High-Speed Rail

Envisioning a Future Interstate Rail Network

This is the first piece on a national rail network. The second, revised version is here.

A public-works precedent

On June 29th, 1956, the creation of a 41,000-mile U.S. Interstate Highway System was authorized by the Federal-Aid Highway Act. Today, the system of grade-separated, high-speed highway corridors extends across more than 47,000 miles of the national landscape. For a large percentage of Americans, the Interstates have come to define daily life. Largely because of the Interstates, travel within metropolitan areas and between cities is effected primarily by automobile.

In an age of increased environmental consciousness and in an era in which more and more people desire an alternative to the car for commutes long and short, it is time for the Interstate’s heir. That successor must be a system of interconnected high-speed and standard-speed railways. the transport politic’s vision for such a network is presented here.

The need for a national system

Today’s Amtrak system, pictured below in thin black lines, is anemic at best. It is slow and inefficient, and it doesn’t even serve a significant number of the nation’s largest metropolitan areas. All of the nation’s metro areas with populations of above 100,000 people are represented in proportion to their population sizes in the maps below. Amtrak is poorly connected to the Canadian VIA Rail system, and it is hindered by reliance on freight-owned tracks. It must be improved, or the American passenger rail network will never advance to international standards.

The National Association of Railroad Passengers recently published its Grow Trains report. NARP suggests expanding the existing Amtrak system to serve many of the cities and towns that trains simply skirt by currently, arguing that the national rail provider could relatively easily expand along existing rail right-of-way. But NARP’s vision doesn’t go far enough.

Similarly, the Federal Railroad Administration has authorized the development of high-speed rail along a number of the nation’s corridors, illustrated in bold yellow in the map below. The States for Passenger Rail Coalition has recently been pushing for the development of medium-to-high-speed rail along FRA’s proposed lines.  But the FRA’s vision is now decades old and the corridors it has authorized focus either on unpopulated areas – such as the route from Dallas to Texarkana – or miss obvious connections – such as between Pittsburgh and Cleveland. Neither the FRA nor the States for Passenger Rail Coalition have been active in working with Canadian authorities to consider how the two nations’ systems might function as one.

And yet, abroad, governments are planning massive new systems of railways. In Spain, 10,000 km of high-speed rail are under construction; in China, 13,000 km of high-speed and 40,000 km of standard-speed track are being built. The United States, to put matters bluntly, is quickly falling behind. It doesn’t help matters that we simply have no national framework for developing our rail network, and must rely instead on a hodge-podge of local actors to get anything done.

It is time for a new vision, national in scope and ambitious in form.

Existing Intercity Rail Network

Existing National Rail Route Network – Amtrak and FRA-Proposed HSR

A new vision

California is leading by example. After the passage of a $10 billion bond last fall, the state has begun the serious planning of a new network of rails that will extend from San Francisco to Los Angeles and eventually from Los Angeles to San Diego and from Fresno to San Diego. Trains will run at 220 mph, equal to the fastest trains in Europe or Asia today. The system will require a massive investment, but it will bring down travel times between the state’s two biggest metro areas to 2 hours 40, dramatically changing the role of rail there. In a single swoop, California will have made traveling by rail fast and convenient. If the foreign precedent applies, the success of California’s plan is virtually assured.

Below is the transport politic’s vision of the 21st century’s Interstate system: a network of 10,000 miles of high-speed rail and roughly 30,000 miles of upgraded standard-speed track. The system would provide electrified 200 mph service (in yellow) between the biggest cities on the East and West coasts and connect every metropolitan area of more than 100,000 people in the continental states with at least standard-speed rail (in brown). Standard-speed rail could be implemented relatively simply along existing freight right-of-way; in many cases, these tracks only need minor touch-ups to be readied to serve passengers. The system would rely on existing Interstate and rail right-of-way and extends on both the NARP and FRA proposals, but narrows in on the most cost-effective and interconnected corridors, focusing on the most densely populated regions. This is why each map of the rail system included here has as its backdrop the concentrations of population in metropolitan areas in red.

The system would have an emphasis on connecting destinations separated by 500 miles or less; for such distances, high-speed rail outpaces airplanes and in other countries has commanded up to 80% of the market share on such routes. The high-speed system would not traverse the Great Plains or the Rocky Mountains, as such a trip would likely attract few passengers and be relatively cost inefficient. It would not provide high-speed service for Denver or Salt Lake City, but both are so isolated that high-speed rail to and from them would be relatively underused. But the whole system, including standard-speed rail, would allow for a high degree of interconnectivity between the cities in the densest areas of the country and allow for the time efficient replacement of the automobile and airplane on a large percentage of trips.

Such a system would require an active federal government funding an expensive national system, maintaining its infrastructure, and running its trains. Our government is currently not capable of doing as much, but with a defined vision such as this – to provide rail service to all of the nation’s metro areas and to connect the biggest ones with true high-speed rail – Washington could mature to the task. Bck in 1956, the federal commitment to highways was minimal; in one bill, under one president, the system changed.

Proposed North American Intercity Rail Network

Proposed National Rail Route Network – Standard and HSR Routes

The system as developed here would allow for trains to travel throughout the nation, with connections with a Canadian high-speed network at Detroit, Buffalo, and in northern Washington and New York States. An ideal system would mean electrification of the entire network, both standard and high-speed; electric trains are faster, quieter, and better for the environment than their diesel counterparts. Trains on a national network might be equipped to travel on both high-speed and standard-speed lines, thereby giving even that part of the population not directly on the high-speed lines access to the high-speed network.

Below are regional maps of the system, showing stations in every metropolitan area of above 100,000 people.

Western High Speed Rail Network


Midwest High Speed Rail Network


Northeast High Speed Rail Network


Texas High Speed Rail Network


Southeast High Speed Rail Network


The proposed high-speed network

The spine of this new interstate railway system would be the high-speed rail network, totaling 10,352 miles in length and directly serving metropolitan areas representing 60% of the population of the United States and 46% of the population of Canada. It would include large upgrades to the two currently electrified portions of the system – the Keystone Corridor and the Northeast Corridor – and the creation of brand-new railways along the rest of the system.

A high-speed system of more than 10,000 miles in length would be the longest in the world. But the United States remains the world’s wealthiest nation, and its need for improved mobility is great. An investment of this sort would not be out of line.

Construction of the system would be implemented in four phases, each with 2,000 to 3,000 miles of service, and ordered according to relative merit within the network as a whole.

In order to evaluate the different lines, the transport politic developed a system by which it could examine the cost effectiveness of each line both in terms of travel within the corridor alone (the Corridor Score) and within the system as a whole (the Overall Score). Travel between every city pair in the system between 50 and 500 miles apart was evaluated, and the results were compiled by corridor, whereupon they were divided by route mile to appraise potential ridership by mile of new construction. The results provide the basis for prioritizing routes and suggest a method by which the federal government could begin to imagine how such a high-speed rail system might be developed. (PDF with description of methodology, evaluation of every city pair, and scores for each corridor or here.)

High Speed Rail Network Phasing

HSR Route Network – Phasing

High Speed Rail Network Routes

HSR Route Network – Corridors

Above left is a diagram of the high-speed rail network’s phasing plan. The first phase would provide service in the Northeast, in the Midwest, in Florida, and in California; subsequent phases would fill out the system in the Northwest, the Southwest, and the Southeast. Above right is a diagram that demonstrates the locations of the 15 high-speed corridors as defined by this proposal.

Below is a chart that provides data on the 33 different sub-corridors evaluated for the system, providing the mileage, Corridor Score, and Overall Score for each of the lines. The data demonstrate why an initial focus on routes in the Northeast and the Midwest makes the most sense: the corridors there are likely to attract the highest ridership, because of the dense concentration of large metropolitan areas in those regions.

California, Florida, and Texas all, too, deserve strong high-speed connections within their respective states. The Florida and Texas plans are similar to those proposed in the mid-1990s. The California CrossState proposal is the same as that proposed by the California High-Speed Rail Authority; the first phase from San Francisco to L.A. has been prioritized ahead of its higher-scoring (according to the transport politic’s criteria) L.A.-San Diego and Fresno-Sacramento extensions because of the political realities in California and because of the text of last fall’s Prop 1A, which required that the trunk line between the two biggest metro areas in the state had to be built first.

The research done for this proposal does suggest that certain lines, such as that connecting Macon, Georgia, and Jacksonville, Florida and that connecting Los Angeles and Las Vegas, are relatively expendable. Lines in Phase 4 are simply not as cost-effective in terms of likely passengers per route mile as corridors in the other phases. The ordering proposed here, however, is simply a suggestion based on a series of a few objective criteria. Other issues were not considered, such as a desire for an East Coast north-south route as soon as possible or a connection between New Orleans and Houston. So there’s no reason why this proposal would be the be-all end-all.

That said, the vision presented here of a unified national system with a phased-in order based on expected ridership, makes sense. A well thought-out high-speed rail system could be an effective tool in fighting climate change and in providing an alternative to automobile and airplane travel. Any such plan must be implemented with consideration for the system as a whole and with a greater vision for serving the populated areas of the entire country.

First, however, we need a national government that’s willing and able to take charge.

Phase 1 Corridors – 2,757 miles
Route Order Mileage Corridor Score Overall Score
NEC / Boston-NYC-DC 1 454 2,486 3,436
NY CrossState 1 / NYC-Albany 2 141 781 3,592
NorthEastern / Boston-Albany-Hamilton 3 499 829 2,091
MidWestern 1 / Chicago-Cincinnati-Louisville 4 350 557 1,456
MidWestern 2 / Cincinnati-Columbus-Akron 5 258 553 1,800
Lakes 1 / Detroit-Toledo 6 57 454 3,133
Florida CrossState / Jacksonville-Orlando-Miami 7 498 878 907
California CrossState 1 / San Francisco-LA 8 500 516 1,091
Phase 2 Corridors – 2,272 miles
Route Order Mileage Corridor Score Overall Score
California CrossState 2 / LA-San Diego 9 167 1,408 1,877
Lakes 2 / Chicago-Toledo-Cleveland 10 371 477 1,879
California CrossState 3 / Fresno-Sacramento 11 120 636 1,342
CrossCanada 1 / Hamilton-Detroit 12 197 407 2,074
Crescent 1 / DC-Raleigh-Charlotte 13 409 556 1,092
Crescent 2 / Charlotte-Atlanta 14 252 489 1,039
Lakes 3 / Cleveland-Pittsburgh-Harrisburg 15 357 396 1,277
NorthWestern / Portland-Seattle-Vancouver 16 301 564 564
Keystone / Philadelphia-Harrisburg 17 98 198 1,474
Phase 3 Corridors – 2,405 miles
Route Order Mileage Corridor Score Overall Score
Texas CrossState / Dallas-Austin-Houston 18 412 485 581
CrossCanada 2 / Hamilton-Toronto-Montréal 19 434 318 793
MidWestern 3 / Chicago-Milwaukee-Minneapolis 20 438 429 655
Crescent 3 / Atlanta-Birmingham 21 142 268 710
SouthEastern 1 / Chattanooga-Atlanta-Macon 22 193 233 761
NY CrossState 2 / Albany-Montréal 23 205 128 1,321
MidAmerican 1 / Chicago-Springfield-St. Louis 24 290 271 579
SouthEastern 2 / Louisville-Nashville-Chattanooga 25 291 217 705
Phase 4 Corridors – 2,918 miles
Route Order Mileage Corridor Score Overall Score
SouthWestern 1 / LA-Phoenix-Tucson 26 434 186 366
MidAmerican 2 / Dallas-Oklahoma City 27 217 169 351
MidAmerican 3 / Oklahoma City-Kansas City-St. Louis 28 615 189 226
CrossCanada 3 / Montréal-Québec 29 168 161 338
Crescent 4 / Birmingham-Jackson-Dallas 30 639 158 260
SouthWestern 2 / LA-Las Vegas 31 227 113 294
Crescent 5 / Jackson-Baton Rouge-New Orleans 32 296 108 170
SouthEastern 3 / Macon-Savannah-Jacksonville 33 322 43 220

(PDF with description of methodology, evaluation of every city pair, and scores for each corridor or here.)

42 replies on “Envisioning a Future Interstate Rail Network”

Amazing work (and amazing PDF), I’m definitely very impressed by all the calculations to achieve those rankings. If only policy makers actually listened to transit advocates and looked at such proposals.

I would maybe say that there should be some type of bonus added to corridors in which a small city has a bigger stature than its population size indicates. For instance, I’d expect the Albany-New York line to have more ridership than its population would suggest since all the state’s power brokers transit between those two cities, and there are obvious reasons for New York City-ers to go to Albany.

Yonah- wonderful presentation. FYI- I tried to download the PDF for methodology but got file errors using Firefox 3.0.

On the substance, here is another question. You list the Raleigh-Charlotte corridor as a candidate for 150-200 mph service, mostly tracking the NCRR railroad.

The NCRR is an old line with lots and lots of twists and turns. Also, in terms of station spacing, the benefits of the high cruising speeds may be considerably reduced by putting stations in beyond Raleigh, Durham, Greensboro and Charlotte. (Standard-speed trains could connect other cities to these hubs)

So here’s my question to you. I like high-speed rail much better than medium-speed rail, especially having ridden the Thalys from Paris to Brussels in 2007. What’s even better about the system of HSR in Europe is that you get off and have access to robust local transit networks. In the US, most of our local transit networks are relatively weak.

A HSR program without a corresponding boost in transit development may generate a considerable amount of travel based on park-and-riding and car renting at destinations.

Would we be better off pursuing a strategy of mostly 125-140 mph diesel service using light, powerful rolling stock and using the money we would have spent straightening all tracks to TGV-esque standards on beefing up local transit along the HSR lines?

With the FRA moving towards positive train control by 2015 and the safety management that would be needed for HSR, it is possible that a train like the British InterCity125 might be the right fit.

What do you think?

Patrick –

I’m not sure what’s up with the PDF in Firefox; it works in Safari, but the new “here” link after the PDF link should work.

In terms of the North Carolina corridor – I know it well, obviously, being from Durham.

I compared the Raleigh-Durham-Greensboro-Charlotte corridor to a new line from Raleigh directly to Charlotte, which would be better in terms of being straighter and faster… but the results are dramatically less promising in terms of ridership. Simply put, the intermediate stations at Durham, Burlington, and Greensboro make this corridor worthwhile for HSR; a Raleigh-Charlotte direct connection wouldn’t be good enough.

In terms of whether to invest in full-out HSR or not:
> This is a compromise one must always make. If we choose to invest in HSR, we are choosing not to invest in something else, maybe improved local transit.
> On the other hand, true HSR at 200 mph is likely to be more cost-effective in the long term in terms of attracting riders (and being profitable) than a slower service, so this is an open question. Faster service also makes longer distance journeys more reasonable for passengers.
> Third, diesel rolling stock is fine… but not particularly environmentally friendly. If this is supposed to be about carbon reduction as well, electric trains are where this is at. That said, diesel trains that could also operate on electric power would make a lot of sense. In part of the corridor, they would run on catenary, while in less developed areas they would run off diesel engines.

It seems to make more sense to connect Binghamton/Ithaca, NY, to NYC instead of Philly. The traffic/movement/connections are a lot stronger that way.

One suggestion would be to connect DC to the Great Lakes through PIT instead of having travelers go to PHI. The corridor of CHI-DET-CLE-COL-PIT has a large population and a direct link to the Capital would be good.

Additionally, anyone from NC/VA needs to go to either PHI or ATL to travel west. NC has shown that it is growing, therefore, without blasting another trail through the Appalachians, a DC-PIT line would give them quicker access to this population corridor.

All that being said, I don’t know if there is too much twisting and turning in the mountains between DC and PIT. Maybe it would follow I-68 across MD/WV and then cut up I-79?

Already the comments section betrays the political difficulties of HSR; no one can agree on where to run the train and no one can agree where to stop the train. The politics of funding the train then become substantially problematic. I hope Yonah addresses these roadblocks and shows how the Federalist structure of government can deliver this service.

Diesel is not the way to go as was suggested above. Electrifying the entire system has multiple other benefits including a stronger power infrastructure in this country.

I like the plan but see some flaws in it. First was mentioned above. What happens when you get off? You need to have transit. Its not a one or the other. Its a both. Also. You show both Poughkeepsie and Kingston on the NYC-ALB line. Not possible. The ALB station is already on the east side of the river and between Poughkeepsie and Kingston is not a great place to swing from east side to west side and make the rest a new run.

I like the idea. I hope you have a follow up to it that looks at once this is built this is our next plan.

Check out this HSR network I have been tinkering around with for the past few years:

Your idea is a solid effort but I think some routings are too zigzaggy to get full user buy in. Take Texas for example. The route makes it likely that users will see that between Houston and Dallas that it goes too far out of the way by heading to Austin first for it to be worthwhile vs driving. A version of the Texas triangle would be better (with the points of the triangle being Waco, College Station, and Austin [huge college ridership guaranteed here already]) with lines continuing out to the major cities of DFW, Houston, and San Antonio from there.

Also, in your close up of Canada, you depict a secondary line from Oshawa to Ottawa that goes south of Kingston. I believe that would be very difficult to pull off as both Oshawa and Kingston already are located on Lake Ontario.

Thanks Chris…
You’re right about Poughkeepsie and Kingston being located on opposite sides of the rivers, and this would be a problem. However, it could theoretically be solved simply by having a new station on the opposite side of the river from one of the towns… which would, as you say, require good transit connections.

Nick – you’re absolutely right about the line problem in Ontario… my limited knowledge of Canadian geography got me confused. Maps fixed. (Route shown as HSR originally is now standard-speed to Peterborough.)

Thanks for the hard work putting this together.

I’m wondering if you have data available on routes that were not chosen. I would think that Fort Collins–Pueblo, CO (possibly south to Albuquerque or El Paso—don’t forget connections to Mexico) would have a higher overall potential for ridership than the section from Oklahoma City to Topeka as the population is concentrated linearly along the Front Range. I realize that this line is connecting Texas cities to Chicago, but perhaps a more direct direct route from OKC to Saint Louis or a route from Houston through Memphis would attract higher ridership.

The West presents difficulties due to its dispersed population, but it includes some of the fastest growing cities of the country and future growth should not be discounted. El Paso, Denver and San Antonio (another city that would provide travel options south to the Rio Grande Valley and Mexico) are the three largest metropolitan areas in the country not included in your system and scores should at least be calculated to show why they do not necessitate HSR service. Phoenix to LA is included and that line would travel nearly 300 unpopulated miles between the two. (I’m not arguing that it shouldn’t be included, but it provides another example with few population centers between destinations).

I would also look at a direct connection between SF/Oakland and Sacramento, although this may be difficult due to terrain and density.

And echoing pagespages, I agree that a line from DC to Chicago could also be beneficial. The Harrisburg to Baltimore/DC section is not marked with standard rail and should be included as it is also a heavily traveled route. I would connect State College through Harrisburg as well; the routing you have to Hagerstown would be nearly impossible due to the ridge and valley terrain that the line would have to cross plus would have little ridership. Going through Harrisburg would connect the student population to Philadelphia and the East Coast.


Great article and vision, are you interested in doing advocacy for this network with other as part of a national campaign?


In CA the Oakland – Sacramento connection has been found to be most efficient through the Niles-Altamont alignment, Oakland-Pleasanton-Livermore-Stockton-Sacramento. Unfortunately that alignment was not selected for the initial system and has been placed in a separate project.

Alan, the Binghampton line runs from Philly to Scranton to Binghampton to Albany. Philly to Scranton makes a lot of sense, and once you’re in Scranton it’s silly not to connect to Binghampton.

So you’re question is really, why do you connect Binghampton to Albany instead of running a direct route to NYC. There are two routes to get from Binghampton to NYC. The first would be via Port Jervis, and the second would be via Scranton and the Lackawanna cutoff. I’m not sure which route would be quicker/cheaper to build, but what you have to remember is that this would be an entire standard speed line vs. a high speed line from NYC to Albany and then a standard speed line from Albany to Binghampton. Depending on the exact distances and speeds, it may actually be faster to get to Binghampton from NYC via Albany. And even if it’s slower the marginal difference might not make up for the possible traffic gained by having the Albany/Bingampton link.

This is great! I think anyone living in the US who is familiar with what the Europeans and Japanese have accomplished has envisioned a fantasy map like this.

I thought it was interesting that a recent post on this site mentioned the “Texas T-bone” but this route map doesn’t show that. Instead, you are showing a very awkward alignment going right through Austin. The “T-bone” alignment connects the major cities in the Texas triangle (Houston, Dallas, San Antonio) with three lines that all converge on Killeen, a town in the middle of all three.

Please don’t forget the awesome extension to Monterrey from San Antonio. As the industrial capital of Mexico, it woudl be incredibly effective to provide rapid transit to and from the first major city in the US. This also goes for San Diego/Tijuana, which is probably even more critical

Nice site. However your analysis of the routes and their relative importance has some very serious data flaws in it. Particularly, the SF-LA route you ranked less than Toledo-Detroit or Fresno-Sacramento. Are you nuts? Fresno and Sacramento are podunk rural hamlets compared to the Bay Area and LA, and they’re far apart at that. I think there are several zeros missing your population assumptions for the Bay Area. The San Francisco area has 7 million people in it. You must be assuming that “San Francisco” means just the city of San Francisco with its 800,000 people. You also have to understand that a station in SF links to the entire 7 million inhabitants of the Bay Area by local transit. I think if you used real figures for the Bay Area you’d see that connecting these two metropolitan areas would rank much, much higher. I can tell you there ain’t much demand for that Fresno-Sac route, while several airlines run hourly or half-hourly shuttles between SF and LA shuttling thousands of people daily and that the interstates connecting the two areas, while 500 miles apart, is constantly congested.

Thanks for the comments, Josh. However, the ranking is not based on the number of passengers per route but rather number of passengers per route mile, which gives a better understanding of the relative cost effectiveness of each route.

There is no doubt that more people will ride from San Francisco to Los Angeles than from Fresno to Sacramento. However, the route is so much longer between LA and SF that it would be less cost-effective to build per mile, according to my calculations. Take them as you may.

If you go over the data in the PDF linked to above, you’ll see that the data I used does not misrepresent the population of the San Francisco metro area.

Yonah –
well, ok, but that’s still not right. First of all, Fresno-Sacramento is not a actually phase at all. In fact, Fresno in in the middle of the SF-LA phase and is part that phase — there’s no way that it won’t be! Look at your own map. The Sacramento extension only get you the short distance from Merced to Sacramento, including Stockton, which has limited value. The SF-LA route INCLUDES stops in Merced, Fresno, and Bakersfield, capturing the vast majority of the the Central Valley except Stockton and Sacramento. You have to include the population and ridership of Merced, Fresno and Bakersfield on the SF-LA line, or else the anlaysis is just plain wrong.

Josh – You’re right – the entire distance between Fresno and Sacramento would not be in the “CA CrossState 3” labeled above; however, the calculations linked above consider only the new 120-mile section from between Fresno and Merced to Sacramento.

If you look at the California HSR site (, you’ll notice that Merced is on the CACS3 phase, not on the CACS1 phase… I’m sorry if the maps here do not reflect that as well as they should; I’ll update them to make that clear.

If you look at the calculations, you’ll find that the following journeys are included in the calculations for CACS1:
– SF-Fresno; SF-Hanford; SF-Bakersfield; SF-LA; SJ-Fresno; SJ-Hanford; SJ-Bakersfield; SJ-LA; Fresno-Bakersfield; Fresno-LA; Hanford-Bakersfield; Hanford-LA; Bakersfield-LA
The following journeys are included in the calculations for CACS3:
– Fresno-Merced; Fresno-Modesto; Fresno-Stockton; Fresno-Sacramento; Merced-Stockton; Merced-Sacramento; Modesto-Sacramento.

Journeys shared between the two corridors and with CACS2 are proportioned according to their relative share of the journey distance.

Seriously, look at how these results were formulated before you start questioning methodology.

A lot of the corridors you propose are really weird and convoluted – for example, Houston to Dallas via Austin, or Miami to Jacksonville via Tampa. The less straight the routes are, the fewer people will use them.

Your choice of intermediate stops is often questionable as well. For example, for the Empire Corridor, you propose a stop at Kingston and another in Poughkeepsie, which makes it two stops in Dutchess and Ulster Counties (combined population 480,000) but none in Westchester (population 950,000, plus about 380,000 in Rockland and Putnam Counties, whose nearest Amtrak station is Croton-Harmon). Similarly, on the VIA Corridor, your line west of Toronto skips all the important suburbs, like Mississauga and Oakville, while serving the relatively marginal industrial town of Hamilton.

Nicholas: you’re not honestly proposing to connect New York to Montreal via Vermont, are you?

I love the fantasy maps! I do have one suggestion for recalculation though. Your midwest 2 route (Cincinnati-Dayton-Columbus-Akron) should actually go to Cleveland instead of Akron. This is the known as the 3C corridor (Cincinnati-Columbus-Cleveland) that is a part of the Ohio Hub proposal. The existing rails just don’t go from Columbus to Akron. In fact, Akron is kind of left out in the cold completely:

Interesting reading…I can only hope that any of it will come to pass. FWIW there is no Southern Coast route anymore. IIRC it was suspended just prior to Hurricane Katrina and has not been reinstated. MOE is no longer a valid Amtrak stop. I utliize Amtrak for my business travel as much as possible, but am usually thwarted by lack of available ground transportation on the far end. My “home” station is FAY, which has pretty decent access, comparatively speaking.


Having looked at the methodology now, I think there are two big methodological problems with your idea.

First, you don’t justify the choice of the formula, either theoretically or empirically. This is wrong; normally, when an economist wants to invoke a formula, he’ll typically argue for the form of the formula theoretically, and then try to estimate constants. In your case, there are at least two unexplained constants, the power to which you raise D and the power to which you raise Cs/Cl; you assume both are 0.5. You also ignore other important constants like economic integration, as measured for example by the volume of air travel between each city pair; this arguably measures demand better than distance.

And second, your metro populations are based on MSA numbers. The MSA measurement is a bad proxy for the number of people with good access to a station. For example, it underestimates the Bay Area, which is really one metro area consisting of both SF and SJ, and South Florida, which is on a long, narrow corridor. It also ignores the feasibility of station placement; there’s basically no way to serve the Inland Empire without missing half the population there.

It is going to be a tough transition from “America runs on trucks” to “America runs on rail and trucks”. The better the preparation, especially during the temporary low oil prices, the better.

BTW: I disagree w/ Alan about Binghamton connect. In term of passenger connect, it would be better to have some intermodal transit to Syracuse from Binghamton, then rail East or West. Ithaca is even easier, either transit to Syracuse or a closer stop where there is existing rail traffic.

Historically, trains ran into Pennsylvania from Binghamton, so as long as rail interconnects to NYC before Philly, it makes more sense to work on rail development South rather than Southeast of Binghamton.

While the PDF focuses upon American rail development, it may be shortsighted to ignore our neighbors to the North and South. In other words expand the view so that the greyed out area is further away and, at the very least, show some of the interconnects with Canada’s better rail syste,.

Great job on this piece. For the last couple of weeks I’ve been working on a similar map of rail in the U.S. The only routes that I think should be looked at are your Texas routes of San Antonio-Austin-Houston. San Antonio to Houston should be linked by HSR as should Austin to San Antonio. Also, College Station to Waco for those Houston travelers going to Dallas. A HSR line from Houston to Baton Rouge is also ideal, and would significantly boost the economy of the cities along that route.

My other suggestion would to run a HSR line from St. Louis to Jackson, Mississippi through Memphis, but I can see the other side of the argument as well with their being very little population between these cities so standard rail would be more cost-effective.

I just finished my master’s thesis evoking your exact sentiments but expanding on the angle that our bi-modal system currently is failing us and we need a circuit-breaker between the airlines and the endangered auto (ie: nearly bankrupt auto companies and fossil fuel issues). The airlines are not profitable (compare post-deregulation stats quarter to quarter vs. pre-deregulation); the air traffic control system is from another eon; and if more airlines go bust then our traveling public will be stranded (see Hawai’i; ATA + Aloha Airlines goes bankrupt stranding passengers and damaging tourism industry). With the current credit situation none of our major airlines could step up and purchase a failed airline as was the case with PanAm and TWA. Imagine if Delta/NWA went bankrupt? Who could buy it and maintain their passenger share? No one! It’s against the law for foreign ownership of an American airline don’t forget. Just one of the angles by which we can move forward with rail is pointing out the fragility of the air system.

Overall, the thesis took the tack that the policy orientation of modes competing for funds will continue to hinder the overall efficiency of our transportation system and that has to change. In order to arrive at such a goal however, planners need to link a future rail system with major airports ie: Newark, Dulles, Atlanta, Houston, Dallas, Denver, SFO, LAX etc. If true high-speed rail doesn’t connect to major airlines the myriad of airline/pilot/Boeing constituency groups will “rail” against rail.

Luftansa was successful with linking their major hub in Frankfurt with a rail operation from two nearby cities to feed their profitable int’l flights. Few airlines make that much money on regional jets anyway (see: Independence Air). IATA reps have said multiple times that they wouldn’t mind if rail could take over for this segment so that airlines can focus on profitable routes (TransCon and Int’l). Better yet rail and air can partner up; meaning efficiency.

So my point is these these modes need to connect and be non-rivalrous to an extent. Additionally, I love the graphic representations.

It’s interesting that almost half the comments are quibbling over details, and there will be more than enough time for those quibbles in the development of each corridor. In the case of the California corridor, you are wise to defer to the Commission’s existing work. Once an idea is that far along, it deserves support even if it’s only 80% right.

(Personally, as a longtime Californian, I think the Commission’s alignment is excellent given the obstacles. This amazes me, as I usually can find something to quibble on. I had to settle for wondering if they shouldn’t have a Burbank Airport station, and that’s an outrageously tiny quibble given the scale of the project.)

There’s a serious question about the importance of a “national” network, since HSR’s optimal performance is for trips of 300-600 miles or so. Federalizing a problem is not always the answer. I’m hopeful that, as in the EU, this can be a project that state governments lead, forming multi-state groupings only ad hoc as they already do in the NE Corridor. It’s striking how many of the strongest projects are largely within one state (CA, TX, FL, and NY to name only the most obvious) and in those cases the Feds’ best role is to offer a fair share of the money and get out of the way.

This plan is useful not so much as a Bible telling us exactly what to build, but as a general sense of the size and shape of the challenge. Good, work, Yonah, and thanks for the effort.

That’s nice work, but Alon Levy has an important point about the scoring formula. How does one derive it?

I think that some such formula could be derived empirically if one had Amtrak’s passenger data available, sorted by riders’ origin and destination stations. One could then find the nominal scheduled time and perhaps also the average actual time for each pair of stations, and use that as an additional input. The number of trains per day could also be significant, so we could use that also. The result would be

(Number of riders) as a function of (origin-city population), (destination-city population), (travel time), and (number of trains per day)

Failing access to complete ridership data, one could use overall data for various trains’ riderships, especially trains with big destinations at or near the ends of their routes.

Turning to Metropolitan Statistical Area population as a proxy for potential-passenger population, I agree that it is not the best possible choice, but any improvements on it will require analysis of local population densities and access convenience.

I wonder about the logic of excluding the possibility of 500+ mile high speed rail lines. I would propose that at least one transcontinental HSR line should be built. You never know if someone might be more willing to take a high speed train locally if they know, psychologically, that they could, if they wanted, take that same train to the other coast.

Also the bias against long distance HSR is predicated on the idea that the public would prefer air travel over those distances and while that may be true in the present there is nothing to guarantee that situation will stay the same. Consider two things. We are assuming people will consider long distance travel more convenient and competitive in cost. Neither may stay true in the long run. Air travel may become less and less convenient via increased security and future fuel prices may make it too expensive. Imagine what would happen when the middle class gets priced out of flying? Or what might happen if businesses, that are watching their bottom line, start sending their people on trains instead of pricey airlines.

Thanks everyone for all the comments.

In terms of the methodology, I fully recognize that the formula I used for comparing different lines is no panacea; however, I simply wanted to take advantage of the data I had: the population of metro areas according to the U.S. Census.

While Alon and others make good points, I would like the point out that I derived this formula in order to take into account three things in considering ridership between city pairs: the population of the two metro areas involved, the proportion of the small metro area to the big metro area in each pair, and the distance between the downtowns of each city. My formula assumes that there is more travel between two big cities than two small ones, and that there would be proportionately more travel from a small city to a big one than from a small city to another small one. It also assumes that there is more travel between cities the closer they are (between 50-500 miles), though I squared this ratio, because I don’t think that distance increases and ridership decreases proportionately.

You may quibble with this formula, but it provides a simple method to compare ridership. By using Census MSAs, I could estimate ridership for cities and their suburbs between than using city populations or Census CSAs would allow. I would like to point out that using MSA numbers rather than city populations does seem to be appropriate considering that a high-speed rail system would exclusively provide inter-MSA travel, never intra-MSA travel, meaning that there’s little reason to calculate populations within the MSA. I’m also not particularly convinced by the argument that travel might depend on density of MSAs; for inter-city travel, people from all metro areas are able to get to the airport – why wouldn’t they be able to get to the train station? Density I think is far more important when considering transit than HSR. For those concerned about the Bay Area, let it be known that San Francisco and San Jose are in separate MSAs and I calculated them separately.

I do not have flight data between city pairs – I looked for it, but unfortunately could not find it (if someone has it or can tell me where to find it, I’d be interested!). I do not think existing Amtrak ridership statistics are applicable, because honestly they present very small numbers for a service that is far from high-speed rail. Using existing Amtrak numbers, in my opinion, would sort of be like using the transit ridership of a city with only a bus system to estimate how many people will ride a new light rail system; we’ve seen in numerous circumstances that the two numbers aren’t necessarily parallel at all.

Anyway, thanks again for all the comments. To say the least, we can be sure that the discussion level here proves the great unmet desire for HSR in this country!

An interesting idea.

One nitpick: you show “standard speed service” along the North Shore of Lake Superior between Duluth, MN and Thunder Bay, ON. Except for a short distance between Duluth and Two Harbors, no rail line has ever existed along that corridor, and due to the topography along there, no rail line likely ever will.

Local information. You’ve got the Ithaca, NY connection badly screwed up, so please revise it.

You need a standard-speed line from Ithaca to Syracuse. Connecting via Albany (!!!!) is insane. There’s a lot of traffic in direction from Ithaca north to Syracuse and points beyond.

Based on the topography and minimizing the need for new ROW, this would most likely run east from Ithaca to Cortland (also in bad need of rail service), and then straight north to Syracuse.

The map distorts the geography of central NY in a distracting way: Ithaca is actually *NORTHEAST* of Elmira, while the map shows it *WEST* of Elmira. Given that, service from Binghamton to Ithaca via Elmira would be vastly inefficient And it doesn’t even follow the existing line (which departs at Sayre, PA). Instead, Ithaca to Binghamton needs to be connected more directly. Even a route via Cortland (which is northeast of Ithaca) would be better than one via Elmira in terms of speed. If the goal were cost containment, the branch from Sayre would be used, and would still be better than an Ithaca-Elmira train.

Regarding Binghamton, NY (which has NO P, by the way — another map error) — you want the direct Lackawanna Cutoff connection from Scranton to NYC. The route from Binghamton to Albany is actually not really very appropriate. Binghamton-Albany-NYC would be significantly slower than Binghamton-Phiiladelphia-NYC. And the intermediate-point ridership would be very low; that’s a genuinely low-population route.

Binghamton-Cortland-Syracuse, on the other hand, would provide faster access to “the north and west” for Ithaca, Elmira, Binghamton, and Scranton. Therefore it should be included.

Eventually, both Scranton-NYC service (Lackawanna Cutoff) and Allentown-NYC service (any of several routes) should be added to the map too.

And in *another* area — the existing Amtrak Adirondack route is not the greatest even though it’s the one usually proposed for high-speed service. The west (NY) side of the NY-Vermont border simply has fewer people than the east (Vermont) side. Running up the east side of Lake Champlain is the logical thing to do, along the Rutland-Burlington-St. Albans corridor, before reaching Montreal. This is approximately the Vermont Railway route occassionally proposed for upgrades to passenger service.

Yeah, these are all quibbles. But getting the city connections right *matters*. The fact that Detroit currently lacks a reasonable rail route to NYC — change via Chicago?!? — is very important. Building a rail route to Ithaca but making it stupid would be a way to handicap it badly.

Thanks for all of your hard work, this looks great. I too have quibbles with certain connections, particularly in Ohio and the greater Midwest. I think that we should remember the potential for high-speed rail to create new economic growth. Therefore, somewhere like the Youngstown-Warren metro area in Ohio would definitely need to be connected two of the following; Cleveland, Pittsburgh or Akron. A high-speed network would create strong competition between Midwestern cities located close to one another, and could help revive dying towns like Y-town.

We all hope for a Federal program, but the Governor of Ohio has authorized Phase I of the Cincinnati-Columbus-Cleveland route in this year’s budget. Let us hope it passes the State Legislature.

Here’s the deal with airports. Lot of planes fly from O’Hare to Milwaukee, but no PEOPLE fly from Chicago to Milwaukee (O’Hare-Milw. used for illustrative purposes–same idea applies lots of places). Those flights are bringing connecting customers from, say, Seattle or Denver. People are not going to ride trains from Seattle to Milwaukee; nor are they going to get off at O’Hare and travel 20 miles by bus or cab or subway to the train station to then ride a train to Milwaukee.

Building a HSR link from O’Hare to the Milwaukee airport (and continuing downtown) makes enormous sense. Integrating baggage handling and ticketing into the transportation service is even smarter. If United thought of itself as a transportation company rather than an airline, they could expand their service and reduce their costs. They could sell their takeoff and landing slots in Milwaukee, free up their planes for longer haul (more profitable) flights and actually provide more reliable service (think snow closings of O’Hare in winter and thunderstorms in summer–probably won’t stop the train).

The airports already have infrastructure in place that does not exist at most rail stations:: hotels, rental cars, parking, etc. that make these kinds of connections the smart way of doing business. And conceived intelligently, it could draw the airlines in rather than making them opponents. Add Minneapolis, Indianapolis, Detroit, and even St. Louis to the O’Hare network (with a few intermediate stops) and you’ve just eliminated the need to build the $20 billion third Chicago airport. Now that’s a smart investment!

While the general idea of a high speed rail network is something nearly all of us here would support, I find this map curious.

Look at Ohio. You can’t go from Columbus to Toledo, Detroit, Chicago, Indianapolis or St. Louis. There’s no direct Detroit-Toledo-dayton-Cincinnati route and no direct Columbus-Pittsburgh route.

At the same time, there is a curious stub end route to Lima and a weird line from Ashtabula to Charleston, WV via Youngstown, Wheeling and Parkersburg. Both of these would be dogs as far as ridership goes and the latter would cost a fortune to build in the mountainous territiory south of Parkersburg. Ironically, Youngstown would have no direct east-west service, just this odd north-south route.

Even the 3-C Corridor does a dog-leg at its north end, which will slow running times and push up costs. Same for Cleveland-Toledo-Chicago, which does a dip down to Ft. Wayne and back up to hit South Bend.

It looks to me as though the map was drawn without regard to present and past travel patterns. It also looks like costs of building in some locations were not taken into account.

I think we’d be better off to stick with a map we already have—The Ohio Hub—and go from there. If were up to me, I’d start with what we had running in 1962 (to pick a year—the map is at NARP’s website: and overlay that with the Midwest High Speed Rail Initiative, the Ohio Hub and other services. That would be much more logically laid out, I believe.

Finally, I think that we are fast approaching a time where we will be forced to make radical changes in the way we move people and freight. There will be a massive shift to rail and that means we will be more focused on moving the masses and keeping the nation mobile, rather than building a lot of very high speed lines.

We likely will build a lot of 90 mph lines, 110-125 service on key corridors and a few 125+ mph routes. Frequency will by a key part fo this. Many routes could support a dozen or more round trips even now. Some could carry service 24/7.

In addition, we could finance these improvements by having some passenger trains carry containerized, high value shipments (think FedEx, USPS, UPS, etc). Bringing more players into the mix makes financing easier, since costs would be spread more.

Containers could be roll-on/roll-off for service to smaller towns and intermodal containers for larger shipments.

Electrification? Yes, at least on the busiest corridors, especially if we get into a Peak Oil scenario. We will have to move away from dependence on diesels. Stiil, electrification is expensive and can’t be justified for secondary routes, at least not right away.

Finally, I just want to that I am not here to pan what’s been done. I do think the site is well put together and could really be of value as we move forward. I would suggest that the author study plans put forth by others and incorporate them into what he has.

Very nice article and nice feedback. And an important step toward a solution to a critically important problem that we all share.

A few comments based on a several decades of exposure to ideas and discussions re transportation.

1. As noted previously the solution requires really big thinking and really big ideas. Apollo Project-sized thinking and ideas. The existing rail system and its 150 year-old concept cannot be allowed to be the tail that wags the dog. The problem requires Jeffersonian-level visionary thoughts to be applied.

2. NASA has had a mandate for decades to study transportation issues ranging from earth orbit to our front doors and everything in between. They’ve studied everything from SST’s, SuperJumbo jets and airport capabilities to process 800 people and baggage unloading from one flight to connecting multimodal transportation and infrastructure. It would seem logical to have NASA deeply involved in the planning of this enterprise. The benefits being access to decades of investigation, NASA’s left-brainpower (good critical thinkers) and their (theoretical) separation from political and business influences.

3. An Idea Too Big: A Short Story
In 1991, a woman stood up in the Hingham, Massachusetts town meeting. The topic under discussion was the town’s position on the planned reactivation of a pre-Civil War rail route that ran through the town. The “Greenbush Line” was planned for commuter service. The woman, a critic of the Greenbush Line, cited a proposal made by a European company back in the 1970’s to the Massachusetts Bay Transit Authority. It was proposed as a solution for high-speed transit for Boston’s expanding South Shore communities. The concept proposed a monorail that ran in the median of Massachusetts Route 3, an important north-south limited access highway. It took advantage of all of the connector routes already in place. It settled NIMBY issues. It was a green solution. It would involve high-tech MIT-level problem solving that Massachusetts has long been famous for. Best of all it could become a revolutionary national model for high-speed transit solutions. It made such terrific sense.

Of course the idea was dismissed and ultimately “the easy solution”, the diesel-powered, 40 mph, already obsolete, rail line was built and opened a few years ago. The Big Idea was dismissed because it was literally too big an idea for the small-thinkers and vision-impaired that I normally associate with most public problem-solving. Certainly it was a concept with difficult technical problems to overcome. But so was the Apollo Project. The missing ingredient was a JFK-level vision with its “we do it not because it is easy but because it is hard” inspiration. I’ve often wondered if that project had been built how much we would have learned from that model that could have been applied to high-speed national transit system today and in the future.

4. The above concept would well apply to the Boston-Washington corridor since Amtrak’s existing rails are speed-constrained due to their ancient location. A totally new solution would allow truly high-speed travel here. And the problems it would be forced to solve could be applied elsewhere, even globally.

5. It’s time to reread JFK’s speech at Rice University made on September 12, 1962.

Comment #39. Nick. You got that right.

I hate to say it though, I think that America has passed away. Long dead and now the new America stands confused and in disarray.

Simple question… why does the plan not call for a connection between Dallas and Tucson?

ALSO, we need a President calling for a national system… and a legislation to write a federal law for it, including an SERIOUS eminent domain clause and a large budget. This SO could be done within 20-25 years with the right people running it and the right technology applied. America needs the economic stimulus and American people WOULD LOVE to take an alternative to flying these days.

Somehow, greed and corruption in the automotive and airline industry will prevent it though. You need a megalomaniac business man (who digs trains) to get behind this, to manipulate the nation’s status quo in the right manner. Who is that gonna be?

I like the idea, but I would change one thing in the Western Section. Instead of two routes leaving Los Angeles, one to Las Vegas, one to Phoenix, I would put only one. Eliminate the route to Phoenix and build a Phoenix to Las Vegas route. There are interstates from LA to Vegas and from LA to Phoenix, but there is no interstate from Phoenix to Vegas. These are the two largest neighboring cities in the country without an interstate link. This would be a route monopolized by the HSR. Also, it is 275 miles from Vegas to Phoenix and 357 miles from LA to Phoenix, so this would mean less miles of rail and therefore a cheaper cost to build and maintain these routes. At speeds of 200 mph, this would add not even an extra hour and a half of travel time between Los Angles and Phoenix. Driving the shortest of the two routes from Phoenix to Vegas takes about 5 and 1/2 hours, and thats on US-93, with no interstate the entire way (there is another route that uses some interstate travel that is an extra 63 miles and an hour & 10 minutes). Vegas to LA is 215 miles, so that would be a little over 1 hour travel time. The total trip from Phoenix would be 3 hours travel time, tops. That’s better than the interstate travel time now on I-10, which is almost 6 hours.

Leave a Reply to Buckeyeb Cancel reply