ROW for HSR to piggyback.

  7. Alex B.

    I don’t want to speak for all of those who dunked on Matt’s post, but would note that the sheer volume of dunks means there’s no single, unifying complaint.

    I have a couple of criticisms about Matt’s post – the main one being that he’s directionally correct but wrong about important details.

    For example, he directs the criticism at Amtrak for a bad map, when the actual news story that he links to isn’t even an Amtrak proposal, it’s a map of the corridors that states have submitted to document their interest. Amtrak presented these at their board meeting; someone on twitter then converted the photo of the screen into a map with the same style as Amtrak’s earlier graphics.

    This might seem pedantic, but it matters: the source of the bad proposals is directly attributable to the laws passed by Congress.

    Same thing with the original proposal; much of Matt’s tone is demanding what Amtrak should do, without either speaking more broadly (e.g. what the United States should do) or getting into the actual legal reasons Amtrak must partner with states for any sort of corridor service.

    Asking the question “why aren’t Amtrak’s plans any good?” ought to invite more curiosity about what Amtrak’s actual planning authority is.

    It’s great to see you highlight what the best practices for low-speed rail are; now the key question is getting the relevant entities to understand and implement them. Which will probably require getting Congress to change the law – and that’s why it’s frustrating to see Matt misdirect his critique.

    • Alon Levy

      Yeah, the bad practices tend to be staple jobs of state plans. The problem is that federal attempts at coordination have amplified this problem – the Northeast Corridor projects are a staple job of railways each wanting the others out of its way, and even where the feds do have formal and informal planning power, they don’t use it. The basic problem is that when they put together something like the NEC Commission, they hire the same hacks who’ve mismanaged the existing agencies, or maybe private-sector managers with little experience with good public transport planning and a long rolodex of hacks.

      As far as I can tell from the Northeast (well, Germany, but I have a lot of ties to New York and Boston), the rest-of-US plans for daily and twice daily trains suffer from the same problem. Nobody in the loop understands good practices, so they think in terms of daily trains as a starter service rather than as a, so to speak, finisher service.

      • Alex B.

        Yes, amplifying the problem is a good way to put it. The NEC Commission is a good example of the Congressional problems – it was created by Congress as a coordination body of the states. It’s starting from the premise that the states are the relevant rail network planners.

        The legal structure of the NEC Commission is basically inviting the staple job. And that’s what Congress created, rather than build up the specific network planning authority and expertise in either Amtrak itself or the FRA.

        Anyway, I wish the discussion of politics were directed at Congress (and Congressional staff) from people like Matt Yglesias, who has a key audience with many of these folks.

        • Alon Levy

          Coordination of states isn’t even bad – it can be a Northeast-wide Verkehrsverbund, it just isn’t because none of these people knows what a Verkehrsverbund is or how schedule planning is coordinated.

          • Borners

            Heck Japanese inter-prefecture ad-hocery would be a revelation to Anglo-ingnorami (e.g. Tsukuba Express has 4 prefectures working together). That said Verkehrsversbund is better than that.

      • Matthew Hutton

        To go back to the London Barcelona example I think a daily train that left London at (say) 10am would be OK. If it stopped at Paris – Disneyland, Montpellier Sud De France and Perpignan it would take around 8 hours. So it would arrive in Barcelona at 7pm local time – and it could then go onto Madrid for 9:30pm.

        So for the US for long distance services a single daily train isn’t that bad if the departure time is good – especially if the service is reliable.

        • Richard Gadsden

          If they are end to end services (ie intermediate stations are not important), then a sleeper train leaving around 10pm and arriving 7am can be really useful as a single daily service.

          But note that across most of Europe there are relatively frequent services over shorter distances, so you can reach the destination by changing train a few times. London-Barcelona is an exception because there isn’t a relatively frequent Barcelona-Marseilles train, but London-Nice can be done with a change in Paris and often a second change in Marseilles. A direct train would clearly be superior, but is not necessary, especially if ticketing and timetables are coordinated (which, admittedly, they aren’t). Direct sleepers are key (no-one wants to interrupt their sleep to change trains in the middle of the night) but daytime trains don’t have to be direct.

          • Matthew Hutton

            Given lots of British people like going on holiday in the Mediterranean I don’t think the via Paris option works really as no one really does it. Maybe through ticketing would help – but I think having to take luggage through the centre of Paris on top of through the centre of London is a dealbreaker.

          • Richard Gadsden

            I have done it, and yes, crossing Paris is a huge pain in the butt. There is bypass track around Paris (“LGV Interconnexion Est”) that would make it possible to avoid, but there are so few trains using it that it’s nearly useless.

            At the moment, you can change in Lille instead (which is much better), but there aren’t nearly enough trains through Lille that go to places other than Paris – plus the London-Lille service isn’t regular enough without timed connections.

            More trains running on routes like Amsterdam-Lyon or Köln-Avignon would make that much better; and the long-promised London-Köln and London-Frankfurt services (which would stop in Lille) would definitely help with the connection there.

            The other problem specifically with the Spanish Mediterranean is that the connections beyond Barcelona (e.g. to Malaga) are not good either. For most of Europe, the connections are reasonably good other than crossing Paris – e.g. connections in Milano are pretty good, also Brussels, Köln, Frankfurt and, obviously, Zürich, all of which are the key nodes for transfers when coming from the UK. But from Barcelona, you generally have to go via Madrid to get to anywhere in the south of Spain, and that adds hours and an additional transfer to the journeys relative to what there will be when the (planned) Catalonia-Andalusia HSR line is built.

            If you had three changes from London, all of them within the same station (Lille, Lyon/Avignon and Barcelona) and you could complete the journey in one day then I think there would be more people interested. At present, you have to change in Paris, Barcelona and Madrid, with station changes in both Paris and Madrid and an overnight in Barcelona which makes it feel like a station change.

            But direct sleepers are still what I think the right solution is to long-distance trains. There are probably 20-30 routes across Europe for a high-speed sleeper; I figure there are eight routes from London.

          • Matthew Hutton

            Don’t forget three changes from London is still not great. The median customer surely has two changes of train including the tube before they get to St Pancras.

            There’s also not that many destinations that you can get to in one long day or overnight from London.

            For night time trains we are probably talking the following destinations:
            * Barcelona/Madrid/Seville
            * The south of France
            * Milan/Bologna/Florence/Rome/Naples
            * Munich/Salzburg/Vienna?
            * Berlin?
            * Hamburg/Copenhagen?

            And then some of the places in between you’d probably go in the daytime, but plenty of them could be reached if you change trains once I’d hope with good service.

          • Phake Nick

            The thing with daily direct sleeper train is they aren’t really significant in the grand picture despite their cost. Each of them would carry as much people as one or two LCC planes, when there’re dozens flying every day across all cities along the route, and the night train still wouldn’t attract most people. It’s best to describe such night train as vanity niche product, and should only be conducted if its revenue can help improve other transit offer.
            The night train from Sapporo to Osaka which I quoted in the other reply was really popular, yet it was only 1% of the travel market between the two cities. These travellers are now lost as JR West no longer offer such train after Shinkansen enter Hokkaido, yet those losses are rounding error level compared to the growth in aviation market.

          • Matthew Hutton

            While the Japanese railways are super reliable which is great they have big flaws too around fixed pricing and the difficultly in booking tickets online. You’ve been able to buy tickets online in the UK for 20 years if not more.

            If you live in Sapporo or Osaka and only get the train occasionally having to go to the station to buy tickets is quite annoying.

          • Eric2

            Why not run more such trains if they are profitable and there’s demand? You could run them one after the other, 5 minutes apart. That could become a huge part of the travel marker if there’s demand.

          • Sassy

            @Matthew Hutton

            What? You’ve been able to buy tickets online in Japan for at least a decade. There is a regional fragmentation issue, but there is some cooperation for single tickets across several operators, even if it’s not as good as the integration when you buy at the station.

            It’s also normal to buy tickets right before you depart, so online ticketing has relatively small convenience benefits. With online ticketing, I buy my Shinkansen/Limited Express tickets on my phone as I’m walking towards the gates, instead of stopping by the ticket vending machines. It’s more convenient, but not by that much.

            It’s also possible to get discount tickets online, however I’m not some penny pinching housewife planning family vacations. 30% off is not worth the airline-like stress of being tied to a specific train. The fixed pricing is reasonable, predictable, and stress free.

          • Sassy

            The general ease of showing up to the station, buying a fixed price ticket, and getting on the next train, is probably why online ticketing didn’t really take off. People who take subways don’t book a ticket for a specific train weeks to months in advance, and neither should most people who take intercity trains, unless they have specific seat preferences or want a guaranteed seat at a very peak time.

            Even though it’s been available for a long time, I think it’s only gotten more popular recently, with aggressive promotional campaigns and closure of staffed ticket counters at smaller stations.

          • Andrew in Ezo

            “The general ease of showing up to the station, buying a fixed price ticket, and getting on the next train, is probably why online ticketing didn’t really take off”
            Pretty much this. I almost always use one of the in-station ticket machines to buy long-distance train tickets, and only use the staffed counters for more convoluted journey itineraries. Online only for the aforementioned high demand cases. It’s quick and the machines are almost never out of order (unlike my experience with ticket machines in the U.S.). Also, culturally train stations in bigger Japanese cities are a center of *daily* social and business activity much more than than in N. America or even Europe, perhaps having a significance greater than their purely transportation function, and buying tickets there are no more inconvenient than going grocery shopping.

          • Phake Nick

            JR Central have said that they intend to only allow online reserved tickets for Chuo Shinkansen so we might be able to see how that fair in Japan compares to other Shinkansen systems with conventional ticketing methods

        • Phake Nick

          Japan have online ticketing now and they also offer discount ticket through those online channels and apps.

          >Why not run more such trains if they are profitable and there’s demand
          Because there are such demand have not been demonstrated even in places where night train are successfully run, like for example OBB’s night train network can only be grow this successfully by splitting capacity into two half and sending them to different directions. This speak about the demand of night train.

          • Matthew Hutton

            Pleased to hear the Japanese are now selling tickets online and are offering discounts.

            I do appreciate your much more realistic take on the world!

  8. Lee Ratner

    Many American transit advocates seem obsessed with covering the entire country geographically rather than focusing on the areas with the most people and letting cars and buses cover those areas. I think this is wrong. You should focus where most people live and also where you have a greater chance of getting people out of cars.

    • Matthew Hutton

      Buses are pretty slow and expensive though. That’s the difficultly.

      The political reality is probably that if you want the federal government to fund spending on the north east corridor you’ll probably need to fund stuff in the rest of the states too.

      • AJ

        Are buses expensive? The highway network already exists, seems like building out rail to small towns in America (unless they are on-the-way between major cities) would cost far more than any operational savings of rail vs bus.

        • Matthew Hutton

          Alon do you have any theoretical costs for how much buses should cost to run in a city and in the countryside?

        • Phake Nick

          The cost comparision would depend on frequency, and pretty sure it was comparing how to support transit demand of an entire city instead of merely a few greyhound buses each day from small towns to their neighbourhood

      • onodera

        Existing Amtrak long-distance rail is equally slow. Why would anyone ride Texas Eagle or Empire Builder if they can fly? “See the natural beauty of the USA from a luxury train car” would work but wouldn’t require daily or even hourly service.

    • Tony

      The US political system results in outsized political influence for places which have little population which lie between the more populated coasts. High speed rail is one component of the urban-rural divide, and politicians who represent these rural areas have little interest to vote for something which provides little benefit for themselves. (Plus you have all the Koch funded groups that oppose passenger rail)

      • Alon Levy

        The Koch-funded groups are attacking trains on benefit-cost grounds and not rural populist ones, and most of the people within that orbit have admitted that they support investment in the manifestly profitable Northeast Corridor. The politicians who represent rural areas love to grandstand but they too take the train up to New York to beg for money from the finance industry.

        The other issue is that very few people in developed countries live in genuinely rural areas. Most so-called rural voters live in suburban areas with anti-urban idpol, using names like “exurban” (US version) or “periurban”/”rurban” (French version), and those still benefit from high-speed trains. In France the biggest opposition to the TGV comes from NIMBYs, but in politics it’s EELV and not various center-right pols trying to compete for the right-populist vote.

        • Matthew Hutton

          It’s not urban vs rural in terms of individual places. It’s the fact that by my count your plan doesn’t benefit 16 states out of the lower 48 and there’s several more where the benefits are marginal, such as Nevada where you can only go from Las Vegas to California.

          I appreciate that you aren’t going to be able to do anything that benefits states like North Dakota that gets anywhere close to positive cost benefit, but I suspect there are a few where the subsidy would be small.

          Don’t forget the plan would have to clear the senate!

          • Matthew Hutton

            More than I expected. So “marginal” is harsh for sure.

            Still you can only go in one direction so it’s not like the Ohio service for example.

          • AJ

            There’s good reason the interstate system skipped Reno-Vegas. Outside of Reno/Carson City and Clark County, Nevada is truly empty. “Rural” overstates the population density.

          • Matthew Hutton

            Yeah but to be fair you can go north towards Salt Lake City on the I15.

          • adirondacker12800

            It’s 420 miles from Las Vegas to Salt Lake City. The biggest metro area along the way is Saint George with a population of 177,000. It’s 420 miles from Saint Louis to Columbus and Indianapolis is along the way. So is Dayton. Keep in mind that it’s very likely anything east of Indianapolis will be there anyway for service to Chicago. Which one should be built?

          • Matthew Hutton

            Why do it? To clear the senate.

            Happy to be wrong, but I don’t think only benefiting 31 states is enough. Especially as some of the states get partial benefits that might not satisfy their senators.

  9. AlternativeTransport

    Very interesting. Have you got travel times for your map?

    I have been working on a theoretical medium speed clock face scheduled system with a two hour takt west of Houston – Minneapolis and a one hour takt east of that. Most of the interchange nodes in the east also line up with a 2h takt but sometimes not. The takt in Montana, Dakotas, Wyoming is thinned out to 3h or 4h. The idea was to create an idealised theoretical system to the standard of a DACH country with 160km/h and 200km/h upgraded lines where possible (if the radii and grades allow it) and occasional new routes with 250 to 300km/h speeds. With acceleration, deceleration, station time, a 15% time reserve the nodes are supposed to line up with each other to enable a network with timed connections in larger cities. This isn’t really the approach of what city pairs would have the most potential but more like a what if scenario the Federal Government under Eisenhower put together a Interstate road and rail package where states could choose what was best, and the existing passenger rail network had not collapsed.

    • adirondacker12800

      There aren’t any people west of I-35, they can get on the twice daily regional jet to a hub airport.

        • adirondacker12800

          No there aren’t. Being very generous the combined population of all of New Mexico, Colorado and Wyoming is 9 million.

          • adirondacker12800

            It’s metro Chicago. Wikepedia says I-25 is 1,062 miles or 1,719 kilometers long. About the same distance as Chicago to Jacksonville. Longer than Chicago to Boston, or Chicago to Dallas. Or Chicago to New York or Chicago to Washington D.C. or anything you want to pick east of I-35 or along it.

          • Matthew Hutton

            The question is how expensive are the flights to those regional airports. Because if they are Ryanair prices then you can’t compete, but if the flights are expensive, well maybe you can make a train work.

            If driving is 40 cents a kilometre and flying is also 40 cents a kilometre, well maybe you can charge 25 cents a kilometre or more for the train – especially if it’s overnight and reliable.

          • adirondacker12800

            you can’t make it that price unless Bill Gates and Warren Buffet decide they want a full scale sized high speed model railroad for the 700 miles/1,100 kilometers between Denver and El Paso. It doesn’t have to be overnight if it’s 700 miles.

          • Matthew Hutton

            If you need federal money you’ll need to do something for most of the lower 48 states to clear the senate. Maybe that’s better buses. Maybe that’s 100mph running on legacy lines. Maybe that’s high speed rail.

            But to be honest whether it makes a profit isn’t that important.

          • Phake Nick

            @Matthew Hutton The flights to regional airports are expensive because of the cost and lack of competition. This condition is created because these airports can barely support any volume of travel demand more than a few regional jet per day, sometimes even fewer. And a few regional jets per day mean like 200 people at most, far from suffficient to justify a high speed rail station even if there already exist a line, let alone building an entire line for them.

            Alternative could be injecting more funding into Essential Air Service program, which would be far more economical.

          • Santa Claus

            Then inject the money into an Essential Rail Service program.

          • Phake Nick

            For what for who? A 1x daily train mostly empty will be a worse service than regional jet coming a few times, and cost massively more. The only people who will be happy will be construction worker.

          • Matthew Hutton

            To be fair from say Saint George Utah being able to get a train to LA, Las Vegas, Salt Lake City, Denver, Kansas City, Oklahoma City, Dallas and Chicago at 300km/h would probably be an improvement over the current service.

          • adirondacker12800

            There aren’t enough people in Saint George to spend 100 million dollars to build a train station. But that would be very very silly because no one is going to be building high speed rail in Utah.

  10. IAN! Mitchell

    I understand this post isn’t about the Deep South, but I was fiddling around with some possibilities between Atlanta/the rest of the eastern seaboard, Florida, and Texas, basically trying to figure out whether there was a way through that would have sufficient opportunity to compete with long drives and connecting flights.

    One that I found which I liked was Atlanta-Columbus-Tallahassee-Pensacola-mobile-new Orleans-Baton Rogue-Houston.

    That stop frequency likely favors value-engineering over highest possible speeds, but I think with the air markets dominated by hub-and-spoke services to ATL and DFW, and with large military bases, universities, and state capitals as well as large populations of single-vehicle or no-vehicle households compared to poverty rate, that there’s opportunity there, particularly in comparison to Orlando-Gainesville-Atlanta, a route I (despite my hometown being Gainesville) consider totally preposterous.

  11. Peter

    I don’t get the stop in Sandusky, Ohio. Sure, it’s right between Cleveland and Toledo, but Sandusky is very small. Can someone explain the reasoning there?

    • Alon Levy

      It’s just on the way and the expected traffic is so high that, worst case scenario, some trains will skip it, same as the stations shown for the Northeast Corridor suburbs.

      • adirondacker12800

        And they will let people who don’t live in Sandusky drive to the station and use the train. Just like they let them drive to the airport in Toledo or Cleveland.

    • John D.

      If memory serves, Sandusky is home to Cedar Point (the famous theme park resort), which could justify a limited service during its season.

  12. adirondacker12800

    Michigan, in contrast, is not on the way between much,

    Detroit is on the way to Toronto. The state of Michigan already owns a decent ROW. There are a bunch of wide places in the road between Toledo and Chicago. They don’t think trains are a communist plot to sap and impurify the precious bodily fluids of Real Americans(tm), as evidenced by already owning a decent ROW. And it’s on the way to Toronto and they already own a decent ROW.

    This, in particular, makes St. Louis-Kansas City and Atlanta-Birmingham, omitted last year, and Nashville-Memphis, suggested last year as a maybe, solid propositions.

    Your are missing things. Indianapolis is as far from St. Louis as Kansas City. It’s very roughly the same size as Kansas City. West of Kansas City there’s a whole lot of nothing. East of Indianapolis there will be high speed tracks to Ohio. St. Louis-Dayton would have more ridership than St. Louis-there’s nothing west of Kansas City. Terre Haute to Dayton, more than St. Louis-there’s nothing west of Kansas City.

    Make Indianapolis-Cincinnati-Columbus, “T” shaped. It makes trips to Chicago faster for more people with less track. And perhaps someday Columbus to Pittsburgh, roughly along 1-70 because Columbus-Pittburgh will carry a lot more people than Kansas City-Denver. ( St. Louis-Pittsburgh is the same distance as Kansas City-Denver. )

    • Henry Miller

      Maybe, but not until the rest of this is operating, plus local transport means people east who drive this now (someone who lives in Ohio going to Topeka, because of the need to transfer ina hub flying is not competitive) no longer have a car.

      Until the rest of the network works well enough to out preform the models it doesn’t work out

      • adirondacker12800

        Topeka doesn’t have commercial airline service. Likely because Topeka isn’t very big and not enough people want to go there. It’s a ten hour drive from Columbus to Kansas City and a five-ish hour flight connecting in Chicago or Atlanta or….

        • Henry Miller

          It is more like a 7 ish hour flight, plus all the hassle of flying and dealing with airports. There is a reason people in the midwest often take 16 hour road trips: you get in your car (which you already have anyway so you just pay the marginal costs of fuel) and relax on a freeway for a few hours. No worry about TSA, missing a connection, and all that, just go.

        • adirondacker12800

          They let people in Topeka travel outside of the Midwest. They don’t have commercial service for those trips either. Driving isn’t relaxing, for most people it’s work. 16 hour road trip gets you beyond the Midwest.

          • Henry Miller

            Compared to flying, driving is relaxing. You go when you want to go, stop (eat, play…) when you want to stop. So long as you avoid rush hour in major cities there are clear highways all over.

          • adirondacker12800

            If you are taking a 16 hour road trip, stumbling across rush hour somewhere is unavoidable. Unless you are driving from Montana to Las Vegas. Likely a morning one and an afternoon one. People take their bladders and bowels along with them on long road trips. I suspect they stop most often for that. Quickly because loitering around in a highway rest stop slows down the trip to the destination.

  13. df1982

    The logic of Mat Yglesias’s is that if you focus initial funding on the NEC and get service down to 90-120min for NY-BOS and NY-DC, then demand will be so high (and operating costs comparatively low), that you basically have a money-printing machine. You can then use the annual profits off the NEC, which would likely run to several billion dollars, to build out HSR to other areas, both as extensions of the NEC (Pennsylvania, Virginia, upper New England), and to other regions, e.g. the Midwest, the South (centering on Atlanta), etc. You turn the vicious circle of present-day rail in America into a virtuous circle of investment and profitability.

    By contrast, if you spread the infrastructure funding across the whole country, then you don’t really have enough to properly fix the NEC, and the improvements you make elsewhere are not transformative enough to make intercity rail profitable. So the whole thing remains a permanently subsidized money sink.

    Shaving 15min off of the NEC timings, plus having two trains a day running from Pittsburgh to Philadelphia in six hours rather than one train a day in seven hours isn’t really going to make much of a difference to anything.

    • Alon Levy

      Yeah, exactly. Starting with the Northeast is fine. The problem is with all the Politics Knowers (by which I don’t even mean Matt, but random trolls at much smaller scale) who say it is Obviously Impossible to just build in the Northeast without spreading the money around, when Chicago-Detroit/Cleveland is right there.

      • df1982

        Well, now this makes the issue you have with Matt clearer (I was struggling to see it from the earlier post). To forestall the “why does the elite north-east get all the money” complaints, you can fund a starter line for Midwest HSR going from Chicago-Detroit/Cleveland (splitting in Toledo I assume), which probably stacks up well in its own right, given the size of the cities, and the fact that there is nothing between them except flat, unpopulated farmland (compared to the densely populated coastal terrain of the NEC).

        Combine these two projects with some federal support for state-level or private initiatives in California, PNW, Texas and Florida, and you have what looks like a national HSR program without dropping money on improving Kansas-Cheyenne service or whatever.

        • Matthew Hutton

          I think an additional phase with a Kansas City – Denver – Salt Lake City – Las Vegas extension and a Dallas – Oklahoma City – Kansas City extension would be a wise thing to include in the initial plan.

          Yes neither might meet a cost benefit analysis but it means you hit all the top 50 metros and means that you hit more of the states and it’s very much benefiting people across the country. And that’s probably essential to get federal money

          I also think you might do OK with the network effect of connecting the east coast/mid west, Texas and the west coast. It’s also a good way to persuade the states in the middle to support rail improvements in the north east as they will support future rail improvements for their states.

          • Eric2

            KC-Denver-SLC-Vegas is basically impossible due to the distances and mountains. You could include it on the map, but it would be an outright lie, unlike Dallas-OKC-KC.

          • adirondacker12800

            Geography is a cruel mistress. Salt Lake City isn’t along the way between Denver and Las Vegas. Not that anybody in their right mind will be building high speed rail to Denver or Salt Lake City.

          • Phake Nick

            It would be wiser to include them as initial “plan” like some sort of MoU without any term that say when such should be funded.

          • Henry Miller

            HSR in Colorado might make sense, but only connecting Pueblo to Cheyenne, and cities in between (mostly Denver). The only reason to get Cheyenne is it is close to the last place in Colorado you would seriously consider, and now you bring HSR to Wyoming which is worth something politically.

            If I ran the numbers correctly it pencils out as an okay route.

          • adirondacker12800

            Your pencil is broken. There aren’t many people in Colorado and two thirds of them live in metropolitan Denver. There are almost no people in Wyoming. The biggest metropolitan, area, around Cheyenne, is less than 100,000. Almost all of the own cars they can drive to the train station in Fort Collins. If there ever is a train station in Fort Collins.

          • Matthew Hutton

            That’s not how politics works. You need to benefit as many states as possible. Now sure in the rural states you’ll build as cheaply as you can and the stations will cost under $10m, but still.

          • adirondacker12800

            Politics works both ways. The people in Wyoming who think trains are a Commmmmunist plot to sap and impurify the precious bodily fluids or Real Americans(tm) don’t want trains. The ones who think all government spending is bad will be against it too. The three dozen people in Laramie who think Cheyenne is the big city, not a wide place in the road along the railroad tracks, will be disappointed.

          • Matthew Hutton

            The rural senators at least aren’t that intractable.

            And it’d bring a lot of tourism dollars in. By car you can do maybe 500km for the weekend, by high speed daytime train it’s probably 1000km and by high speed sleeper it’s probably 2500km.

          • adirondacker12800

            There has to be something for the tourists to tour.

          • adirondacker12800

            Cheyenne isn’t in the Rockies. It’s out on the plain east of them, So is Denver. Where there are frequent flights and a short walk to the car rentals.

        • adirondacker12800

          We tried spreading the money all around in 2009 and yokels from the hinterlands refused to take it.

          • Matthew Hutton

            All you’d be committing to at the beginning would be a few million dollars to study the options as part of the bill to build stuff in the east.

            If it turns out the cost-benefit ratio is 0.5 or better then maybe the trans continental bit would be built. If it turns out to be 0.25 or worse then it probably would be forgotten.

          • adirondacker12800

            Anyone who wants to, can use Google maps to find out Kansas City is 600 miles away from Denver, for free.

          • Matthew Hutton

            Why would you build with a cost-benefit ratio below one?

            Some reasons you might are:
            Benefiting a disadvantaged region
            Votes in parliament/congress
            Maintaining a sense that the infrastructure programme you are doing is going to benefit the whole country
            Safety and other non economic improvements

            In Scotland they are improving the A82 near Loch Lomand with a cost-benefit ratio of 0.21 – https://www.transport.gov.scot/media/50955/dmrb-stage-1-assessment-report-february-2014-volume-1-a82-tarbet-to-inverarnan.pdf (page 68) – main project page – https://www.transport.gov.scot/projects/a82-tarbet-to-inverarnan/project-details/

          • Phake Nick

            B/C below 1 mean the outcome of money you are investing in total is going to be less than the money you put in. Aka if you hand the money directly to citizens they would gain more from the benefit from such rail.

            There are many ways to serve the needs (including but not limited to transportation needs) of people who live in small and faraway places. Rail being a tool to transport many people at once is not a single solution that can cut everything.

          • Nathanael

            If Indiana could be convinced to sell its coastline to Michigan, I think we could get a Chicago-Detroit line. :sigh:

          • Matthew Hutton

            I’m sure most of the road improvements in the highlands have suffered a cost-benefit ratio below one if they are happily going ahead with one with a cost-benefit ratio of 0.21.

            However together they create a perception/reality that driving to and through the highlands is straightforward. This has I’m sure helped the growth of tourism in the region. And that tourism helps the highlands to be more sustainable overall.

            Even if it’s a bit difficult to pin down.

            Being able to do TV ads showing people eating dinner in Chicago and waking up in the rockies or vice versa would be a huge tourism win that’d be difficult to do with existing transport options.

          • Phake Nick

            It is hard to justify a high speed railway with 1000km dedicated new track serving mainly night trains, which will be in the number of single digit per day even in ideal case, and demand will be significantly lower mid-week.

  14. Eric2

    1) Detroit-Columbus and southwards looks awkward. Maybe the route should go due north from Mansfield to form a T junction slightly east of Sandusky to allow for Detroit-Toledo-Columbus-Cincinnati trains?

    2) How do you deal with three-way junctions that meet at a station? For example Toledo or Cincinnati. Is the station effectively a Y (so Detroit passengers need to transfer in Toledo in order to reach one of either Chicago and Cleveland)? Or is there is true three-way junction with complicated flyovers and/or separate platforms for different directions?

    • Alon Levy

      1. Yeah, maybe. But that gets into second-order analysis and requires a less primitive ridership model.

      2. Toledo is on the Cleveland leg of the Y; Chicago-Detroit trains pass near it but do not stop there. Cincinnati I’m less sure about; ideally it should be on the Cleveland leg as well, because Indianapolis-Columbus and Louisville-Columbus are both good trips to serve whereas Indianapolis-Louisville is direct, but the infrastructure around Cincinnati might force it to be on one of the other legs.

      • adirondacker12800

        Michigan didn’t ask you and seems to be aiming for the Y, if Ohio and Indiana ever get over their fear that trains are a plot to sap and impurify the precious bodily fluids of Real Americans(tm), to be in the general vicinity of Michigan City, Indiana.

          • adirondacker12800

            I’m sure lots of people asked lots of questions and got lots of answers, when they upgraded track to 110 mph.
            Or when MIDOT bought railroad. I’m sure almost all of them knew, off the top of their heads, that Detroit is the second largest metro area in the Midwest and if they didn’t, someone told them. And other interesting things like Grand Rapids is a bigger metro area than Toledo and Kalamazoo has almost as much Amtrak ridership as the whole state of Ohio.

  15. Phake Nick

    -If the proposed Atlanta-Charlotte high speed rail that take 2 hours to connect the two cities materialize, how will the travel time of this section change the forecasted demand on the map?
    -Can Texas’s network support extension to Monterrey?

  16. Allen

    It would be cheaper for Mayo to build a campus in the Twin Cities than build a new rail line from Rochester to Minneapolis.

    • Onux

      The point isn’t to build a new line from Rochester to Minneapolis, the point is to build an HSR line from Minn. to Chicago. Rochester happens to be along the way from Madison to Minn. and is larger than Eau Claire so it makes sense to put a station there and serve the market, the same way that Trenton gets a stop between NYC and Phila.

      Whether or not Minn./Rochester can support an HSR line beyond Madison is another story. It is arguably the weakest of the Chicago Hub lines, with Minn. much farther than Detroit, Cleveland, StL, etc., Minn. not as big as Detroit, Milwaukee not as big as Indy, Cleve., etc.

  17. Onux

    I am glad to see that Hartford gets a full HSR line, given its equal prominence to Providence, but the Harrisburg to Pittsburgh routing to connect the NEC to the Midwest is not optimal. DC-Baltimore is now as large as Chicago, while DC is actually farther West of NY than it is south. This means that any connection to the Midwest to “get to Chicago” that doesn’t serve DC is putting the cart before the horse, and that sending trains from NY through DC-Balt isn’t hurting NY service much, but sending trains from DC through Phila. is hurting that service A LOT (a DC-Pitt or DC-Cleve service would start out by going *farther* away from Pitt or Cleve than when it began). The justification is always to get trains to NY faster, but this is a chimera. As noted, DC is as large as Chi, and Cleve is about halfway between them (Pitt is closer to DC than Chi) so if you are building a network to connect the midwest to Chi, you should want to connect it to DC as much. Second, Cleve to NY via Phila. is about the same distance as Chi to Chattanoga or Boston to Richmond and less direct than driving or flying, in other words not a great HSR distance or routing. Killing DC-Cleve service (a viable distance) or marginally improve Cleve-NY service to a still-not-viable distance isn’t a good trade.

    Ideally service would through route in DC then cross the Appalachians around Winchester-Frostburg, however, others in the past have convinced me that a line from Baltimore to Hagerstown, crossing the mountains Mercersburg to Schellsburg is best, representing a compromise that services both DC-Balt and Phila-NY without going too far out of the way for either. Alon’s map has the line from Harrisburg continuing west to around Mercersburg anyway (the Appalachians are narrowest here, so tunneling cost are minimized) so it just doesn’t make sense to lay tracks heading west through the lightly populated Shenandoah valley when you could lay tracks heading west that provide better service to the heavily populated DC-Balt area.

    I have also seen a connector Harrisburg-York-Balt suggested to address this issue so DC trains don’t have to backtrack as far, or even a line Wilmington-Waynesboro-Mercersburg with a similar I-83 connector to Baltimore.

    • Phake Nick

      Eh, the added distance of DC to Pittsburgh via Philadelphia is just about as much as it would be to NYC if routed further south, no way it is “killing” the service by making such routing choice, just like Osaka to Nagoya on Tokaido Shinkansen have to first route North through Kyoto isn’t killing the service either. But from the point of view of travel time, the key market of getting to Chicago is on very edge of competitiveness of travel from any single point of NEC no matter how optimal the track is, and thus any further detour risk exponential reduction in share, hence the largest markets should be served in the most direct way if there’re indeed intention to serve this market, and while DC-BWI is large undoubtedly NYC is still larger.

      • adirondacker12800

        Metro DC is almost exactly the same size as metro Philadelphia. Metro New York is almost 7 times the size of metro Baltimore.

      • Onux

        @Phake, I need time to run numbers but the short answer is that DC-Pitt and Cleve is inside of great HSR distance to total travel capture, but NY is outside the optimal range even with a Phila-Harrisburg routing. Making DC-Balt riders *go backwards* to get to the midwest moves those cities outside of good range, while travel from Phila & NY doesn’t get much worse because it started out as not great to begin with.

        Going to the midwest via DC (or Balt or close to it) isn’t really a detour for NY, remember that while going to DC you are going west the whole time (more W than south from NY!) but going to Phila from DC means going east which is a non-starter for getting to cities to the west. Nagoya to Osaka “goes north” to Kyoto but Osaka is west of Tokyo not south of it. Osaka/Kyoto/Nagoya/Tokyo are almost on a straight line, its the geography of Japan (mountains and coastline) that causes the Tokaido to wiggle the way it does, and the Chuo is being built in a near straight line and may miss Kyoto (a mistake) so your Japanese example is not relevant. The deviation of CAHSR from LA-Palmdale instead of straight to SF via Tejon (universally recognized as a mistake by everyone except CA politicians) is the relevant example. Don’t repeat that with DC-midwest to chase NY-midwest ridership that won’t materialize because NY is so much farther east than DC.

        The DC-Balt CSA is the same size as Chicago, and 2.6M larger than Phila. It is 45% the size of the NY CSA.

        • adirondacker12800

          Click your heels harder. Washington’s MSA is 6.3 million and the 6th biggest MSA in the country and Philadelphia’s is 6.2 million and the nation’s 7th largest. Going to Washington gets you Baltimore with 2.8 million or going to Philadelphia gets you New York’s at 20 million. You might want to try clapping too.

          • Onux

            Per 2020 Census, Wash-Balt CSA is 9,976,383; Phila CSA is 7,214,065; NY CSA is 22,491,979. My statements were accurate, no heel clicking needed. Yes, NY is much bigger, but distance is a factor. NY is much farther than DC and there won’t be huge mode share capture vs air to Cleve, Detroit or beyond. DC can have great capture to the near Midwest, but not if trains backtrack 140 mi East before going west. Adding a little bit of distance to the NY and Phila traffic by going through Balt isn’t as significant because you are not moving the journey time outside the optimal (it’s already outside optimal in the best case) and because the relative penalty is less: every mile towards DC/Balt the train is going west and closer to Pittsburgh, not away from it.

          • adirondacker12800

            and clicking your heels doesn’t change that DC’s MSA is the same size as Philadelphia’s give or take 100,000 or so. Or 7 plus 22 is 29 and that’s a lot more than 10.

        • Phake Nick

          eh, the range where HSR can be competitive is not clear cut. It’s a range. And the distance and speed of NEC to Chicago is also unknown. It depends on what alignment is chosen and what trainsets are used, and even frequency and stopping pattern.

    • adirondacker12800

      Hartford doesn’t need a high speed line. “Faster than driving” to Springfield or New Haven would be good enough.

      • Onux

        People in Hartford want to go places other than Springfield or New Haven. Places like Boston, NYC, and Philadelphia. Being able to get on a high speed train in Hartford that takes them right to these places will mean much more ridership (and benefit to those people) than if you make them ride a “faster than driving” train to Springfield or New Haven and make them change. The same way that there are a stations in Providence and Trenton, instead of telling people to catch a commuter train to Boston or Providence and changing to the HSR there.

        • adirondacker12800

          Why would they have to change trains? There are enough people in Worcester, Springfield and Hartford to have train once or twice an hour, it can just pass through New Haven and keep on going.

          • Onux

            There are enough people in Worchester, Springfield, Hartford to justify a HSR line connecting them to NYC, not just a “faster than driving” connection that continues on. For it to continue on would mean electrifying the NewHaven-Springfield line and why would you spend the money to do that but not upgrade to high speed when there are high speed lines at both ends. Hart-Spr-Wor have 2.9M people between them, which is larger than most Midwest cities (Indy, Columbus) people including yourself assume should get HSR. In fact, the H-S-W complex is almost exactly the same size and spread as Cinncinatti/Dayton. And if you have the Bos-NY and Bos-Albany already planned as part of the network you gain the advantage of connecting those 2.9M for only 64 mi of track (although why would you build the 87 mi of HSR track through the hills from Spr to Albany to connect Bos (4.9M) to Albany (0.9M) before you built 64 mi to connect NY (19.8M) to H-S-W (2.9M) plus Hart (1.2M alone) to Bos?)

          • adirondacker12800

            CTDOT wants to electrify between New Haven and Springfield because, long term, that saves them a lot of money. And makes the commuter train trip slightly faster.
            It would be high speed between Boston and Springfield because the demand in Boston is for Albany and beyond, not Hartford. Not so high speed between Springfield and New Haven and high speed again beyond New Haven.
            It’s one continuous suburb between Springfield and New Haven. Screamingly fast high speed saves 15 minutes and would cost billions and billions. 75 minutes from Worcester to New York versus 90 minutes, or 2;45 to Washington D.C. versus 3:00, isn’t worth billions and billions. It’s still faster than anything else except maybe Marine One landing on the White House lawn.

          • Onux

            “the demand in Boston is for Albany and beyond”

            The demand from Boston to Albany (0.9M and 170 mi away) should be less than the demand from Bos to Hart (1.2M and 117 mi). Beyond Boston? Syracuse, Rochester and Buffalo combine for 2.9M, between 312 and 456 mi from Boston. As I noted above, the demand from NY to the 2.9M people in Hart-Spring-Wor should be more than the demand from Bos to Upstate NY, NYC is larger than Bos, and Worcester is only 196 mi from NY, far closer than Syracuse to Bos (not much farther than Alb to Bos).

            HSR through Hartford would save much more than 15 min. It is a 1h24m journey now. Very reasonable HSR should do it in 40m or as little as 30m. That is a savings of 45-55m, which is incredible significant.

          • adirondacker12800

            Today’s train is slower than driving. I suggested faster than driving. Part of it being slower than driving is that it makes a lot of stops. A through train wouldn’t. Fourth grade arithmetic is a cruel cruel mistress. It’s less than 60 miles from Springfield to New Haven. 40 minutes is an average speed of 90 which isn’t high speed. It’s “faster than driving”. Something they could do with electrification, some curve straightening and maybe some grade crossing elimination. Which avoids spending billions and billions digging a tunnel through solid suburbia.
            Once the train gets to New Haven they can use the tracks that are there for really fast trains between Boston, New York and beyond. Once the train from Boston is in Albany it will be able to use the high speed tracks that are there for the New York to Montreal trains and the New York to Toronto trains. It’s 700 miles from Boston to Detroit. Flying looks better if you are starting out in Boston. It doesn’t if you are starting out not-in-Boston. There’s a lot of New England that is not-in-Boston. And Cleveland. And Columbus.
            There’s a lot of New England that is not-in-Boston and there is even more that is not-in-Hartford. Getting Springfield to New Haven down to “faster than driving” is good enough, they can spend the money other places.

          • Onux

            A train averaging 90mph implies a top speed of 125-150 mph (depending on how express the service is) which is 200-240kph which meets the definition of HSR for upgraded lines. For reference, Acela averages 68mph north of NYC and 72mph south of it. I don’t think you can do this with just elect, curve straightening and grade crossing elimination, but if you can fantastic. Alon has presented plans to get the NEC to this level of service along those lines using a few tens of billions not hundreds of billions. I also suggested traveling the route in 30 min, which is 120mph average, which is 170mph/275kph average.

            New England has a total population of 15.1M, Boston has 4.9M in the metro area, or 8.9M if you include Providence and Worcester. Hartford, New Haven and Bridgeport are about 3M more. There is in fact NOT a lot of New England outside of Boston, what is outside of Bos is mostly along the Hartford/New Haven/Bridgeport corridor to NYC, and what is outside of Bos is definitely not between Springfield and Albany.

            You are correct that no-one is going to fly from Boston to Cleveland/Detroit or beyond. But that means there is little case for a line from Boston to Albany. Long distance travel from the midwest will not materialize, and the traffic between N. England and upstate NY will not be larger than traffic from the heavily populated parts of NE (which includes Hart/Spr/Wor) to NYC and points south, most of which is closer to NE than upstate NY.

          • adirondacker12800

            Vermont, New Hampshire and Maine are never going to have high speed rail. There are no straight right of ways and there aren’t enough people to be carving new ones through mountains.
            There are enough people in Massachusetts, Rhode Island and Connecticut for there to be TWO high speed rail routes. One to New Haven and beyond and one to Albany and beyond. Anything west or north of Albany will already be there for the passengers to and from New York, Philadelphia and Washington D.C.
            60 miles of high speed rail between Springfield and New Haven gets you Hartford. 90 miles of high speed rail between Springfield and Albany gets a similarly sized metro area, Utica, Syracuse, Rochester, Montreal, Buffalo, Toronto and unless you live at Logan Airport or close to it, Cleveland and Detroit.

          • Onux

            The point isn’t connecting Bos to Hartford’s 1.2M, vs 2.9M in upstate NY. The point is connecting 2.9M people in Hart-Spr-Wor to NYC. Buffalo is already 456mi from Bos, the same distance as DC, but with 8.7M fewer people and without 30+M other people in between. Toronto is 550 mi, Montreal 521mi but only 319mi/5hr direct (i.e. not HSR). A 5h HSR ride with a transfer(?) in Albany will not get ridership vs a 1h flight or 5hr drive. Cleveland? Detroit? HSR will not get any ridership at 639-800+mi, stop dreaming. The ridership potential through Hartford to places south on the NEC is greater than the potential from Bos to places much farther west.

    • Phake Nick

      I was going to check out how Alon’s model say about it, but then I find the model say both connection via DC and connection via Philadelphia is going to generate 1000 daily ridership between the city pair of Atlanta and Pittsburgh on both options.

      • Onux

        Atlanta to Pitt is an 1h45m flight but at least 886mi or 6-8hr HSR through DC. I’m surprised in generates any daily ridership.

          • Phake Nick

            People who “can’t fly” are absolute minority of the entire demography, they use Amtrak service now, and together with leisure travellers they support 1x daily train service in most of America. Passenger volume for 1x daily train service is not sufficient to justify constructtion of new high speed rail. Not to mention it doesn’t make sense to maintain the current Amtrak long distance service after long distance high speed rail line enter service, and passengers who really “can’t fly” along such legacy track and are faraway from high speed rail stations will be left with only a feeder bus connection.

          • Matthew Hutton

            I think it’s difficult to know either way how much ultra long distance train routes will generate in traffic.

            In Japan the prices for the ultra long distance train routes are too high compared to flying.
            In Europe the service is too balkanised between the different countries.
            In America Amtrak isn’t reliable enough and is probably too slow.

  18. Rewenzo

    Curious about international HSR to Toronto. How would you recommend handling customs for an HSR train that crosses real borders, i.e. no Schengen Agreement? Currently at Niagara Falls VIA and Amtrak make everyone disembark with their luggage go through customs in the station, walk back outside and around the building to re-enter the waiting room, wait for everyone else to go through customs and then re-board the train.

    • Henry

      Would most likely be solved by station clearance and pre-clearance on the Canadian side of the border, which is what is currently done in Vancouver for Cascades service and is in planning stages for Montreal. Trains that arrive in Canada would do CA customs at the stations on arrival, and Trains departing Canada would have US pre-clearance in the stations they leave from. A similar situation exists in the UK for Eurostar trains, though those trains have about the same amount of stops on both sides of the channel, so customs exists at all stations for pre-clearance, arrivals customs does not exist. Once past customs, there’s basically a mini-station within a station with its own waiting area, restaurants, and platforms while you wait for your train.

      The Toronto-Detroit line would face some issues though if trains would continue all the way to Chicago, as both sides of the border would have too many stations to easily run pre-clearance. It would also make domestic trips impossible on international trains, which would be non-ideal for a corridor like Toronto-Windsor. Realistically, trains would probably terminate in Detroit from both directions (no Chicago-Toronto direct trains), Montreal from both directions, and Toronto for all trains from Buffalo, with all customs facilities for both countries centralized in Detroit, Montreal, Toronto, and Niagara Falls. In an ideal (but, unfortunately, unrealistic) world, the US & Canada would just sign their own version of a Schengen agreement and fix all the messy problems with the border in one go.

    • Sassy

      If you want the international train service to be most useful for domestic trips as well, the way VIA/Amtrak does it might actually be best, assuming you have similar requirements in border controls. If you have border control at the stations instead of the border, that limits what stations people can get on/off the international train at.

      You could probably streamline by having everyone get off the train, but having people done with immigration and customs procedures to immediately get back on the train. However that is a pretty marginal improvement.

      Having immigrations and customs officers board the train instead of getting all the passengers off and back on again, can be better and possibly done in motion. However that gives an opportunity for passengers to potentially shuffle items around to sneak them past officers, likely a completely unacceptable situation for the US and Canada.

      In some world where cross border travel is high enough to support show up and go frequencies despite the complicated border crossing, then just turning trains around at the border might be better. Passengers would get off the train, go through immigrations and customs procedures, and get on the next train.

      Since trains come every 3-5 minutes, turning around trains actually lets the passengers who complete immigration and customs procedures faster get on earlier trains onwards to their destination, and prevents rolling stock from sitting around not rolling while passengers are going through immigration and customs procedures.

      • Matthew Hutton

        The thing is if you have a single train ticket turning up a bit early to go through security costs you maybe 15 minutes. Whereas if you have to get the whole train through security on route it costs you maybe 30 minutes.

        Plus it interrupts whatever you are doing and you have to make sure you don’t forget anything.

    • Alon Levy

      It depends on stopping patterns. If there’s enough time (as in 20 minutes, not hours), they can do passport checks on a moving train; that’s what was done here pre-Schengen. Much of the problem today boils down to border control and immigration enforcement attracting sociopaths at all levels from the minister of the interior down.

      • Rewenzo

        Not entirely sure what you mean. Would they start doing passport checks for everyone on a Toronto-bound train at Rochester? What about people getting off at Buffalo? What about people who get on at Buffalo, which is basicaly on the border? Some people have suggested pre-cleared cars.

        • Alon Levy

          They’d start at Buffalo; the point is to do the checks between the last station in the departure country and the first one in the origin country.

          • Rewenzo

            I guess I’m skeptical you could check an entire train’s worth of people and luggage in the time it takes a high speed train to go from Buffalo, NY to Niagara Falls, ON.

          • Alon Levy

            It’s a matter of throughput. You need a certain number of border control agents to process a certain number of people per hour, and if the headway between trains is not too much longer than the distance between these stations, it’s just a matter of passenger convenience.

          • adirondacker12800

            The Niagara River isn’t that wide, It takes moments to cross the border. Toronto is a big enough origin/destination for there to be two different kinds of trains. Long distance trains where they do customs and immigration on the train while it’s in motion, without stopping, between Rochester and Toronto and regional trains, that make a lot of stops, where everybody does it conventionally in Niagara Falls.

      • Sassy

        If they do border checks on the train, that gives opportunities for passengers on the train to collude to subtly shift illicit goods around the agents, as people who have yet to be processed and people who have already been processed, would be in the same room.

        That was okay for pre-Schengen Europe, but that sounds very not okay for more paranoid North America.

        • adirondacker12800

          They don’t have to be. It would need an empty car. That people move into as they are processed.

        • Phake Nick

          Maybe if the train design can be changed so that it only consists of compartment, like some European trains, then this problem can be avoided.

        • Alon Levy

          >paranoid North America

          Nobody checks anyone’s bags on the buses between Seattle and Vancouver, nor does anyone check cars on Rainbow Bridge. The empty car system, the train holds, the random delays – these are just hostility for the sake of hostility. At some point, a leader needs to tell them “no.”

  19. Dave Peticolas

    I know this post was concentrating on the Midwest, but I see you have a bit of rail in the Pacific Northwest. My local senator is pushing HSR in the PNW and I’m not sure what to think about it. Is it really economical for that region?

    • Alon Levy

      I don’t know. My model says it’s not, but there’s a specific limitation in my model that suggests it’s overrating ridership in large areas (like all of Western Europe if I run it here) and underrating it in small ones (Taiwan massively overperforms the model).

      The issue, though, is that even if it works well, it is marginal – this isn’t the Northeast Corridor. So careful attention must be paid to cost control and value engineering, which is difficult because the legacy approaches to all three major cities suck and so it’s very easy to come up with proposals with 50 km of tunneling in metro Seattle.

      • Matthew Hutton

        I’d guess Western Europe underperforms the model because the international connections suck. And that also means smaller places close to the borders also get weaker traffic than they otherwise would.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.