High-Speed Rail Followup

My high-speed rail map exploded, thanks to retweets on social media from the Neoliberal account and Matt Yglesias, who posted a cleaner map of my proposal made by Twitter follower Queaugie, who even called me a transit guru:

So, first of all, thanks to Queaugie for making this, it’s much cleaner than my drawings on an OpenStreetMap base; I keep advocating for geographically accurate maps, but schematics do sometimes have their uses. But more to the point, I’d like to give some context to why some lines are and are not included.

Some examples of past maps

Mapped proposals for American high-speed rail go back a while. On the Internet, interest exploded in the 2000s, leading to high hopes for California High-Speed Rail and the Obama stimulus. Yonah Freemark made one at the beginning of 2009, which played a role in his rise to become a superstar public transit wonk. The RPA had its own plan rooted in the concept of megaregions: see here for analysis from 2011 and here for a synthetic map. But the map that’s getting the most airplay is by Alfred Twu, which is very expansive to the point of having two transcontinental connections; it was most recently covered in Vox and tweeted by Secretary of Transportation Pete Buttigieg, which generated so much discussion that I chose to crayon US high-speed rail rather than my original intention of picking a city and crayoning urban rail for it.

How my proposal differs

My map differs from past ones in visible ways – for one, it is not connected. At the time I started to make it, I believed there would be four components: Florida, Texas, California, and the general Eastern network. It turned out late in the process that there’s decent demand for Atlanta-Florida travel, enough to justify connecting Florida to the general network. But Texas and California remain disconnected, as does the marginal case of the Pacific Northwest.

Analytically, I project traffic by a gravity model, depending on the product of two metro areas’ populations; Yonah and the RPA have different methodologies. But the emergent difference is, first of all, that I have a less connected network, and second, that there are some glaring omissions. I believe those omissions are justified and would like to explain why – in effect, why other people overrate connections that I do not include.

Amtrak and stagnating regions

New Orleans was the largest city in the South until overtaken by Houston around 1950. This means that the historic rail network of the United States served it amply, as it was large relative to turn-of-the-century America. Amtrak, formed to preserve a skeleton of the preexisting passenger rail network, retained the importance of New Orleans and gave it three distinct long-distance routes: one to Atlanta and New York, one to Chicago, one on the way between Florida and California. This way, there is more Amtrak service to New Orleans today than to Houston, whose metro area is around five times larger.

Proposals tend to build upon what exists. So most people recognize that at transcontinental scale, high-speed rail is uncompetitive, but at the scale of Atlanta-Birmingham-New Orleans it looks like a reasonable line. It should get decent modal split, if built. The problem is that not many people live in New Orleans today. The population one needs to sustain high-speed rail is large, larger than that of your typical early-20th century city. This can be done either via a megacity that drives ridership, as in France or California, or via high population density so that many midsize cities are close together, as in Germany or Florida; the best geography is when both are present, as in Japan, South Korea, China, and the Northeastern US.

The growth of the South in the last 70 years has not been even. Texas has exploded, and so have Atlanta, Nashville, and the cities of the North Carolina Piedmont. In contrast, New Orleans is stagnant. Farther north, on the margins of the South, Missouri has had about the same population growth since 1920 as New York, and has been steadily losing seats in Congress. St. Louis and Kansas City, like New Orleans, were huge hubs for early-20th century America, but their populations are just not good enough for high-speed rail. Chicago-St. Louis can squeak by, but Kansas City is too far. Memphis is in relative decline as well, but manages to piggyback on Nashville, albeit marginally.


  1. Matthew Hutton

    Surely the Great Plains lines could be built cheaper making Kansas City make sense?

      • Eric2

        Southeast Missouri is mountainous (the Ozarks) but StL-KC is very flat. Not quite as flat as Illinois or Nebraska, but flat enough for HSR without any significant earthworks.

      • Andrew

        How would Chicago-Quad Cities-Des Moines-KC-Topeka-Wichita-OKC-Dallas do? (Could also do a spur from Des Moines to Omaha.)

        • Alon Levy

          Very poorly, even Chicago-St. Louis-Kansas City doesn’t justify construction costs, and that’s with a piggyback on St. Louis, not a new line via much smaller Des Moines.

          EDIT: okay, I actually went and checked. Dallas-OKC profits $120 million/year, but construction cost would be more like $7.5 billion, and 1.6% is a tad too low. OKC-Tulsa adds another $100 million/year, but then Dallas-OKC-Tulsa would cost more like $11.5 billion, and then you return 1.9%. It’s very close to 2%, so it could go in blue, but the trouble is that the line would rely exclusively on travel between pairs of decentralized, sprawly metro areas at distances where the main competition to the train is not the plane but the car; even at 2% I’d be uncomfortable adding this to the map.

          • adirondacker12800

            travel between pairs of decentralized, sprawly metro areas
            Which is why their airports are such miserable failures.

          • Alon Levy

            US-wide, the distance at which the modal split between car and plane is 50-50 is something like 900 miles. Midwesterners routinely drive for 12 hours even when flying is available.

          • Andrew

            Thanks! I assume bypassing Tulsa and going to Kansas City still wouldn’t truck?

          • Henry Miller

            What us the potential for HSR in a sprawly city? 100 km of line with 3 stations is 10 minutes between stations, and half an hour off peak driving. Needs TOD near the station, but seems workable.

          • Alon Levy

            I don’t know. I try to distinguish sprawly cities with bad transit, like Chicago and San Francisco, from cities with very weak centers and no transit, like Los Angeles or Dallas. Dallas’s city center is about the same size as Vancouver’s, and the non-city center destinations aren’t really near DART.

          • Matthew A da Silva

            What are the regional variations in how long Americans will drive? NY-Chicago is about 10 hours, but I don’t know anyone who’d do that vs flying. Hell, I barely know anyone who would drive NY-Cleveland or NY-Pittsburgh.

          • adirondacker12800

            Using I-80 between New York and Chicago Google Maps says it’s 789 miles or 1270 km. With a drive time of 12:33. That is an average speed of 63mph. Might be possible if you leave your stomach, bladder and bowels at home. Most people won’t. And most cars would have to stop for fuel at least once. I guesstimate trips at a 50 mph average, it would take 16. Spread that out over three days I might consider it. I would have to be really desperate to do it over two.

          • Alon Levy

            Yeah, NY-Chicago is a flying city pair, but US-wide the 50-50 point on car vs. plane is if I recall correctly 900 miles, because Midwesterners are used to all-day driving with pit stops, and flying between Tiz-al-Nabi and Pizdeloch is indirect and/or expensive.

          • Onux

            “Spread that out over three days I might consider it. I would have to be really desperate to do it over two.”

            You are very much the outlier in this case. I know people who would drive from Virginia to Indiana (600+ mi) without even thinking about it. They brought snacks in the car, they went to the bathroom when they needed gas. People speed, making the trip shorter. For most Americans, especially as Alon noted those in the Midwest, 8 hours driving a day isn’t desperation (like it is for you) it is normal.

          • adirondacker12800

            There are masochists everywhere. I know people who drive to Florida a few times a year. I think they are nutz. It explains why I can fly from Albany to multiple destinations in Florida.

          • Herbert

            Why are you so focused on “return on investment” to the exclusion of externalities?

          • Matthew Hutton

            To be fair my experience of flying in the US is that it’s more hateful than flying in Europe. Plus the roads are quick. If the speed limit is 65 you can average 65 – you can’t in the UK.

          • Henry Miller

            As a mid westerner my in laws are 20 hours driving. Flying is 7 hours (not a direct flight so transfers and eat in an airport ). I’ve done both.

            When we drive I don’t worry about missing the plane. We go when we want to.

            When we drive we make time for something along the way. Flying is about the destination, driving is about something along the way. There are many nice parks in the middle of nowhere that we have seen.

            When we drive we have our car when we get there. Transit in the US is universally bad, so this is important. My inlaws need to take two cars to pick us up at the airport, and then we are unable to go.

            The car plus a few hotel rooms and meals is a lot cheaper than four plane tickets and airport meals. Plus airport meals are generally bad.

            There is no clear winner, but driving has advantages.

          • adirondacker12800

            Apparently you run out of Great Britain before you run out of day. Google says it’s 827 miles or 1330 km from Penzance to John o’ Groats and takes fourteen and half hours.

          • adirondacker12800

            As a mid westerner my in laws are 20 hours driving. Flying is 7 hours (not a direct flight so transfers and eat in an airport ). I’ve done both.
            Gawd it must someplace small with almost no commercial service. Or you obsess about the airport close to your parents. Check the destinations for airports across the Northeast and Midwest, that are secondary or tertiary, the destinations are hub, hub, Florida, Florida, Florida, perhaps Las Vegas and Florida
            My inlaws need to take two cars to pick us up at the airport
            The market for families with six minor children is small. Assuming your in-laws are up to it, they could come visit when the heat index is 110 in Florida. May-October?

          • fjod

            Alon, you say “the social benefits of HSR are proportional to ridership” – but I’m not convinced. The social benefits of HSR depend on what your social priorities are. A crude economic calculation of ‘social benefit’ (i.e. extra wealth generated) means that the high earner’s work trip is worth more than the low earner’s. If you want to prioritise a left-behind region or demographic, trips within that region or demographic have higher social benefit. Is the vulnerable teenager forced into taking the train to supply heroin to the countryside (commonplace in the UK) delivering any ‘social benefit’ at all?

          • Henry Miller

            I ran the numbers, driving costs 3 extra day and saves my family of 5 $400 (would be even more but my youngest is free). Most of the $.54/mile the IRS allows for that trip is a fixed cost that I pay even if the car stays home. While I make more than that, the costs of time are not as bad. When I fly it is a lot harder to stop at a national part along the way, just one such stop more than makes up for any time savings.

            @adirondacker12800 My in-laws live in El Paso: no other airports close, and not even Chicago gets a direct flight. We have had them come visit in the past, but for medical reasons they cannot now (in these Covid days we wouldn’t have traveled if the medical reasons were less serious).

            2 adults and 3 children is all most cars fit (5 passengers), and even that requires special reduced sized car seats. I expect most grandparents would need two cars to pick up family at the airport. Families use SUVs or mini-vans for this reason (in the past station wagons, or larger cars – and they often didn’t put all kids in seat belts which allows for more crowding in the car)

          • adirondacker12800

            Silly me I assumed you were hypothesizing about some place that will have train service someday. El Paso won’t. And it’s been my experience that the kids outgrow the car seats in a few years which mean it doesn’t affect many people for very long.

  2. oevans82

    Thanks for the follow up Alon!

    I’d like to hear an explanation of the Louisville-Indianapolis-Cincinnati triangle. Firstly I am curious why a triangle is warranted here, but other triangular city groupings (Houston-Dallas-Austin/San Antonio; Washington-Pittsburgh-Philadelphia) get a different sort of network topology. Secondly, why did you pick Cincinnati as a vertex instead of Dayton? This seems to prioritize Cincinnati to Chicago over Columbus to Chicago and Indianapolis to Cleveland (and thence to Pittsburgh or New York)

    • Car(e)-Free LA

      I have no idea why Alon chose a triangle instead of a wye in Southeast Indiana, but serving Indy-Cincy-Columbus makes a lot more sense than a wye at Dayton–especially if a wye in Indiana instead of a triangle is selected. Splitting the terminus for Chicago-Columbus/Cincinnati really hurts operational patterns on a not-very-busy route while it is much easier to fill trains which route Chicago-Indianapolis-Columbus-Dayton-Cincinnati. It also allows savings on construction if an Indiana wye is selected because trains can be routed New York-Cleveland-Columbus-Dayton-Cincinnati-Louisville-Nashville instead of splitting service patterns.

    • fbfree

      Wyes vs. deltas is a tricky problem.
      If we consider the case of an good geometry for a wye, three equally sized cities located on the corners of an equilateral triangle, the travel time penalty imposed by using a wye vs. a delta is 2/sqrt(3), or around 115%. Using a gravity model, the ratios of the ridership is approximately the inverse square of this value, or 75%. The relative cost of construction is 1/sqrt(3), or 57%. Therefore a wye would have an ROI 30% larger than a delta configuration in this case.
      However, the total profit may still be maximized with a delta configuration, depending on the exact ROI for the leg. Assuming a 2% ROI is necessary to cover financing costs, profit is maximized for a delta configuration whenever the ROI for the delta exceeds 3.4%. There’s thus a fairly narrow range of ROI, for marginal lines, for which constructing a wye makes sense given this geometry.
      There is a caveat that long-distance travel from outside the triangle is less sensitive to the travel time penalty of a wye.
      As Alon is using a modified gravity model, with a 500 km cutoff, his model will generally prefer wyes over deltas for city distances under cutoff value. San-Antonio – Dallas via College Station is around 500 km vs. around 410 km on a direct line, the particular choice of cutoff he uses may be producing an artifact in the analysis of ROIs for a Texas Triangle HSR.

      • Sascha Claus

        Wyes allow for timed connections at a single, central location (if you want to stop there), like in Herlasgrün. With a delta, you need an integrated, clockface timetable and nodes/knots at each corner (or slightly further out) serve all sides equally.

  3. Richard Gadsden

    Interesting. The city that gets the best connections relative to size (because it’s on the way to lots of other places) is clearly Cleveland here. I wonder if Clevelanders know that their city will gain more from high-speed rail than any other?

    I appreciate that the terrain is very difficult, but a straight line from Cleveland to New York would knock quite a bunch of minutes off Chicago-New York, which I think is marginal and a route where every minute is of particularly high value as a result.

    Possibly a Harrisburg-New York chord could bypass Philadelphia (which would still be served by Washington-Cleveland and Washington-New York services, so would lose no connections) and gain similar time for far less capital cost.

    • RossB

      From a political perspective, making Cleveland (and Pittsburgh, Detroit, Milwaukee, etc.) part of a high speed rail network would be huge. These are Biden areas inside Swing states (although Ohio has swung to the right more recently). Done right, and you not only employ a bunch of workers building this network, but industrial cities get a lot of new business. Businesses close to the train station (right in the city) are suddenly more attractive. I’m sure there are people in Cleveland who understand this — it is just a matter of getting the word out.

  4. Phake Nick

    Would it be justifiable to extend the Pacific Northwest line to Calgary and Edmonton, the California line to Mexicali, the Arizona line to Ciudad Juárez, and the Texas line to Monterrey?

    • RossB

      >> Would it be justifiable to extend the Pacific Northwest line to Calgary and Edmonton

      No. The mountains are huge, and the cities are small.

    • Paul

      I wondered about Mexico connections too. San Antonio – Monterrey is probably possible (metro of 5M, straight line, not much terrain). But anything beyond Monterrey and the border cities would be hard: Mexico has no passenger rail service and the highly populated parts of central Mexico are likely too far south of the border with high mountains in the way.

      • Herbert

        Unfortunately for Mexico, its busiest air route by far is MEX-CUN which is far too far for hsr even absent the effect of geography.

        But there are other routes where hsr would make sense. Peña Nieto cancelled one with no good reason and amlo is running into some problems with his “Maya train”, but the problems to me seem political… And political is always fixable.

        • barbarian2000

          The reason the Maya Train is such a mess is because it’s a diesel line built exclusively for tourist use that goes directly through both rainforest and (as the name implies) indigenous Maya lands.

    • J.S.

      The same can be asked about from Seattle to Spokane (or really the whole of E.Washington). The answer is the same the Cascades are huge resulting in either a really windy, thus slow crossing, or a very expensive tunneling operation to get a straighter line. Both could kill a project because of the time or cost.

      That said there is lots of cross city flights between Seattle and Spokane. I know that Spokane at least has been trying to pitch itself as a cheaper alternative to Seattle for tech companies. Though its hard to win that fight against just driving north/south along the I-5 instead to Tacoma/Everett for cheaper prices.

  5. PeakVT

    As internet maps go, yours is a good one, but of course it is a fantasy because it waves away the politics of having 50 states and assumes hard-nosed rationalism that is rarely in evidence in American transportation planning. Without the PNW corridor, your map touches 27 states, so that corridor will very likely have to be built. A more durable coalition probably requires adding Houston-Baton Rouge-Jackson-Birmingham-Atlanta, even if it is less viable than Dallas-Little Rock-Memphis-Nashville. (I think the geography of New Orleans makes it marginal in just about any case.) That would bring the pork to 32 states by my count, which would probably be enough to get funding through both chambers on a regular basis. Politics will probably put Kansas City-St. Louis-Nashville on the final map as well. But yours is still a good map, especially in contrast to the Twu map, which is hot nonsense.

    • Eric2

      I think I would divide the map into “Stage 1” and “Stage 2”. Stage 2 would include both lower-priority lines (like Pittsburgh-Louisville-Atlanta and Buffalo-Boston-Cleveland) and rejected-by-Alon lines (Houston-NOLA-Atlanta, Chicago-Omaha, StL-KC-OKC-Dallas). This would make another ~7 states feel included, but investment would begin with Stage 1 where it belongs.

      • df1982

        Really though, American HSR activists should focus on the quick wins that can actually show HSR works as a concept. This means city-pairs at distances where rail offers a massive advantage over air or road and with the populations to sustain patronage, and this basically means the following:
        – NEC upgrade
        – California (+LV)
        – Texas
        – Florida
        – PNW
        – And maybe a Midwest hub centred on Chicago (probably starting with Detroit-Chicago)
        Of course, noises are being in all of these regions. But you actually need to have bonafide HSR up and running somewhere first before you start proposing lines like Louisville-Nashville or Buffalo-Cleveland.

        • michaelrjames


          I make that point every time this issue is discussed. Logically it should have been the NEC –which actually has the trains but not the exclusive ROW or good track–but no one is holding their breath for that even though NYC to DC is a puny 360km.
          Texas is notionally private (we’ll see …) but was always a risk to HSR enthusiasts because 1. there are reasons why HSR is publicly funded around the world (and no I don’t accept Japan as a contrary example; let’s not go into that can of worms) as I believe the whole pax-rail experience in the US attests; 2. it’s Texas so who knows about how those deplorables will react even if it notionally saves them a few hours compared to driving; 3. doesn’t terminate in Houston CBD so will be a very poor precedent for the US if it gets built like that.
          Nothing needs to be said about CaHSR … With its long route, challenging geography and complicated politics it was always a risk of poisoning public perceptions of HSR. Despite everything it will still be a huge success if it ever gets completed (ie. between LA and downtown SF).

          Alon has previously made a case for LA-San Diego here, the argument being it is only 204km and long parts of the existing ROW are straight. Alon said it could have a “trip time is about 1:45 or 1:50, which would be competitive with driving even outside rush hour” but I’d say if it was true HSR (the only thing worth building) it could easily be 1.15 making it closer than many LA suburbs by car, and would be a knockout blow over any alternative, especially car but also flying, probably including helicopter!

          But I think a process of elimination leaves LA to Las Vegas as the best bet. At ≈400km it is not the shortest but, at well under 1:45, actually very good to compete with flying. But most of all would be that it would be used by large numbers (in a pandemic-free world) across the classes–in fact compared to most other city pairs it would have a much broader class of (American) users than most other routes which by itself is important in spreading the message that travelling by HSR is a great experience. Also it draws those visitors from all over the US (and the world of course) so it would have impact on the national psyche, in a way the NEC wouldn’t (being a bunch of expense-account elites travelling between a few east coast cities). If smart they could even do a bit of substantial value-capture at both ends, but especially in LV where a mixed residential and shopping precinct around/above the station would have to be a big success.

          • Phake Nick

            1. Using legacy rail to model HSR financial performance would be wrong given how even in Japan virtually all long distabce non-high-speed lines are still lose-making, but that doesn’t stop high speed lines from gaining as much their operation cost as their operation profit
            2. It is Dallas and Houston not people in-between that would be the potential riders of the train, and these cities in themselves are not dissimilar with other US cities.
            3. It’s always a cost vs return analysis on how far a train station should be away from city center, especially in Texas Central’s case where they do not plan to share track with existing system.
            4. That analysis on Las Vegas route you given is a self-fulfilling one, which I don’t think is a good idea

        • Brian Farrington

          Five of the twelve largest cities in the US are in Texas–Houston, Dallas, San Antonio, Austin and Fort Worth. Dallas – Houston is about 240 miles, Dallas – Austin – San Antonio is about 275 miles, and Houston – San Antonio is about 200 miles. There’s already a rail link between Ft. Worth’s and Dallas’ RR stations, although it adds an hour to go 30 miles. The point, though, is that whether you design a triangle or a wye, there’s nowhere else outside of the NE corridor which allows you to serve so many people and big cities so close together. And there are no real topographical challenges. So the Texas Triangle seems like a no-brainer. The problem, however, is not getting from city center to city center. It’s that Texas cities are so spread out and have local transit which ranges from inadequate to completely inadequate. Texans used to driving anyway will not take HSR unless and until there are links to local transit which allows them to get around easily without cars.

          • adirondacker12800

            They let people from the suburbs into the city to use the train station.
            Texas airports have equally bad transit but people still fly.

    • Car(e)-Free LA

      Counterpoint: Alon’s map serves every Biden state except CO and NM and every state with a Democratic senator except MT, CO, WV, and NM. I think non-HSR projects could buy off those states. Why bother trying to flip Tom Cotton?

      • Phake Nick

        This is just a result of cities voting Democrats and cities are where High Speed Rail demands are at. And Republicans representing mainly rural communities aren’t going to benefit from it much so it make sense they aren’t a big supporter of High Speed Rail.

      • RossB

        I agree. There are a lot of Swing states and Democratic states in there. I also think there is value in leveling with America, and saying “sorry, we can’t build everything”. This “best bang for the buck” approach is what makes Biden popular. It is likely that he will get $10,000 of student loans wiped out, even though many in his party want him to go further. This is not only a compromise, but a very good one, as it helps those most in need (which surprisingly enough are those with relatively little debt). The same can be said with this proposal. It goes further than just serving the Northeast (which is what many have suggested) and as a result, serves many swing states and Democratic states, while avoiding the coast-to-coast silliness.

        • Herbert

          HSR is very much a thing that once you get the ball rolling, it has a certain momentum to it and bit by bit “marginal” lines become “politically conceivable” to “politically necessary” to “built”

        • Herbert

          You don’t start with the compromise. You start with the impossible and haggle from there

          • Henry Miller

            You need to know what compromise is acceptable to haggle to. Otherwise you risk agreeing to something like Minot ND to Spokane Washington. Probably a good distance for HSR, but between the lack of people along the route, and the mountains it is one of the worst routes I can think of.

            Also if you aim too high you get rejected outright instead of get any comprise.

          • adirondacker12800

            Bismark is the other metropolitan area in North Dakota after Fargo. There are only two in North Dakota. Minot is a micropolitan area Bismark would be a better choice. Until the oil industry collapses. It could go through Billings and Missoula! The two places in Montana that manage to have enough people for the Census Bureau to call them metropolitan.

          • adirondacker12800

            The city buses run once an hour in Minot. Which is probably excessive for a region with 75,000 people in it. What do you think the chances of there being a busload of people who want to take a 900 mile ride to Spokane?
            Greyhound has a bus a day from Bismarck to Spokane with a change of buses in Billings Montana. A bus, it only offers one option. Apparently Minot is so small there isn’t Greyhound service.

          • Henry Miller

            That is why I said Minot: I wanted to come up with the most worthless possible compromise, and Bismark would get too many riders for my point. I’m sure there are worse cites in Washington to choose from, but I happen to have family in Minot so I know how bad a choice it is, while I can’t think of any cities in Washington.

          • adirondacker12800

            Ask Google the right question it’s much easier to come up with the answer.
            Apparently Amidon in Slope County which has a population of 727.
            Works on Canadian Provinces too, asking for statistical areas.
            If you want to limit it to places that manage to make it Metropolitan status it can be helpful.

    • Lee Ratner

      Many of Amtrak’s existing lines come from nearly every state needing to have an Amtrak line in order to justify it’s existence even if these lines get little used. There is no way that a big chunk of states are going to voluntarily go without HSR.

      • RossB

        A big chunk of states won’t vote for anything Biden proposes. I mean anything.

        Like most legislation it will pass the House along party lines. As the previous commenter noted, you only need the votes of a handful of senators. Most of them will support it on principal, or with the agreement that they get other projects. Manchin could be tough, but the president has dealt with him before. That assumes that no Republican senator — even those from Texas and Florida — would vote for this. That would be a political gamble, and if nothing else make it easier to run against them. It is easy to argue against a coast-to-coast high speed rail line (I could do that, easily). But if I’m from Texas, I would hate to run against this map.

        • Phake Nick

          It would be easy to run against this in Texas if your main voter base isn’t near any of the four cities on the map – And that’s how states like Texas are still dominated by Republican despite those cities mainly vote Democrats. As you can see from numerous local politicians in Texas are now voicing their oppositions against the Texas Central.

          • Car(e)-Free LA

            Mhm. But who cares? Because Biden+50 Dems can force it through anyway. What Texas wants doesn’t actually matter, politically.

          • Herbert

            The question of Texas turning blue (absent GOP shenanigans against voting rights) is not one of “if” but one of “when”. Certainly the recent winter storm and its abysmal handling by Texas Republicans hasn’t helped their case…

          • Phake Nick

            Again, the divide stem from demographic and urban-rural divide. Individual political event might change individual political view but that doesn’t change the representation.

          • RossB

            In close races, individual political events can make a huge difference. The Georgia runoffs are a good example. It is quite possible that the Democrats took both Senate races because of Trump’s behavior after the general election. I’m not the only one who feels this way. Political experts do as well. To quote Micah Cohen, from 538.com:

            “I guess what I’m arguing is that (i) yes, most partisans are locked into their views, but (ii) there’s a small share of voters who do view events like Jan. 6 and Trump’s role in inciting the insurrectionists as beyond the pale and move accordingly, and that (iii) there’s enough of those voters that they’re important electorally.

            … Democrats won both U.S. Senate races in Georgia because of this kind of thing.”

            Texas is rapidly becoming a swing state. If Cruz (who won by 50.9% in an off year election) loses additional votes in those five cities, then he could very easily lose the election. A small event (like not supporting high speed rail) could be the difference.

  6. DB

    Why no Montréal (4.4M) – Québec City (0.8M) and no Calgary (1.5M) – Edmonton (1.5M), perhaps at least as thin gray lines? City size, other factors?

      • df1982

        Vancouver-Portland has nearly 10m people in a 500km straight line, with Seattle bang in the middle, and it’s growing fast. It’s perfect for HSR, and has a lot more going for it than Atlanta-Jacksonville or some of the other corridors you have there. In fact both legs would outperform Berlin-Hamburg for ridership on the basis of population/distance.

        • Phake Nick

          It would still be less than the Dallas to Houston line, which have 15 Million population, and according to estimate by the author in linked post, the ROI of that Texas line would only be 1.2%, using cost estimate from Texas Central. Pacific Northwest will be less than that.

          • df1982

            You’re right, Dallas-Houston performs better. In fact I think it would have an ROI of a lot more than 1.2%, given that it has attracted private capital to construct a line from scratch, supposedly without any government subsidies (probably the only example of this happening in the world, since Miami-Orlando is largely re-purposing an existing freight railway, and LV-LA will require publicly funded HSR to actually reach the LA core). But PNW is up there. A model that downplays both of them is a wonky model.

          • Herbert

            If you want your rail line to earn back infrastructure cost with nothing but fares and don’t include any other costs or benefits, you get a wonky model with wonky results.

            It’d be interesting to see what this map looks like with a 50$ a ton carbon cost or recouping 30% of positive externalities…

        • Phake Nick

          JR Central behind the project is themselves building a new high speed maglev across the Central part of Japan using entirely private funds, although government-issued loan have been considered to speed up the construction time.
          Texas Central still have gap in funding source pending to be identified, and from what I have read, they wouldn’t reject government funding if conditions are appropriate.
          As for demand estimate, I think the estimation on this blog is in fact even slightly more optimistic than Texas Central themselves, since this blog anticipate “2 short tph Houston-DFW”, while Texas Central’s official information say they will only offer 2 short train per hour during the peak hour, and off-peak they will only offer one train every hour. And their trains will not just be short but also use spacious seat pitch with 2+2 seating in same cabin as Japanese Shinkansen in order to attempt attracting passenger using comfort factor by sacrificing capacity.
          The variable here is the construction cost and fare rate.

          • Herbert

            Adding service to a line that is not at capacity is pretty cheap.

      • adirondacker12800

        The distances have a bit to do with it. Google says it’s 2100 km between Toronto and Winnipeg. Sudbury and Thunder Bay are along the way. Slightly shorter if you go through the U.S. I suspect the Canadians aren’t terribly interested in high speed rail to Duluth. Or Michiganders

        • Henry Miller

          If you are in Thunder Bay, where else would you go? The city isn’t extremely big, and everything close enough to be worth thinking about is in the US. In Minneapolis the big retail destinations bring in extra staff for Canadian Thanksgiving. If you have HSR to Duluth, connecting to Minneapolis, and onto Chicago is a no brainer.

          Seniors in Minneapolis sometimes travel to Canada for their prescriptions, although most of them now go to Fort Francis (probably the first time you have heard of that tiny city), HSR could attract them.

          In the end though the population just isn’t enough. That is before you account for international links under performing. Worse, most of the proposed line is in the US, but there is hardly even any small towns along the line. The US ends up subsidizing a line that is mostly used by Canadians, so it is really hard to find political cover. Most of the Canadian shoppers are traveling from Winnipeg.

          • adirondacker12800

            There are so few people out there that the Trans Canada highway isn’t even a four lane – two in each direction – limited access highway. It looks like this.

      • Ryland L

        Alon, again what is the threshold of Large/small. Sweden does high-speed-lite between Stockholm (pop: 1 million+) and Malmo (pop: 316K). Finland was considering HSR between Helsinki and Turku (just 200K)!

        • Alon Levy

          Sweden is starting to build HSR, yes, but my impression is that even within Sweden people treat it as a marginal value proposition, even with future growth, TOD, and decent interregional connections. (The existing trains have a timed transfer at Alvesta.)

          • Herbert

            Sweden is rich, has comparatively easy geology and a population that wants to shift from air to rail. That to an extent balances their geographic disadvantages.

            On the other hand, it is interesting just how much passenger ferry traffic the Baltic Sea has and that only some of those routes are in any serious stages of being converted to fixed crossings (Øresund built, Fehmarnbelt to be built, Talsinki planning)

          • Sascha Claus

            Many of these routes (Ystad / Trelleborg – Germany / Poland, let alone Germany – Finland / Baltics countries) seem to be way out of the usual fixed-link range. Maybe something for April, 1.

  7. adirondacker12800

    Houston-Birmingham-New Orleans it looks like a reasonable line.

    There are almost as many people in Metro Charlotte as there are in the state of Mississippi and more people in metro Atlanta than there are in Louisiana. Birmingham-Atlanta makes more sense than Nashville-Memphis or Birmingham-New Orleans.

    • Ryland L

      Also, the strong pattern of migration from Coasts to Texas (much of it in the mid- to lower-income brackets) could justify a line that integrates Texas into eastern network

      • adirondacker12800

        The stereotype is Rustbelt to the Sunbelt. Saint Louis, on the southwestern edge of the Rustbelt, to Dallas is 662 miles. Through Little Rock! It’s 179 from Texarkana at the border between Texas and Arkansas to Dallas. The distances are huge.
        It’s a two hour flight on American non stop to DFW or 1:45 on Southwest to Love Field.
        Saint Louis isn’t very coastal. Philadelphia to Dallas via Atlanta is 1560 miles. Through Shreveport not New Orleans. Los Angeles to Dallas is bit better. 1437. Through Abilene not San Antonio.

  8. adirondacker12800

    Cheyenne? Why not Laramie too?
    Be really generous and call it 182,500. There isn’t much of anywhere to go except Denver, two round trips a year for all of them?
    ….. or 365,000 a year. Average of 1,000 a day. One 1,000 passenger train or lots of buses that go to the conventional rail station in Fort Collins? If Fort Collins ever gets rail service.

    • Herbert

      Interestingly enough the entirely arbitrary 1000 pax number is what Bavaria uses as a threshold for rail reactivation

      • adirondacker12800

        Fort Collins has active railroad

        So do Cheyenns and Laramie. 182,500 people is being overly generous and they aren’t carfree Japanese

  9. joeshupac

    Good stuff! Very interesting about New Orleans and the way that US population patterns have changed a lot more than rail patterns.

    I see that the Lake Erie area is the only place where two lines run closely parallel to one another. If you were forced to pick just one, would you pick the Ontario side of Lake Erie or the American side?

    That also made think of a hypothetical question. If, for political and/or technological reasons, it becomes common for people to take fast ferries or very short airplanes or even ground effect vehicles from Florida to Havana, Toronto to Niagara/Buffalo, or Milwaukee to Michigan, how might this impact the map, if at all?

    • Alon Levy

      The Canadian side is stronger. The American side connects Cleveland (small) and Chicago (far) with Upstate New York (small), the Canadian side connects Detroit with Toronto, which the model thinks is stronger than Toronto-Montreal.

      • adirondacker12800

        Cleveland-Buffalo connects Cleveland to Toronto and Boston and at the outer edge of train trips, Montreal. Boston-Cleveland Is a bit iffy if you are staying in a hotel at Logan but places other than Boston itself getting to Logan comes in to play. Last time I checked Hartford to Cleveland, keep in mind that apparently Hartford is the center of New England’s universe, there were two flights a day at “we are the monopoly airline” fares. I’ve never checked PVD to CLE.
        If there wasn’t so much border drama, Buffalo-Detroit is much shorter through Ontario. Google road distance is 707 from Detroit to Boston via Hamilton Ontario. That might be interesting to more than a few people. Especially if you are starting out someplace other than Logan Airport. 808 through Cleveland and Toledo on the south side of the lake.

        • Phake Nick

          Cleveland to Toronto can also be connected via Detroit. Cleveland to Boston can also be connected via Pittsburgh. So as any destinations beyond.

          • adirondacker12800

            They could take a grand detour around Lake Huron and meet the Toronto-Winnipeg line in Sudbury too. I think is very very unlikely the Canadians are going to get the urge to build a line between Winnipeg and Toronto because that’s 2100 km.
            Road distance between Cleveland and Toronto via Detroit is 400 miles/ 643 km. Or 292miles/469km via Buffalo. Buffalo was lousy with railroad yards back in the day. From satellite images it looks like the route from the west to Niagara Falls has been erased. That might be a difficulty. Cleveland-Philadelphia-Boston is 740 miles/1191km or through Buffalo and Albany 640 miles/1030km. And avoids all the congestion between Philadelphia and New York. Well Harrisburg and New Haven.

        • Phake Nick

          Yes, travelling from Cleveland via Detroit to Toronto will take an additional half an hour, probably a bit more given the need to cross border but dependent on which method of border crossing is chosen, but when one want to prioritize this lost would be less than longer link from Montreal – Toronto to Detroit – Chicago

          As for capacity problem East of Pittsburgh, the proposal only anticipate 8 trains per hour there. Still ample capacity to add additional trains. Spare line can be build after other line actually hit capacity limit in operation.

          • adirondacker12800

            Crossing the border will take time whether you are doing it via Windsor Ontario or Niagara Falls Ontario, call that the same.
            It’s going to take longer that half an hour to go 100 miles/160 km if the train is actually going to stop somewhere along the way.
            There’s 12 trains an hour passing through Trenton NJ. 8 from Washington DC and beyond and 4 from Pittsburgh and beyond. 12 may be too low. There are all those pesky Philadelphians only 90 miles/150km from New York. That has astoundingly bad traffic and no place to park. And tolls, lots of tolls!

          • Matthew Hutton

            Surely we can agree that Niagara Falls will get enough custom to warrant a station as a major tourist destination?

          • adirondacker12800

            You take the train to goes to Niagara Falls New York from U.S. origins and the train to Niagara Falls Ontario from Canadian origins. The train that goes from Toronto to Boston or Washington D.C.or vice versa, can seal the doors between Toronto and Rochester. So they can do customs and immigration while the train is surfing along at high speeds.
            There’s enough demand around the western end of Lake Ontario to have two different kinds of trains. Express trains to other places and local trains serving the local markets.

          • Alon Levy

            Yes, you could do that. Or you could avoid overly splitting the market just because of CBP butthurt and process people faster.

          • adirondacker12800

            Doing customs and immigration while in motion has a wait and processing time of zero. Processing hundreds of people all at once needs platoons of agents.
            Downtown Buffalo is unfixable. Slow down to go through curves to mindlessly serve downtown Buffalo ( I’ll give you hint here. people traveling to or from places not Buffalo are blissfully unconcerned about Buffalo. ) and then make everyone loiter around while border control does its things wastes too much time. Don’t fix it, run two different kinds of trains.

        • Phake Nick

          > Crossing the border will take time whether you are doing it via Windsor Ontario or Niagara Falls Ontario, call that the same.
          Not necessarily. The distance between Niagara Falls and Toronto is short enough and without much other things in the way, that it might be possible to adopt the sort of border control similar to Eurostar, where passengers only need to clear the procedure at the terminal in London (or in Toronto in this case), omitting the need of interrupting the train at national border and allow individual passengers to continue their journey using transit option in destination city immediately after clearing the border control instead of having to wait for everyone else come back to the train.

          > It’s going to take longer that half an hour to go 100 miles/160 km if the train is actually going to stop somewhere along the way.
          Fair, but not necessarily. Especially since you’ll also be combining the service with service heading toward Detroit and thus increasing the frequency offering, providing more room for trains with different stop pattern class

          > There’s 12 trains an hour passing through Trenton NJ. 8 from Washington DC and beyond and 4 from Pittsburgh and beyond. 12 may be too low. There are all those pesky Philadelphians only 90 miles/150km from New York. That has astoundingly bad traffic and no place to park. And tolls, lots of tolls!
          Commuter railway usually have more traffic than long-distance railway. And “4 New York-Pittsburgh-Cleveland, 4 Washington-Philadelphia-Pittsburgh-Cleveland” totalling to 8 trains per hour was from the original estimation posted on the blog.

          • adirondacker12800

            What part of “Trenton NJ” got you confused? The trains from Harrisburg and beyond to New York will be on the same tracks through Trenton NJ as the ones from Washington D.C. and beyond to New York. 8 from DC and 4 from Harrisburg. 12 maybe too low but there can be trains that make all the major stops between New Haven, Wilmington or Harrisburg, that weave themselves between the express commuter trains. Or use the West Trenton Line. Someday far in the future when there are six tracks under the Hudson River to Penn Station New York.

  10. tvdp

    Can you explain more about the PNW line? How does population growth factor in? Isn’t Seattle/Puget Sound a space where lots of people will move? Moreover, given that HSR will take over a decade, how should we factor climate change into long-term plans?

    • tvdp

      Additionally, how do we factor in transit at each location? One reason Seattle is such a great destination (as are NYC, DC, Boston), is you can get from the train station to your destination without needing a long car ride.

      • Phake Nick

        It is modelled after Japanese and European cities with high speed rail coverage, according to my understanding.
        Even NYC, DC, Boston have relatively great transit, most of them still seems less comprehensive than many Japanese or European cities, so I doubt that would give much bonus over the model

  11. Benjamin Turon

    Overlaying this map of “Very High Speed Rail” lines with “Higher Speed Rail” corridors would be interesting. Just because you can’t justify a new build 200-mph rail line doesn’t mean you can’t justify a upgrading of existing rail infrastructure for a 90-125 mph high frequency service. For example the Boston-Portland ‘Downeaster’ is a corridor that doesn’t need 200-mph bullet trains, but you could boost average speeds up to what they were during WW2 and even better, a intercity service on par with London-Birmingham by Chiltern Railways.

    • Alon Levy

      90-125 mph is what we call low-speed rail here. The average speed of lines with American marketing names like “higher-speed rail” or “emerging HSR” is low and actual HSR should not be brand-poisoned by association with them. Boston-Portland is a great low-speed line, EMUs extending the Haverhill line could do it in around 1:45. Ditto New York-Allentown and, at lower average speed because mountains, New York-Scranton-Binghamton. I didn’t depict them on a national HSR map, but I did make those New England and Upstate NY maps, it’s just complicated to draw an entire system with timed connections over a region so big.

      • adirondacker12800

        When Binghamtonians can get a bus to the HSR station in Syracuse, going to Syracuse by bus will be faster than taking a train through Scranton. Or just drive the enormous parking lots at the station next to the airport. Next the airport may not do much for them but it doubles the frequency of the bus for places like Auburn, Ithaca or Watertown And it avoids digging up downtown.
        Electrification and other improvements will make it better than this but not a whole lot. …..Bus to Syracuse..

        There can be froth from foamers about Philadelphia-Allentown-Scranton-Binghamton-Watertown.
        Allentown/Lehigh Valley is a million people, Scranton/Wyoming Valley is half a million, Binghamton is a quarter of a million and Syracuse is three quarters of a million. It’s 250miles/400 km of mountain. …. and if the high speed train to New York does that in two hour it means it’s less than three to Philadelphia going that way.

        If there are 20 intercity-ish trains an hour at Penn Station New York you can’t time connections. Or need to.

  12. Anton Babadjanov

    The PNW has near-hourly flights between its cities and multiple private bus connections. Ridership stats on the low-frequency and very low average speed Amtrak Cascades service just don’t tell the whole story.

    Seattle is one of the fastest growing cities in the US (population wise and economically). Vancouver is in the top 10 tourist destinations in the world. One of the best lines on the entire map is the PNW line and a gray line can only be drawn if one doesn’t have full information about the region (didn’t you live in Vancouver, Alon?).

    It’s not a line that would need true high speed, but still definitely a new ROW and higher frequency (e.g. 1 tph rather than 2 tpd). A new ROW will by default also increase speed to a significant extent (without it being the goal). Current average speed is in the 40 – 50 mph range (which is also why current service ridership is barely a useful indicator) and a new ROW will push it in the 80 – 100 mph range (avg speed).

    Also urban transit mode share in those cities is better than many other city pairs on the map, and that also feeds into future ridership projections.

    I am waiting for an update on the PNW line from Alon.

    • Alon Levy

      Yeah, I lived in Vancouver, and was stricken by how small it was compared with New York or even Boston. I’m not basing my analysis on Amtrak, hence the omission of the NOLA hub and the different route used for Chicago-Detroit; I’m basing it on raw metro area populations, with data trained on existing Japanese, Spanish, French and to some extent German and Korean ridership.

      The problem with the Pacific Northwest is that it lacks the drivers of ridership on this side of the Pond. Population density is low, which in practice translates to there not being many cities around – there are just three and they connect to nothing else, whereas what may superficially look like a connection between three cities in Germany also speeds up travel to many other nearby cities. Now, you can support HSR on lower density, as in Spain, but Madrid is bigger than Seattle, Barcelona is a lot bigger than Portland or Vancouver, and the connection between them gets about an hourly train.

      • df1982

        Berlin-Hamburg works without any significant population centres in between (or that much in the way of through-traffic), and that city-pair is comparable to both Seattle-Portland and Seattle-Vancouver.

        • Alon Levy

          It’s not an NBS but a legacy line running tilting trains, with express passenger rail priority and much better right-of-way geometry than on the Cascades. Moreover, there is no proposal here to replace it with an NBS to cut the trip time to an hour, and rail advocates here do crayon things like Erfurt-Frankfurt, Hamburg-Hanover, and Ulm-Munich.

          • df1982

            A lot of money was spent on upgrading the line in the 1990s, which had originally been built to high standards precisely because it was such an important corridor (it had the fastest regularly scheduled train in the world in the 1930s), but fell into disuse during the Cold War for obvious reasons. Since the upgrade trains can do the journey in 90min at an average speed of 190km/h (although they presently run somewhat more slowly because Deutsche Bahn) and demand is high enough that it will have a half-hourly frequency in the Deutschlandtakt.

            There’s no need for a brand new line that would only shave at best a half-hour off the travel-time (compared to something like Erfurt-Frankfurt, which would have a hugely beneficial effect on Berlin-Frankfurt travel times), but it shows that high demand is already there for a city-pair that your model seems to dismiss as insufficient for a high-speed connection.

            Portland-Seattle and Seattle-Vancouver are pretty much perfect matches for Berlin-Hamburg in terms of both population and distance (and the proposed HSR journey times would be similar to what is already possible on the Berlin-Hamburg line), and indeed each leg would probably outperform Berlin-Hamburg because they would also have Portland-Vancouver passengers boosting ridership.

  13. Eric2

    It seems all the calculations here depend on 1) a certain level of construction costs, 2) the expectation of economic profitability.

    If construction costs were cut in half, then a 1.5% return would become a 3% return and an unviable line would become a viable line.

    If costs remain the same but the government gives developers a 50% subsidy of construction costs (let’s say to combat climate change), similarly a 1.5% return becomes 3% and the line is viable.

    So with some reasonable changes in the starting assumptions, you can get a lot more (or a lot fewer) viable lines.

    • adirondacker12800

      Or the government could build it and charge users fees like we do with roads and airports. Vaguely sorta kinda. But Real Americans(tm) are convinced that trains are plot to sap and impurify their precious bodily fluids turning them into Communists, so that is unlikely to happen.

    • Henry Miller

      Too bad Alon’s study on costs focuses on underground. It is not clear if that will give us anything useful to reduce HSR costs, which should be limited tunnels in general. I’d really like to see Spain level costs here because it makes a lot of things viable that nobody would never consider at current prices.

    • Alon Levy

      Yes. But take the construction costs I’m assuming and cut them in half, and you get to levels that don’t exist even in Spain. And if the government chooses to offer subsidies to intercity rail then by definition the financial ROI falls; you absolutely can look at social benefits and social ROI, but then the minimum ROI needed rises as well because you have to evaluate current costs vs. future benefits at the rate that general society does, which has a higher rate of return than long-term government debt.

      • Phake Nick

        I think in some countries, there are projects being constructed even if ROI isn’t possible, as long as the benefit/cost ratio (B/C) is over 1.0, with the “benefit” being calculated, iirc, by estimating the total amount of travel time reduction for anyone expected to travel on the line, and then calculate the time cost

        • Alon Levy

          Yeah, that’s part of social ROI – I think in those terms and not BCR first because I have concrete ROIs for French LGVs, and second because I think it’s more honest to plan upfront cost vs. long-term benefits this way.

    • Adam

      Here’s a good example of how contractors bake in extra costs for at grade fill or viaducts (not tunnels), by overbuilding concrete superstructure.
      For example, as Clem Tillier recently pointed out, by increasing the structure size (depth of a bridge understructure from soffit to rail) from 1.67m to as much as 5.2m, a contractor can pad a contract to be worth hundreds of millions more.

      Why? because small things add up fast.
      a fatter structure has to have deeper and wider footings/bents –takes longer to excavate those.
      deeper wider footings/bents often ‘successfully’ find more utilities to relocate. takes lots of time and money to remediate the utility relocation problem you caused yourself by having fatter structures!
      a fatter structure takes more materials, and that means more profits for all the contractors. and longer to build and more laborers and machinery rentals! the money fountain never ends does it?
      a fatter structure will requires more embankment and fill when it leaves and returns to grade, A LOT more, that’s more time and trucks and labor to tamp down all those extra meters of embankment you “forced” yourself to build.
      and you guessed it, if your fat structure is vertically higher, you now have to build longer approach ramps to leave and return to grade, now acquire additional property because of that. It’s really a money waterwall more than an fountain…

      Not to mention the rider experience is made much worse at stations like Redwood City as Tillier describes because of the higher platform elevation from fat structures (for example)

      So yeah it’s VERY easy for contractors to force costs to always increase 50% from one in-kind project to the next in-kind project even for at grade and aerial options. they just have to perform all their little ritual tricks such as adding a half meter of fat to the depth of structure and the money tsunami will turn on again with none the wiser from politicians and the public.

      Regulation could help, but sometimes regulation caused the overbuilding problem to begin with. Sometimes the regulation has been captured and the result has been to force overbuilding because overbuilding results in a money tsunami for those who captured the regulation.

      A better path of reform might be for advocates to emphasize heavily how overbuilding bridge depth to 5m (from Tillier’s example) produces much more construction carbon emissions in a huge variety of ways (more concrete for larger structures, more heavy truck trips, more heavy equipment rentals) than a baseline minimal environmental impact, minimal carbon emissions with bridge depth 1.67m standard.

    • michaelrjames


      Our clocks are synchronized again! I think that is a valid point. Plus:

      “… but the government gives developers a 50% subsidy of construction costs…”

      Or even 100% like the Interstate Highway System. The federal gas tax funds about 57% of IHS maintenance, (zero capital repayment), and the remaining ≈40% of costs (maintenance) is pure tax expenditure (ie. “subsidy”) by the states and local gov. Not saying the IHS shouldn’t have been built or that it shouldn’t be subsidised, but probably not at the extent that it has been. But the hypocrisy amongst its users is legion because they are heavily subsidized. (There are some parts of the IHS that are toll-roads but only a minority.) This $66bn per year (maintenance only) is part of the $177bn spent on highways (including construction of new ones) in 2017 according to the CBO. Just stopping spending on building new ones–or just by half–would save enough to build a HSR network over say a decade or two.
      By contrast the ≈12,000km autoroute network of France is almost entirely tolled and was built privately with mixed financing (city and provincial govts often took a share). Of course fuel taxes are much higher but they don’t fund the autoroutes, instead part-funding the ≈30,000km of N-(nationale)-roads which are the older, slower state highways one can always use instead of paying the tolls. Perhaps it’s a factor in why France can afford an excellent HSR and rail network in addition to one of the world’s best road networks?

      Bottom line: it’s ok to take the rather simplistic ROI approach that Alon has applied here as some kind of calculable metric but it’s not the only factor in choosing to build major national infrastructure. I’m not being snarky. If the IHS had or was assessed on the equivalent basis (instead of mostly subsidy by all taxpayers) it would have been more logical and fair to have it tolled. It would add a few cents to transported goods and is not enough to prevent private cars using it (as in France, overusing it). In fact in the coming age of EV, tolling will be inevitable; then maybe the US will be able to afford road upkeep …

      • adirondacker12800

        tolling will be inevitable
        Or they could tax the electricity the car uses. Or the distance it drives. Without a complicated tolling system that reveals where you have been.

  14. Matthew A da Silva

    This uncomfortable reality on HSR viability begs an interesting question, what to do to better-connect stagnating regions to national economic growth? Having whole regions of the US “left behind” creates long-term equity issues – and there is only so much “moving to opportunity” that can be done. Obviously, the current reality of interstate highways + non-hub airports for these areas is not a sustainable approach for greening our transportation sector, and these areas aren’t depopulating anytime soon (nor should they).

    I think there is a strong case to be made, particularly in the Eastern half of the country, for high-quality conventional rail connecting many of these regions to the nearest HSR hub for access to the social, economic, and medical services those hubs provide. These secondary cities often do have important niche contributions they can make in turn, whether tourism-related, or educational, or logistical.

    This brings me to the next idea for your crayon – What would the next level of service hierarchy look like on your map w.r.t speed, frequency, and destinations served?

    • Eric2

      “and there is only so much “moving to opportunity” that can be done.”

      That’s not really true. Moderate upzoning could house all Americans in a small number of coastal metropolitan areas at affordable real estate prices. For example, if NYC (the city, not the metro area) were upzoned just to the density of Paris, that would allow ~8 million people to “move to opportunity”. Repeat for the other top cities and metro areas, and everyone gets opportunity.

      • Matthew A da Silva

        The limits on moving opportunity aren’t economic, they are cultural. People have noneconomic reasons tying them to areas, whether it is taking care of elderly/disabled family, needing to supervise a family business etc. As a result, secondary centers will always be important as stepping stones to opportunity.

        Think about all of the kids who take this path to prosperity:

        Grow up in rural midwest or south => Go to college at state university in small city where they can still commute home every weekend to take care of family issues => get job in Chicago, Atlanta or another big city after college and send money back home

        The second step in that path requires us to continue to support our secondary centers.

        • Eric2

          Every state has a number of universities (and even more community colleges) which locals can go to. I just checked, even in North Dakota the whole state is within 200km of a 4 year university. Yes it would be nice to subsidize education more (even in-state public universities have become expensive in recent years). But I don’t think getting to campus is the main obstacle here.

      • adirondacker12800

        There are 11 million people in metro Paris at a density of 3,800 per square km and 8 million within the borders of New York City at a density of 10,600. I’m not in the mood to add in Nassau, Westchester, Bergen, Hudson and Essex counties and recalculate the density.

        • Eric2

          Paris has a density of 21k/km2. Almost exactly twice NYC. Therefore I said 8 million new people in NYC.

          Of course many more additional people could fit in NYC’s suburbs, though it would be significantly harder to expand the transit system to accommodate them (there are 18 outbound subway tracks from Manhattan to the outer boroughs, only 8 outbound rail tracks which must also accommodate intercity rail and a larger existing population base).

          • adirondacker12800

            If everybody is manipulating symbols for a living along the coasts they starve to death once the food runs out.
            Packing 20 million people into New York City runs into other problems like where to get water. How to move the garbage out. Where to put the sewer plants. And import enough electricity. Manipulating symbols does use electricity. Once the windmills in North Dakota start to fail it’s a reeeeeeeeeeeeeeeeallly long drive from Morristown New Jersey to get to them. they could probably figure out a way to pack enough water, food and fuel onto the truck to do it. Lash the camping gear on top of it all.

          • Eric2

            All those are cheaper for the public, and better for the environment, than new infrastructure for the same number of people elsewhere.

          • Phake Nick

            There are tons of cities around the world having a population of 20 Million, it’s not impossible.
            As for farmers, the question is where the benefit of farming to the society and benefit of jobs in cities to the society balance out. With city jobs still offering higher pay and more opportunities and people still fluxing from rural to city it appears that the balance is still far from reached.

          • adirondacker12800

            There are tons of cities around the world having a population of 20 Million
            There are three of them.

            I would want to know if the city limits sprawls out into farmland like the Phoenix Arizona city limits do. Sumptin’ is going on with Chongqing because the municipality has 30 million people but the urban area is just 7.7 million. And sprawls over the same area as South Carolina give or take a few hundred square kilometers. Decide on what you want to measure and apply it consistently.

          • adirondacker12800

            …..All those are cheaper for the public.
            You want cheaper do it in Chicago. Effectively an unlimited water supply right in the city and it’s closer to the windmills in the Dakotas. It’s closer to the food too.

          • adirondacker12800

            …..All those are cheaper for the public.
            Another one!. Twice as many people on Long Island, including Brooklyn and Queens means twice as many trucks. The entrance and exit ramps from the double decked Long Island Expressway will be long. Not particularly cheap either. And a twin to the Verazanno Narrows Bridge. Double decking the Staten Island Expressway isn’t going to be cheap either. Or the Brooklyn Queens Expressway so those trucks can get to the LIE. And they would need a way to get to the Staten Island Expressway. Or the two new tubes of Lincoln Tunnel. Makes me wonder how far south on the New Jersey Turnpike where there is enough space to stop double decking. Just think of all the extra tolls the people in Delaware could grift!
            Does mean that it changes from twice as many people on Long Island, roughly, as there are in Connecticut to there being as many people on Long Island as there are in New England. The LIRR Mainline goes from being four-ish tracks wide to ten ( two to serve high speed rail for the 14 million people on Long Island. They could sink it in a trench four tracks wide and three levels deep! 12 tracks! On the third hand I’m not sure how far east the six track wide ROW goes in Nassau County Two levels deep and six tracks wide. You want me to keep going or perhaps it’s a better idea to figure out a way to encourage people to move to Cleveland or Detroit where, like Chicago, there is essentially an unlimited supply of water.

          • Eric2

            “You want cheaper do it in Chicago.”

            There is huge excess demand for NYC jobs. There is not excess demand for Chicago jobs. Just look at the housing prices in each.

          • adirondacker12800

            Double the population of New York City you need to reproduce the subway system. They aren’t all going to calling for an Uber to get to work. 3 or 4 cent a kilowatt hour surcharge on the ConEd bill so HydroQuebec can convert people using electric baseboard heat to ground sourced heat pumps freeing up HydroQuebec capacity?
            It all gets very very expensive. An hour to Springfield Mass, well 1:15 because bulldozing 40 miles of very expensive suburb to make it an hour wouldn’t be worth it, makes Springfield or Utica a much more attractive place to have your start-up. Or the developers toiling away at leeching more data from your users. Or Lancaster Penna.

          • Eric2

            The subway has a huge excess of potential capacity, particularly between Brooklyn-Manhattan, if they can just deinterline and improve the signalling and dispatching a bit.

            Residents of NYC use a lot less electricity per head than residents of Nashville or Dallas or whatever other metro area they would otherwise be living in.

            I have no idea what point the second half of your comment is trying to make.

          • adirondacker12800

            Yes New Yorkers use less electricity. Twice as many of them will use twice as much. Shipping electricity from the wind farms around Amarillo to Dallas is a lot cheaper than sending it to New York. Dallas would likely have that pesky problem with water if you double it’s population too. Perhaps someplace else that has lots of water and other infrastructure not being used? Like Detroit. It’s quite windy around the Great Lakes just like it is in Texas The Yoopers would have something to do that pays well. Maintaining wind farms.
            One of the main reasons for Water Tunnel 3 is so that Water Tunnel 1 can be taken out service for inspection after more than a century of use. You want to double the population you need Water Tunnel 4, 5 and 6. To Lake Ontario because draining the Delaware dry isn’t an option.
            Deinterlining moves more trains. The subway, being a passenger railroad is about moving people. There isn’t enough excess capacity for twice as many of them.
            The subway does have excess capacity laying around The four tracks from the Williamsburg bridge to Chambers street is likely the most glaring example. Using them two of them for the Second Avenue subway and feeding that through the existing Montague Street tunnel to the grossly under used Culver and Fulton lines would be nice. Those lines would then have service to the East Side in addition to the West Side. Feed Second Avenue trains over the Culver Line to the West End Line it gets West End trains, today’s D train, out of Dekalb. Today’s F train could run express most of the day! Hmmm I wonder if the six tracks of Hoyt-Schermerhorn could be reconfigured to run more locals on the Fulton line – from the Culver line. Hmm. All of the A trains could go to JFK and Lefferts would get more service by sending some of the C trains there!
            Sending the E through the 63rd Street tunnel to use the excess capacity on the Second Avenue south 63rd gives everything else indigestion. It eventually works out with 60th Street tunnel trains not going to Astoria running along Northern Blvd to Flushing. Does leave some Broadway BMT trains at 57th Street with no place to go. Hmm. You can’t send them all to 125th and Lex because then there is no capacity for Second Ave trains to downtown if you do that. Complicated isn’t it?

    • Nilo

      The region often thought to have had the most severe decline, the Rust Belt gets excellent service in this map. Absolutely brutalized Youngstown has a stop! Pittsburgh has a stop!

  15. Michael M

    HSR in the Quebec-Windsor ‘corridor’, as it’s called, is generally conceptualised as bypassing Ottawa, going straight from Toronto to Montreal via Kingston, with an optional spur to Ottawa – the highways are arranged this way as well, going across the St. Lawrence with Ottawa off to the side. How would that alignment impact your model?

    • Alon Levy

      It would make service worse by requiring more construction and reducing frequency on each segment (with the spur), or by omitting a decent source of ridership (without the spur).

  16. RossB

    Looks good to me. The Northwest line should have speed improvements, but not be high speed. Running trains an average of 180 KMH (110 MPH) would be much cheaper, and achieve similar ridership. Which is to say, not especially high. There already is a plan in place, it just needs to be funded, and could be funded for a lot less money (https://www.aawa.us/site/assets/files/7322/2006_washington_state_long-range_plan_for_amtrak_cascades.pdf). Another improvement would be to speed up the border crossing, which takes way too long.

    The report on what they call Ultra High Speed Rail (400 KM or 250 MPH) shows the folly of pursuing this approach: https://wsdot.wa.gov/sites/default/files/2019/07/12/Ultra-High-Speed-Ground-Transportation-Study-Business-Case-Analysis-Full-Report-with-Appendices-2019.pdf. If you look on page 15, most riders would still drive. On page 47 it shows that price is a huge factor. This suggests that the best approach is to make significant improvements in speed, while keeping the costs low. Given the challenging terrain, it makes sense to work with the tracks they have, rather then build a completely new line.

    • AJ

      If price is a huge factor, then that simply suggests a large public subsidy for O&M rather than capital costs. Even at current speeds, if there were more frequent trains and a trip from Seattle to Porland was ten bucks, ridership would be much, much better.

      I support the existing WSDOT long range plan Ross linked, which is currently estimated at $6.5B (2006 dolllars) of capital investments or a tenth the cost of true HSR on the same corridor. Any further public spending should be on better frequency and lower fares to support a better modal split, rather than more speed, particularly as WSDOT is assuming fares high enough to cover O&M. Arguably WSDOT should focus on boosting frequency first and then invest the speed, as the existing alignment is already competitive with driving (and there remains plenty of low hanging fruit for a fraction of the $6.5B)

  17. Allen B

    The New Orleans discussion is correct re: population stagnation, but I find the population-only analysis misses two factors.

    First, the city bats above its weight for travel demand because of its appeal as a tourist and convention destination. Second, it has a compact urban core that makes it attractive to travel to by mass transit rather than by car.

    This is why the New Orleans airport has more annual passengers than many other cities included in this plan with larger populations, including San Antonio, Jacksonville, Cleveland, Cincinnati, etc. – though it is a hub for no airlines, so it has only local passengers no transfers.

    I’d imagine for these reasons that extending a spur to it from the TX system would attract quite a high ridership; I’m less sure of it, but a spur extending east from it to the gulf beaches could also work, though perhaps only seasonally.

    • Alon Levy

      Tourism at best counteracts the malus coming from not having mass transit. A typical inhabitant of the region doesn’t live in the city but in far suburbs, and travels not to Downtown Houston but to one of its suburbs too, which reduces the advantage of the train over the car. In the NOLA-bound direction you’re right that there’s more centralization, but that’s only for tourism trips and not business trips.

      This is especially true when Houston-NOLA can’t be just about New Orleans – New Orleans is far too small. Rather, it also requires Baton Rouge to fill trains and get to around 1.8% ROI – and Baton Rouge has the full malus for extreme sprawl without the tourism bonus.

      What’s true is that the sprawl malus exists in many other places. It’s really bad in Florida – but in Florida the cities are larger and closer to each other, and the extent of tourism in Orlando is so enormous I’m comfortable just going with the model rather than docking ROI. The sprawl is also really bad in Ohio, but Ohio like Florida has decent population density, and on top of it has value coming from travel to New York and Chicago, both of which have strong city centers close to the train station.

      The Austin/San Antonio connection is a good sanity check on this. It’s 290 km of construction, vs. 530 for Houston-NOLA. Austin and San Antonio are around twice the size of NOLA and Baton Rouge. Austin and San Antonio are close to Houston and Dallas, whereas NOLA is 930 km from Dallas and Baton Rouge 820. So the Austin/San Antonio connection isn’t marginal – the model gives it 4.5% ROI if construction costs can be kept to $25m/km, and even at the horrific Texas Central costs it’s 2.2%.

      • AJ

        But to Allen’s point, the vast majority of ‘business’ trips are heading to either the convention center or the central business district. A suburban Houston to suburban New Orleans business trips you reference aren’t driving the unusually high New Orleans airport numbers (aside from passengers during the Houston hub to transfer); at 5 hours it’s likely time competitive to drive vs fly. I think it’s plausible NOLA is an exception for the same reasons as Orlando, just not to the same extreme.

        As Dallas and Houston slowly create the pale skeletons of real urban transit, rail should be a compelling option for trips between Texas and New Orleans?

      • adirondacker12800

        There are automobiles to be rented that one drives oneself and ones, for very short term rental, that have drivers. It works that way at airports all over. It even works at railroad stations that have a lot of passengers. The short term rental with driver at quite a few. Just like at airports they may even be a designated area for them to loiter at, awaiting the next customer to walk up. without reservations etc. Very busy hotels may even have their own bus that circulates between the hotel and the airport! Not everybody is a frugal masochist scheming to squeeze every nickel till the buffalo um …..cries.

  18. Pingback: New York Regional Rail (not S-Bahn) | Pedestrian Observations
  19. Pingback: News roundup: – Seattle Transit Blog
    • Phake Nick

      It’s essentially improved conventional railway with dedicated right of way. It’s relatively cheap and quick to implement and it’s also an improvement over what was being provided, hence I think it’s a good step, and with increased speed, frequency, reliability, and additional station along new track, it will attract more riders, and such increased level of rider could provide further justification for them to continually improving the line’s service level.
      It wouldn’t replace HSR but it can be a good step toward having HSR in the future, I guess

    • adirondacker12800

      New York, Ontario and Quebec, New York and Quebec, Ontario and New York, come up with some bizarre plan every five years or so and it gets archived on the dusty bookshelves with all the other plans. New York’s last one was to spend eye watering amounts of money to build an electrified 125 mph railroad in nice flat places west of Utica where the tracks would be very very straight instead of a 200, 220 mph one.

  20. Martin Kolk

    It seems to me that a San Antonia- Austin-Dallas line would outperform basically anything else on this map. What is the motivation for not including that? It is also the only part of the map where a normal double track would perhaps not be enough (at least if one expects some semi-regional traffic on the tracks which seems reasonable), as the current plan would maybe put too much stress on the link going to Dallas (the Dallas metro area already is close to 10 million people and will grow a lot, and here all traffic to Houston/Austin/San Antonio will be on a single double track). The Texas triangle seems like much more fertile ground for passenger-rail than most of the lines on the map.

  21. Pingback: High Speed Rail and Conventional Intercity Rail are Not Mutually Exclusive | Sewer Socialists
  22. Ernest Tufft

    I think HSR from Seattle-Vancouver to San Diego is possible because from Redding to Medford, there is longer term population growth, tourism, and ridership friendly to electric “green” trains. The cities line up in direct line within Willamette and Sacramento valleys. So, problem is engineering tunnels through the Siskiyou Pass.

  23. Pingback: Expanding American Rail Capacity – Home Signal
    • Tonami Playman

      Won’t that be 20tph for the new schedule compared to 18tph? The old schedule was 10tph for Nozomi, 2tph for Hikari, 3tph for Kodama and 3tph for Oi depot runs. The new schedule adds 2 more Nozomis bringing the total to 20.

      @Phake Nick thanks for that link, because this is the first time I’ve come across info about the 3 depot runs incorporated into the Tokaido Shinkansen schedule. Makes sense as there are only 2 Shinkansen tracks between Tokyo station and the branch point to the depot. This is a similar situation with the Tohoku Shinkansen compared to Chinese HSR stations that have dedicated depot tracks connecting the stations to their depots.

  24. Pingback: America needs high-speed rail - Student News Today
  25. Pingback: Amtrak Sucks; Publics are Terrible at Choice. – Indefensible Expectations
  26. Pingback: A Realistic US High-Speed Rail Map – Koji's Posts
  27. Pingback: High-Speed Rail Doesn’t Depend on Megaregions | Pedestrian Observations
  28. Pingback: Midwestern Urban Geography and High-Speed Rail | Pedestrian Observations

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.