06 October 2009

Ridership Math

With all the focus on high speed rail, it might be useful to review and recalibrate expectations regarding ridership on the peninsula rail corridor. Fire up your calculators: the conclusion may surprise you.

Caltrain Ridership Projections

According to a recent electrification update, Caltrain is now expecting 72029 weekday boardings by 2035, up from about 40000 today. That pencils out to an extremely conservative annual growth rate of about 2.3%; for comparison, ridership has grown at an annual rate of 8% over the last five years. The number of weekday trains is assumed to grow from 90 (formerly 98) to just 114.

Based on those unambitious projections, and assuming 35000 weekend boardings in 2035 (Saturday and Sunday, in similar proportion to today), the total annual ridership adds up to 20.0 million boardings. Turning back the clock from 2035 to 2030 using the average growth rate, we get 17.9 million annual boardings in 2030.

HSR Ridership Projections

The CHSRA's latest ridership projections were published as part of their 2008 Business Plan. The plan assumes (see page 5) that by the year 2030, a whopping 222 trains per weekday will serve San Francisco. The ridership is estimated based on two different fare levels, at 50% and 77% of air fares. Tallying all the regional ridership totals from the table on page 6 (reproduced at right) that include the Bay Area as origin and/or destination, the total year 2030 ridership is 28.6 million yearly boardings at 50% of air fare, falling to 20.2 million boardings at 77% of air fare.

Not all of these riders will go to peninsula destinations. Those starting or ending their trip in the Bay Area's most populous city, San Jose, should not be counted as peninsula corridor rail passengers. Suppose that 13% of trips that begin or end in the Bay Area do so at San Jose, a portion commensurate with San Jose's fraction of the total Bay Area population. That leaves the remaining 87% to travel for some distance along the peninsula, amounting to 25.5 million riders (at 50% of air fare), or 18.1 million riders (at 77% of air fare).

HSR will be operated to generate maximum yield, just like an airline striving to maximize the difference between revenue and cost, in order to help finance the expansion of the system. Since every ridership scenario shows that revenue increases with fares, even as ridership drops, we can reasonably eliminate the 50% of air fare scenario as a fantasy that no profit-seeking rail operator would ever consider.

So, using the CHSRA's own ridership figures at the 77% of air fare level, annual peninsula HSR ridership (counting only those riders whose trip includes the peninsula) comes to about 18 million.


The ridership estimates, when taken at face value, show that Caltrain and HSR will each have about 18 million annual riders by the year 2030 (as revealed by squinting at the figure at right). Factoring in the conservatism of Caltrain's assumptions and the optimism of the HSR assumptions, one can reasonably conclude:

Caltrain ridership is likely to outnumber HSR ridership on the peninsula for decades to come.

Who would have thought? And dare we hope this will be taken into consideration as the peninsula rail corridor is remodeled?


  1. HSR will be operated to generate maximum yield, just like an airline striving to maximize the difference between revenue and cost, in order to help finance the expansion of the system. Since every ridership scenario shows that revenue increases with fares, even as ridership drops, we can reasonably eliminate the 50% of air fare scenario as a fantasy that no profit-seeking rail operator would ever consider.

    I would have to disagree. This is voodoo economics which assigns zero depreciation cost to the infrastructure.

    Yes, perhaps to the for-profit operator, running 2-hr headways to a place like Merced produces (on paper) biggest operating revenue, but that service level isn't necessarily in the interests of taxpayers who paid to build the system and will be paying the long-term bonds and depreciation costs. As well, it isn't in the interests of highway users, who can expect more traffic from potential customers turned off by high ticket prices.

    Admitedly, yield-management is a complicated issue. Some rail operators do follow the model "70%" airline pricing model -- but they are generally less successful in shifting mode share.

  2. Caltrain ridership is likely to outnumber HSR ridership on the peninsula for decades to come.

    I trust you were reiterating this an obvious point. Of course, the local and regional demand is greater than long-distance travel demand! This is a surprise only to CHSRA.

    Now to achieve platform inter-operability, opening station platforms to all trains... Naturally, during the commuter rush hours, more Caltrain trains will be using the TBT platforms than HSR trains, even assuming the long dwells for the HSR trains. It makes obvious sense to be able to shuffle services on the plaforms according to varying periods of peak demand. Time for PRP staff to earn those salaries.

  3. Clem, while ridership on Caltrain is likely to be higher than on CAHSR, revenue is going to be higher on CAHSR. This is in line with experience in Japan and France - for example, on the JR-West network, the Sanyo Shinkansen contributes 2% of ridership and 40% of revenues.

    Drunk Engineer, running 2-hour headways to Merced doesn't make economic sense. JR Central finds it worthwhile to run about 2 tph on the Kodama, making all stops, including in small, auto-dependent towns. The extra cost of building the tracks through Merced is small, and once the tracks are there, having some trains stop there costs even less, while generating a decent amount of revenue.

  4. @ Clem -

    I couldn't find the reference to the 222 HSR trains/day in+out of SF in the 2008 Business Plan.

    There's mention of 230 weekday trains (115 in each direction) on PDF p11. However, that's for the fully built-out system, so "each way" isn't really meaningful. Traffic between LA and San Diego and traffic between the CV and SoCal will bypass SF.

    Supporting document 5 suggests (PDF p14) that SF would see 9-10 tph each way during a 6-hour peak period on weekdays.

    Not sure how you get to 222 from there.

    That said, I agree CHSRA's math for SF is an unnecessary tour de force. After HSR ridership reaches a sufficiently high level, it would be perfectly acceptable to terminate some HSR trains in San Jose - not because there's Caltrain to transfer to but because the South Bay is the trip origin or destination anyhow.

    Also, there would be political pressure to construct a spur from SJ up to Oakland.


    The fare levels remain undecided precisely because public investors (federal, state, county, city) want to maximize utilization rates and minimize demand for runway/highway expansion and private investors - without whom the whole thing may never get built - will want to maximize profits.

    You may well argue that public agencies will also want to maximize profits because the phase 2 spurs will need to be funded out of operating revenue. In practice, the currency of politics is votes, not dollars. If voters want HSR fares to be affordable for all, they will be - even if that means some public funding for phase 2 after all.


    Your point regarding Caltrain ridership is well taken. It makes no sense to cram Caltrain into just two platform tracks at the TTC. Hence my suggestion of running strictly regional HSR trains between SF and SJ/Gilroy...

    Btw: 72079 Caltrain passengers per weekday in 2035? Rather a precise forecast, don't you think? Let's be honest and call it e.g. 70-75k. Spread over just 114 trains, this implies Caltrain consists will become 40-50% longer.

    Will the platforms at stations up and down the peninsula be long enough to handle those?

    Will pedestrian flow capacity to and from platforms and stations be sufficient?

    Will connecting transportation capacity (car parking, local roads, buses, bike paths) be sufficient to handle the higher throughput peaks?

  5. Be careful in comparing CAHSR with the Sanyo Shinkansen: Japan has very high tolls on intercity motorways, which make for a very different incentive structure than our free interstates. Also, keep in mind that many urban commuter services are operated not by the JR companies but by various privately-owned companies, these obviously not being part of the JR West total. And of course, you have to not only look at revenue per rider but also revenue per passenger-mile and revenue per service hour, as well as just plain old profit margin. That said, the Acela has by FAR the highest revenue per passenger of any Amtrak route, somewhere around $130, about 7 times higher than the $18 that the Surfliner gets. But the Acela also gets only a tiny fraction of the ridership of the NYC-area commuter trains.

  6. I've looked at the ridership numbers in depth, from my perspective as a microeconomist.

    Don't take them too seriously - there are major issues including the use of extremely non-random sampling to generate estimates.

    That said -

    1) There are WAY more local trips taken. A 2000 Caltrans survey referenced in the ridership analysis
    shows an average household takes 8.6 trips a day. Of these, only 1.9% of these trips are between regions and most of these are under 100 miles.

    As a percentage of trips, trips over 100 miles (the primary market for HSR) are only 0.4% of trips taken.

    Yes, a decently run Caltrain (TBT, local bus/ shuttle connections, not exorbitant ticket pricing) should kick the pants off HSR.

  7. @ Elizabeth -

    "a decently run Caltrain [...] should kick the pants off HSR."

    It's not exactly a competition, is it. Of course, long-distance HSR isn't going to match Caltrain in terms of trip units, any more than airline passenger volume across the Pacific is ever going to exceed that between NorCal and SoCal. That's no reason to stop flying across the Pacific!

    To get a better idea of bang for buck, you need to at least look at passenger-miles delivered, if not at profit per passenger-mile. Commuter railroads tend to operate at a loss, so they need subsidies to keep going. Not so with HSR.

  8. I do believe that HSR, as a fierce competitor to airlines and one for which the major costs are capital ones, has the potential to cover operating costs.

    Given the issues with routing and station locations and the statistical problems with current ridership forecasts and given a budget of $40 billion for phase 1, I think the potential for THIS HSR project is very much an open question.

  9. Arcady: in Japan, the Shinkansen has a much higher percentage of passenger-km than of passengers, and an even higher percentage of revenue. On the JR-West network, 2 billion annual riders' worth of commuter rail generate about the same revenue as 60 million riders' worth of Shinkansen.

    The tolled expressways in Japan don't matter too much. Rail market shares aren't higher than anywhere else controlling for travel times - if anything, they're lower than in France and Korea.

  10. Alon, have you ever been to Japan? Seriously, because if so, you must have been drugged or blindfolded while there. You bring up the Shinkansen so much in respect to other rail services and urban transit in Japan, yet your biased analysis consistently misunderstands that local and regional travel demand dwarves long-distance travel demand. The local transit systems of Japan carry VASTLY more riders than the Shinkansen, which is like a cherry on top of the much larger transit grid.

  11. I'll bet twice Bob Doty's fully burdened annual salary that Caltrain will carry twice HSR's ridership on the peninsula even 10 years after HSR opening.

    With the caveat, of course, that Caltrain service isn't so completely fucked over by HSR -- which looks very much like it will be the case, thanks in no small part to Caltrain -- that it is artificially kept under-porforming and artificially inconvenient. Things like terminating 75% of Caltrain short of downtown San Francisco, or not running a coordinated, same direction timed transfer local/express service pattern, or not having a base headway of 20 minutes or less for 18 hours a day.

    Foamers, especially American foamer tourists with Eurail or Japan rail passes, get SO EXCITED about the pointy nosed zoomy trains that they completely fail to see, let along begin to understand, the much larger and much more important role that regional and urban transportation plays. Don't bore me with buses, trams, and commuter trains and all your tiresome carrying on about integrated timetabling and integrated ticketing ... I want an exciting flight level zero airline! And a cab ride!

    PS "passenger-km" is the first refuge of the corrupt scoundrel. It's the way we end up with systematic book-cooked fraudulent "efficient" BART extensions to places like Livermore while we completely screw over bus service in places like San Francisco. Ignore the construction cost, ignore the miserable ridership, just bury it all in a system-wide fare recovery number.

    Transportation is supposed to serve people, not "passenger miles", and it is profoundly anti-democratic -- not to mention anti-environmental and wretched economics -- to privilege the least efficient and most destructive and most fully-internalized-accounting costly modes and passengers over shorter trips made by a different class of human. But hey ... pointy nosed trains! Kewl viaducts! Tunnels galore! Wheeeeee!

  12. "> > a decently run Caltrain [...] should kick the pants off HSR."

    "> It's not exactly a competition, is it. [...]"

    Ummmm... that's is EXACTLY what it is.

    HSR gets 2/3 of Transbay Terminal and Caltrain gets to dump most of its passengers where they don't want to go and where the voters were promised they wouldn't go.

    HSR demands 9tph of capacity, and Caltrain ends up with the all-local dregs, or very little more than that, because we "only" have four tracks available and obviously the vital (ignore that they're empty) trains via Los Banos must have priority.

    HSR dictates the infrastructure for the entire peninsula via fairy-tale levels of "projected" service, shunts Caltrain to the outside.

    HSR (= PBQD) demands that Caltrain not operate between San Jose and Fremont or between Redwood City and Fremont, and instead that a super-costly, intrusive, globally unprecedented level of suburban corridor HSR-regional track and traffic interference (50+ miles SF-SJ) be imposed, to the inevitable cost of the lower-status (albeit far more used!) regional service.

    HSR so over-specifies and over-builds unecessary infrastructure along the peninsula that (a) the money runs out, (b) voters are rightfully pissed off and (c) current and future money (capital and operating) for improvements to Caltrain service that might have had real, local, tangible community benefit goes away because incompetent, American-highway-mentality, engineer-lead mis-design burns all the political capital and voter goodwill.

    (This last is where the huge disaster will occur. Once CHSRA -- explicitly and unambiguously, in the form of Pacheco -- shouted from the rooftops that cost was no object, the floodgates were open to systematic cost over-run and consequent under-delivery and non-delivery of expendable ancillary stuff, like regional train service ie Caltrain. And no, this is no whacky conspiracy theory -- it describes exactly what PB its duopoly BATC partner Bechtel have done over the last 30 years in the Bay Area.)

    There is in fact a real competition here. The money comes form exactly one place -- taxpayer's pockets -- and HSR is successfully demanding that only its contractors and its consultants should wallow in the dump-trucks full of public cash.

    "Not exactly a competition" would start off with an integrated analysis of total corridor travel demand -- not split artificially between artificially separated trains, platforms, stations and operators -- and with an integrated service-AND-infrastructure analysis of how best to serve demand.

    It doesn't start with one party effectively seizing control of a public right of way and unilaterally imposing technical requirements and unilaterally demanding the capacity (= tracks, = percentage of the available public right of way) to run fairy tale numbers of imaginary trains.

  13. PS Drunk Engineer: I'll buy you a drink.

  14. Adirondacker1280007 October, 2009 19:35

    Hence my suggestion of running strictly regional HSR trains between SF and SJ/Gilroy...

    It's your blog. Do whatever you want. Normal people are going to call the train painted red and white that rarely if ever goes south of Gilroy, Caltrain. If they run on the express tracks and go really really fast making few stops they'll call them Caltrain express trains.

    Richard, switch to decaf. :-)
    The underlying assumption seems to be that the red and white trains have cooties and the blue and gold trains are going to recoil in horror or vice versa. What happens if they all share the four tracks?

  15. Richard: okay, forget passenger-km; look at revenues. On the JR Central network, the Shinkansen is 28% of ridership, and 91% of revenues (link).

    And before you complain that connecting transit is so vital, ask yourself how Lyon and Marseille, which by Osaka and Nagoya's transit standards are Houston and Dallas with better weather, generate higher rail mode shares. (SNCF considers its breakeven point with air to be 4.5 hours; the JR companies put it at 3.5).

    Anon: did you even read my comments? Nothing I said in them was incorrect. The vast majority of passengers are carried on local transit, yes, but not the vast majority of revenue, which is what a business-oriented investor would care about.

  16. Alon, you clearly have not been to Japan. You think the JR network is the end-all and be-all of railroads in Japan. The Shinkansen makes the most revenue for JR Central, but that's because it is the "main trunk-line" of the JR Central network. Duh...

    The JR network is an intercity network. It doesn't handle anywhere near the number of riders of the urban and regional transit systems, which includes at least 16 private rail systems! Get that?!? They actually produce positive revenues, and the capital expenses have been paid off. The private railways make their money mostly off property development, but you simply misunderstand transportation economics. And that's being kind.


    Unlike the urban and regional private railway groups in Japan, the Shinkansen HSR has never been able to recover its heavy capital costs, and this is why government support has been necessary. Please don't pretend to know which railways in Japan actually cover their operating AND capital costs.

  17. The vast majority of passengers are carried on local transit, yes, but not the vast majority of revenue, which is what a business-oriented investor would care about.

    The #1 investor in CHSRA (by far) is the taxpayer. What investment return should taxpayers expect? Is it to maximize revenues for the operator, or to maximize number of car/airplane trips diverted to trains?

    I'm unfamiliar with Japanese railway financing...but I don't think any government in Europe would invest tens of billions on new high-speed railway infrastructure without getting something back from the service provider in the form of expanded metro, commuter, and regional service -- even if that service is a money loser for the operator.

  18. The JR network is an intercity network. It doesn't handle anywhere near the number of riders of the urban and regional transit systems, which includes at least 16 private rail systems!

    You're wrong. JR East carries 5.3 billion passengers a year on its regional lines in Greater Tokyo, more than all non-subway operators combined. The Shinkansen is still responsible for 28% of its ticket revenues.

    Similarly, JR West is the main carrier of commuter rail in Greater Osaka, with 1.4 billion passengers a year on its regional lines; nonetheless, the 60 million passengers of the Sanyo Shinkansen account for 44% of farebox revenues, compared with only 39% for the regional lines.

    Unlike the urban and regional private railway groups in Japan, the Shinkansen HSR has never been able to recover its heavy capital costs, and this is why government support has been necessary.

    In New Departures, Anthony Perl explains how both the Tokaido and Sanyo Shinkansen paid off their construction debts within a few years, encouraging the government to build additional lines. Those additional lines, especially Tohoku, get more ridership than Sanyo nowadays and would probably pay their debts back if JNR hadn't been privatized before they were completed.

  19. Drunk Engineer:

    What investment return should taxpayers expect? Is it to maximize revenues for the operator, or to maximize number of car/airplane trips diverted to trains?

    I would say it's a combination. Taxpayer investment should be judged based both on financial metrics and on externalities. All of the relevant externalities can be measured: transportation convenience, development, intercity trade, pollution, emissions. For example, SNCF's study concludes that, in net present value, HSR's financial and social benefits outweigh the costs by a factor of 2.9.

    I'm unfamiliar with Japanese railway financing...

    It's pretty much the same as in Europe. Even now that the mainland JR companies are private, they're considered quasi-public, and invest in both local and intercity transit. (It helps that Japan is so rail-oriented that even its local rail lines make money). If anything, it's more tied together - in Europe, DB and SNCF operate regional lines with regional subsidies and keep them off their main intercity books, whereas the JR companies bundle local and Shinkansen traffic together.

    In fact, the privatization of JNR came from over-expansion of local lines. Beginning in 1964, JNR expanded like crazy, often for political reasons, building multiple unprofitable local lines. Sometimes this over-expansion is blamed on the Shinkansen, but the truth is that the Tokaido Shinkansen was funded by a World Bank loan, which was repaid in full in the 1970s, and the other lines were built only after JNR slipped into permanent deficits.

  20. Sigh. Alon, the JR network is the descendent of the Japanese national rail network, the intercity network, and as you note, this intercity network has been extended well beyond the Shinkansen trunk lines. The JR Central still is a "pure" intercity network cutting across rural areas, so it is no surprise that the JR-Central Shinkansen has 28% of ridership and 91% of revenues. That's equivalent to saying the airlines make more revenue yet carry fewer people than the long-distance buses over this "fly-over" area between Osaka and Tokyo. The JR intercity network has become extremely dense around Tokyo and Osaka, Japan's two primary urban agglomerations, so JR-East and JR-West have evolved in suburban networks around Tokyo and Osaka, respectively. The JR-East and JR-West networks have evolved to carry many urban commuters due to many Japanese workers having surprisingly long commutes. Japanese workers tend to stay in the same house throughout their career, even if their job location shifts farther away. The result can be long commutes within a large urban region. The JR-East and JR-West networks serve these longer commutes, but this only strengthens the basic fact that local and regional travel is far more intensive than long-distance travel.

    So JR East moves 5.3 billion annual riders, but a full 80% of them are on the Kanto Area Network conventional lines! The Kanto Area is basically Greater Tokyo, so the vast majority are commuters and regional travelers!!! NOT LONG-DISTANCE HSR TRAFFIC!!!

    The Shinkansen mainlines only carry 15% of the JR-East volume. Since the Shinkansen is the mainline with higher fares for longer distances and exclusivity (what is an express if not an exclusion of local stops?), any operator would expect higher revenues for their mainline, so the Shinkansen gets 28% of JR-East revenues. That's nothing special, especially since regional commuters make 67% of JR-East revenues.

    The four Osaka-area private railways carry about as many commuters as JR-West, which itself serves mostly local and regional traffic around the Osaka urban agglomeration. The private railways perform better economically by covering all their costs with property development and retail. Furthermore, we can't forget the subway, buses, and taxis either.

    Let's get back to the basic point that Alon consistently misunderstands: without extensive feeder transit, the long-distance Shinkansen could never achieve the numbers it does carry, which depend on feeding off the much larger flows of local and regional transit. The Shinkansen is the cherry on top of a much larger system. You can't substitute this with mega-parking garages or just taxis! This is why it is foolish to think HSR in California will succeed without strong feeder transit. HSR has high capacity, but something has to feed it, especially when long-distance travel demand is a very small fraction of total travel demand.

    JNR actually made a profit in the 1950s, because it essentially had a monopoly on long-distance travel. Automobile infrastructure was very under-developed, allowing rail to dominate. Even today, the tolled expressways absolutely have an effect in supporting rail. Ironically, the first year JNR went into the red was the year Shinkansen opened in 1964. JNR was privatized precisely because it was bloated with too much debt and redundant workers, but the government has always shouldered the capital debt. The World Bank loan covered less than 10% of the original Shinkansen's construction costs. Building HSR has never been cheap, and no HSR system has ever been paid off without government support, period.

  21. Richard Mlynarik said...
    "Transportation is supposed to serve people, not "passenger miles", and it is profoundly anti-democratic -- not to mention anti-environmental and wretched economics -- to privilege the least efficient and most destructive and most fully-internalized-accounting costly modes and passengers over shorter trips made by a different class of human."

    Ridership does not count the number of people served either. A single passenger taking 400 shorter trips in a year compared to a single passenger taking two much longer trips that year does not mean that the first has served 200 times as many people.

    Commuting tends to generate more trips per person per year than shops, schools, restaurants, vacations or anything else, yet a more sustainable city could have a lower proportion of commuting riders because they live within walking or cycling distance of work. So monomanical devotion to ridership as a metric is itself not without its flaws.

    The fact is that effort to squeeze a complex reality into the confines of any one-size-fits-all metric will fall short. Passenger-km are more relevant to the discussion of intercity rail than ridership in terms of how much transport service has been provided to people, and is also more robust to differences in ticketing systems that create one multi-mode trip in one system and multiple trips in another.

  22. @Clem, note that this is all predicated on continuation of ultra-low to moderately-low prices for crude oil. Two or three more oil price shocks, or even just one oil price shock if it was to a genuinely high price for crude oil, and Caltrain ridership (and passenger-miles) will move well ahead of projections.

  23. Anon, what you say about long commutes is just wrong. Look at the busiest lines on the JR-East network. The Yamanote Line doesn't ever leave central Tokyo, and only sees local traffic. The Chuo Line is only two-tracked outside Tokyo and its innermost suburbs, and sees relatively little traffic beyond Takao.

    The distances involved are much shorter than on US commuter rail - e.g. Tokyo-Takao is 53 km and the entire Yamanote Line is 34.5, whereas the Caltrain corridor from San Francisco to San Jose is 74. If those are just intercity services, then so is Caltrain.

    And I don't misunderstand the point about connecting transit. What you misunderstand is my point about what already exists in the US. Los Angeles and San Francisco both have good transit connections to the future HSR stations. The reason Metrolink, the Metrorail, and BART get little ridership is that those systems are monocentric around downtown, where relatively few people work. But if all you want is to go to downtown, the Osaka subway and JR West don't offer a better connection to Shin-Osaka than the Metro Rail and Metrolink do to LAUS or BART and Caltrain do to TBT.

  24. Alas! Alon is slowly getting the basic tenet of travel behavior!

    Yes, local traffic tends to be far more intense than even regional traffic, just as it follows that regional traffic is more intense than long-distance traffic. Travel is a derived demand, and people generally want to go to the places closer to them more frequently. That's the point I've been trying to knock into your head -- bravo!

    what you say about long commutes is just wrong.

    Given the high density of Japanese cities, many Japanese workers do have surprisingly long commutes, largely due to the gradually eroding Japanese cultural tradition of housing market immobility. If you have been to Japan, you would understand this. I never said all, or even most, Japanese workers commute long distances, but it is an interesting trend given the compactness of individual Japanese cities. To compare, an interesting trend in the Bay Area is the very long distance commutes from the Inland/Central Valley to job markets in San Francisco and Silicon Valley. This is largely due to the gradually eroding American cultural tradition of large, single-family houses with big yards -- but obviously not every commuter is traveling these long distances in the region. Many are, however...

    Caltrain is a classic example of an intercity line that has morphed in function into a commuter line, and maybe one day it will become a real rapid transit line. The Caltrain line is the oldest intercity passenger railroad in California, going betweeen San Francisco and San Jose, two distinct urban areas until at least the mid-20th century. The Caltrain corridor was a segment of the SP mainline for all SP intercity trains going to/from San Francisco. What has happened with the JR-East and JR-West lines is similar to what has happened with Caltrain, although the Japanese networks are obviously far more intensive and extensive.

    In urban Japan, you can make a transit connection to just about anywhere. Even outside the city centers, the ease of walking, bicycling, or getting a taxi make the "first and last kilometer" of any trip effortless. This is simply not the case in the vast majority of urban California, so I don't know where you get this crazy idea that HSR station access is equivalent in Japan and California.

    The Transbay Terminal is California HSR's best chance of having a station with a full array of FREQUENT transit connections to MANY destinations, but that may not even be happening now! The 4th and King station has slow, limited transit connections, and the proposed "Central Subway" is a joke in terms of capacity, connectivity, and function.

    LA Union Station does have some transit, but it's still not much of a rapid transit grid for such an expansive urban region. You can only reach a small fraction of regional destinations using transit. LA has quite an extensive bus system geared to low-income riders, but relatively wealthy HSR riders are not going to be attracted to this current service.

    As for the other CA HSR stations? They are barely connected to any meaningful transit grid at all. San Jose has the world's worst light rail system operating in low-density sprawl. Gilroy, Fresno, Bakersfield, and Palmdale have the most rudimentary and down-market of transit systems. Even calling them "systems" is generous.

    Alon, I do hope you realize that Shin-Osaka was built on the edge of the Osaka central city in order to avoid the heavy expense of building HSR through central cities. It's not the busiest station in Osaka by any means. Why? Local and regional trips are vastly more numerous that long-distance HSR trips.

  25. There is a persistent assumption that regional commuter and metro rail systems can never be operated profitably. Try telling that to Hong Kong's MTR or Singapore's MRT, both of which are private companies listed on their respective stock markets and are profitable. The reason nobody runs a profitable transit system in the US is that we have bloated administration and unbelievably high union labor costs.

    Think about BART, where a system that was designed to be driverless not only has a driver on every train, but they make about $100,000 per year for performing a function that should not exist and is clearly not productive. As Dubai's new metro system opens this year, it will be the world's longest driverless system. Don't expect political pressure there to put an unnecessary, $100,000 per year driver into every train.

  26. Yes, I know that why Shin-Osaka was built outside the core urban area. My point is exactly that HSR can still be successful with such routings. If JR West can get more revenues from one HSR line connecting Shin-Osaka with Hiroshima and Fukuoka than from its entire Keihanshin, then there's hope for CAHSR.

    I got the crazy idea that HSR station access is equivalent in California and Japan by looking at which lines connect to the stations. In LA, you have Metrolink connecting most of the region to LAUS, plus the Metrorail for shorter trips. Those systems don't solve the first mile problem, but at outlying stations parking tends to be easy enough that people can just drive. SF has similar access with BART and Caltrain. Both LA and SF have more rapid transit route-km than many successful HSR hubs, such as Lyon (yes, Lyon has more ridership; this reflects access to job centers, not to the train station).

    Now, Shin-Osaka is connected to one subway line, and one regional rail line. People can transfer from elsewhere in the system, but it increases station access time. The American equivalent of that would be to put New York's HSR station in Newark and tell people to take New Jersey Transit or PATH to Manhattan.

    What you say about Caltrain being originally an intercity line is true only historically. Right now the line is not intercity, and neither is the Chuo Line; the Yamanote Line was never intercity to begin with - it was always intended as a loop around central Tokyo. So when you say the JR companies are inherently intercity, you're not describing today's situation, when they're the primary providers of commuter service in their respective regions.

    On another note: who cares that Japanese cities are dense? That doesn't necessarily reduce commute distances. Singapore is dense, too, and it has subway lines almost as long as the Chuo Rapid Line. (In case you're wondering: I'm bringing up Singapore because I lived there for a few years. Tokyo I only visited.)

  27. Andrew: Dubai is a good example of How Not To Do It. It just sank $140 million per route-km into building its automated, driverless metro. For a city with near-slavery, this is very expensive. It's more gadgetbahn than transportation.

  28. While Dubai has plenty of economic migrants (including a lot of Americans in engineering, business, and finance), near-slavery is an unfair characterization of Dubai's huge and vibrant working class.

    The cost overruns were large for Dubai's metro, but a significant part of those costs have come from a decision to build palatial stations with the goal of selling station naming rights to companies. With less than half of those rights sold by the opening of the system in September, we will all have to wait and see if the strategy was foolish or not over the long term.

    The other side of the cost equation as I understand it was the complexity of building a mostly elevated system on highly unstable and shifting sandy soils. Seems like that should have been expected, especially after all the problems they have had with roadway construction and ongoing maintenance throughout the Emirate.

    In any case, there are still very large benefits of a driverless system with respect to long term operating costs.

  29. The cost of a driver isn't high. The rail systems that are designed to minimize cost, such as Calgary's, prefer to put a driver in every car, figuring that paying the driver's salary is cheaper than paying for automation.

  30. The August BART union negotiation disclosure documents would suggest that the costs of drivers and station agents are, in fact, quite high.

  31. Are those numbers just for drivers, or also for station agents, track workers, etc.?

  32. @ transport finance expert Alon Levy: "The cost of a driver isn't high. The rail systems that are designed to minimize cost, such as Calgary's, prefer to put a driver in every car, figuring that paying the driver's salary is cheaper than paying for automation."

    Right. That explains why Muni's TWU-250A fought successfully for years to have a wage-sucking union dues-paying do-nothing parked in the second car of two car trains (and the third and fourth cars back in the dim past when Muni actually ran "trains"). And don't forget the threats of mass carnage, deaths on the track, deaths in the car that would come if this vital job were removed. You need the human touch! THe cost of "automation" (in the form of the inter-car control contacts that already exist) was going to be too high!

    And that explains why the MBTA Carmen's Union, one of the most politically entrenched and sleaze-ridden organizations around, still forces Massuchusetts taxpayers and T riders to pay for dead-weight wage-suckers in the second car on the Green Line. Not to mention the conductors on the other lines.

    And that explains the conductors on the NYMTA subway. (ALL HAIL THE NORTH EAST WHERE ALL IS DONE RIGHT!) Allowing a button in the driver's cab that says "close doors" to close the doors would be a level of automation that just isn't justified in light of how incredibly cheap MTA operating employees are.

    And it explains why Caltrain and Metrolink and all the ALL HAIL THE NORTH EAST commuter railroads cart around conductors and ticket takers.

    All of them are clearly being paid minute, povery-level wages -- and some of them are working for free, just out of the goodness of their harts and a concern for public safety -- because having these highly skilled and alert humans present is CHEAPER THAN AUTOMATION. Just like it is in Calgary.

  33. Okay, so you've established that having a driver costs money. Would you care to explain how automation costs less money? Or does your engineering degree automatically makes you an expert on all matters of public financing?

    And don't talk about conductors or commuter rail. Conductors are easy to get rid of. Drivers aren't. Going down to OPTO (and even automation on short lines) is so easy that even technologically backward cities like New York manage to do it and that most cities have done the conversion; going down to automatic operation requires new infrastructure, which is hard to install on existing lines and bumps up costs on new ones.

  34. Here's a challenge to all you rail types:

    How would you design a commuter rail system so that it would have a high level of service and low operating cots? What would be critical attributes?

    In this bizarro world we live in, capital dollars are actually much more widely available than operating dollars.

  35. "How would you design a commuter rail system so that it would have a high level of service and low operating cots? What would be critical attributes?"

    * One person operation.

    * Equipment procurement based on total life-cycle costs. Design-build-maintain contracts, so that the equipment bidder has significant incentive to control costs.

    * NO LOCAL DESIGN by incompetent, unqualified US rail "professionals". Just look at Acela. Or Muni "light rail". Grown ups must be in charge, because bad equipment is a gift that keeps on giving for 30 years. (Look at Caltrain's gallery cars.)

    * Don't over-specify equipment needs based on dumb operating practices or dumb preconceptions. Dumb preconceptions might (or might not) include "double deck trains are necessary." That should be up to first-world vehicle designer-maintainers and first-world train operators, not made a requirement because that's the way my grand-pappy did things.

    * Caltrain is a small-scale operation by pretty much any metric (other than increasingly outrageous capital program costs and agency overheads.). People with experience do this sort of thing in their sleep. Let them do so. Don't pretend it's unique, of special, or some sort of big deal.

    * No local assembly by unskilled local make-work sheltered workshop shell companies resulting in poor QA and inflated costs. Buy known-to-work equipment off a production line, and have it delivered after being tested at the manufacturer's test facility. (They all have them and know how to use them.)

    * Don't buy from companies with proven records of failure. (Paging AnsaldoBreda!)

    * Very light weight equipment = (a) lower energy costs (which are very significant over its lifetime) and (b) higher performance (= more trips with fewer vehicles and fewer crew.)

    * Minimize non-revenue time. Turnbacks can and should be scheduled for 10 minutes or less, maximum. The US practice of parking trains and crews idle for large percentages of the day is an egregious waste of money.

    * Maximize average speed in general, by speeding up bottlenecks and slow points, which are nearly always a bigger win than upping top speeds.

    The basic equation is that higher average speed and less out-of-service time = more trips provided with the same amount of $$$ equipment and $$$ crew.

    * First world scheduled and preventative maintenance, with on-board vehicle diagnostics. (See design-build-maintain above.)

    * Minimize fleet size, based on first world operating and maintenance practices.

    US rail operators typically have 15 to 20% or more of their fleet lying around out of service even at peak service hours. The capital cost (= STOLEN TAXPAYER FUNDS) is outrageous. And that's leaving aside the theoretically in-service equipment that's sitting around in slow turnbacks or dwelling too long at stations or stopped at signals or creeping through speed restrictions.

    [... continued ...]

  36. [... continued ...]

    * Integrated ticketing and integrated clock-face scheduling with barrier-free platform access and super-convenient bus-train and train-neighbourhood walking routes.

    The train arrives every 30 minutes. The bus arrives every 30 munutes. Step from the train to the bus. The same ticket works. Don't operate duplicate bus service, since east-west circulating feeders and zero-penalty transfers make Caltrain part of regional bus service.

    Inception of such basic, well-proven, repeatedly successful policies typically results in 25+% ridership increases.

    * Station accessibility has a bigger effect that people realize, and Caltrain has gone out of its way to do the wrong thing. Remember that rider trip time value includes the time wasted getting to and from the station; it's far cheaper to save 2 or 5 minutes there than it is to make a train run that much faster.

    * Stations should have a good basic level of amenity (ie a bus shelter plonk on the platform is not a station.) If people aren't comfortable waiting for a train, even for a short time, they're going to choose their comfortable cars. And who could blame them?
    Good lighting (no, it's not a prison yard), good sightlines, shelter, good and plentify and well-located information displays, good neighborhood accessibility.

    Others manage to do it. What the hell is wrong with us?

    * A simple service pattern (2 or 3 stopping patterns AT MOST, not a random hash or randomly-stopping trains), operated on a regular (30, 20, 15 minute) headways and operated over a useful span of service (16 to 18 hours, not the 6 peak hours of "commuter rail") is key to making the service attractive to other than die-hard timetable-perusing bankers-hour masochists.

    * Bums need to be put on seats OFF-PEAK, when the service cost is essentially sunk (all the big bucks go into sizing rail infrastructure, fleet size and crew size for the peaks.) Operate a service which is attractive to users other than peak commuters, and set fare policies to maximize penetration of periodic (monthly, quarterly, yearly) tickets. (Ecopass, etc.)

    Again, this is proven to work.

    * Operate according to schedule.
    4 minutes late is not on time.
    First world maintenance = no in-service failures = higher reliability.

    * Fare machines need to work.
    But ideally 70+% of riders will travel on periodic tickets, and not just daily commuters.

    * Customer service is an asset, not just an onerous cost.
    Basic 1990 stuff like accurate and real-time vehicle tracking and simple, universally-accessibly web sites are part of it.
    Ticket inspection that is frequent and random enough to ensure compliance, random and infrequent enough to not aggravate riders and burn up cash for salaries is another part.

    * Paint blue stripes and the letters "BART" on the side of the trains for all I care. Just don't run god damned 19th century commuter railroad.

    * Create and publicize a PLAN for what service will be provided in the future, and deliver that plan in well-defined, publicly-advertized, reliable steps. Nothing makes people love you (and hence be more willing to tax themselves to keep you going) than delivering something useful how and when you promised.

    Random chaos (hey kids, let's spend $300 million in San Bruno that does zero to improve anybody's train service!) isn't helpful. Public perception matters, because public money needs to be spent up front to create a modern system with much lower ongoing operating costs.

    * And, of course, design a SERVICE PLAN and the ROLLING STOCK and the FIXED INFRASTRUCTURE *together*, to avoid over-building infrastructure (Caltrain and CHSRA are just out of control in the concrete-pouring contractor-pork front), over-buying rolling stock, or over-spending in one area when the same problem (= providing service) could be solved more cheaply by other means.

    How's that for starters. NONE of this is rocket science. And as far as anybody can tell, none of it is happening.

  37. Yeah, but our freight trains are the shit.

  38. How would you design a commuter rail system so that it would have a high level of service and low operating cots?

    To add to Richard's list:

    1. Privitization.
    Put operation of the service out to bid by lowest (credible) bidder. Even "Socialist" Europe does this now.

    2. Modernize railway regulations.
    It is fundamentally impossible to operate cost-effective passenger rail in the current regulatory environment. Trying to build and run passenger trains under 1930s steam-era rules inflates costs by as much as 10x.

  39. To add to Richard and Drunk Engineer:

    - Plan your system ahead of time. Calgary minimized construction costs and maximized ridership by reserving ROW twenty years in advance, well before it even knew it was going to use it to build light rail and not freeways.

    - Keep things at-grade whenever possible: separate the ROW from other traffic with fences, and avoid elevated and underground construction unless necessary.

    - Get rail to work for local service, too. This means making multiple stops in the central city, serving different neighborhoods. It's okay to do this at the expense of fast service from the exurbs, which don't generate much traffic anyway. The RER and S-Bahn do it right, serving every major neighborhood they pass through. The LIRR is a mixed bag. BART is bad, serving only downtown SF and Mission.

    - Through-route services when possible (i.e. no Penn Station-style two-sided terminals). Terminals can turn around trains in 5-10 minutes. Through-stations can get dwells under 1 minute. Again, the RER and S-Bahn are the best practice to follow. This increases capacity, and makes increases usage for suburb-to-suburb travel.

    - Make it easy to transfer from one line to another - again, this increases usage for suburb-to-suburb travel. For SF, that would mean one major station connecting Caltrain, BART, and Muni. The best example of what not to do is Metra, with 4 downtown terminals that not only have no connections to one another but also don't have easy connections to the L.

  40. Richard,s list of desireable transit attributes could be implemented on Caltrain within weeks for no material additional cost: 1. Schedule hourly local trains so all stations from SF and SJ have a train leaving at the the same time past each hour for at least 14 hours seven days a week.
    2. Shorten local trains to match ridership. Each car shorter should save at least seven minutes on the SF to SJ local schedule. Add a Saturday nothbound morning express and a southbound late afternoon express in order to permit contemporary locals to conform in speed to the short lenghth faster locals during the rest of the week. Note: There are plenty of expresses during week days required to consistently draw down local train peak ridership to a level that can be accomodated with two gallery cars per local.
    3.Coordinate local schedules with BART,s Millbrae service. Until September, 2009 BART trains arrival & departure times have been quite stable; Caltrain Millbrae weekday Millbrae arrival and depature times are erratic with no discernable attempt to coordinate with BART schedules.
    These are among the lowest cost initiatives that could produce a material benefit. Numerous other cost-effective changes would be to implement level boarding for at least one car while meeting current PUC regulations and extending King Street light rail service into Caltrain,s terminal in order to provide cross=platform transfers for most trains.

  41. I understand how shortening trains allows running trains more often, but why does it reduce trip lengths?

  42. Faster acceleration, I would guess. Only applies for locomotive-driven trains.

  43. Adirondacker1280011 October, 2009 21:17

    I understand how shortening trains allows running trains more often, but why does it reduce trip lengths?

    Caltrain uses locomotives. Shorter trains accelerate and decelerate faster. That's one of the advantages of MUs their acceleration and deceleration doesn't vary much with train length.

  44. Back to Clem's original topic...

    Some interesting tidbits that jump out from the data:

    1. Caltrain might have more trips, but CHSRA will be running twice as many trains. Undoubtedly, CHSRA will use this "fact" to drive its infrastructure decisions (i.e. CHSRA will be twice as important as Caltrain when it comes to decisions about ROW and terminal space).

    2. In the Business Plan, CHSRA (finally) concedes that it will not dominate the market for LA-Bay Area trips. Unlike every other HSR project in the world, this one will leave healthy mode share for Southwest-type shuttle service. The "Plan" envisions an even split between air and rail. This, no doubt, reflects the fact that OAK will be more conveniently located for East Bay residents than any HSR station. The East Bay has more population than either the South Bay or the Peninsula.

    3. It would be interesting to do calculation of number of seats that will be run on the Peninsula based on the latest published "Plan". The numbers don't seem right to me. I note that the CHSRA Plan shows 4.7 million Peninsula trips -- not clear if these are Baby Bullet trips, or a service that needlessly replaces Baby Bullet service (more likely scenario). If Clem's assumptions of 13% mode share for San Jose is correct, how do they fit those 4.7 million trips into the proposed service plan, without having to turn around trains in San Jose?

  45. Drunk Engineer said...
    "In the Business Plan, CHSRA (finally) concedes that it will not dominate the market for LA-Bay Area trips. Unlike every other HSR project in the world,"

    Well, no. In Europe, rail dominates air when it offers two hour trips, and splits the market with air when it offers three hour trips.

  46. No, at 3 hours rail dominates. On Paris-Marseille, the 3-hour TGV has more than 70% of the air/rail market. On New York-Washington, the 2:50 Acela has 67% of the air/rail market. On Seoul-Busan, the 2:40 KTX has 63% of the total market, including cars and buses, and 79% of the air/rail market.

  47. "On New York-Washington, the 2:50 Acela has 67% of the air/rail market."

    Acela alone? Or total Amtrak?

    Does that compare passenger count LGA/EWR/JFK-IAD/DCA to Amtrak NY-Washington passengers, or does the Amtrak total include NY-Phila etc?

  48. Sorry, I meant total Amtrak.

    This includes only NY-DC traffic. Amtrak reports that it has 67% of the total air/rail market on the NY-DC city pair. Neither the air nor the rail number includes intermediate markets like NY-Philly, as far as I know.

  49. CHSRA propose to run a simply insane number of empty seats up and down the peninsula, thus (a) locking Caltrain out of the downtown terminal which the voters were promised and (b) wasting hundreds of millions of dollars building unnecessary, redundant infrastructure between San Jose and San Francisco just to accommodate empty seats.

    You don't have to take my word for it, just use reptilian-hind-brain level of reasoning: CHSRA propose 8 (eight!!!) trains from San Francisco to Los Banos and beyond per hour. Now either (a) these trains are full when they reach San Jose or (b) they are partially empty. If they are full before San Jose then we shouldn't stop there and should save a few billion not building a two-level, 14-track, Diridon-tastic momument to American Engineering Know-How. If they aren't full, in fact if they're less than 80% occupied, then we're operating too many trains for HSR needs, and we're unnecessarily screwing Caltrain and unnecessarily defrauding public works budgets for no valid reason.

    So how would grown ups deal with this?

    Well, there are a couple approaches:

    The first is to inter-line a subset of HSR trains as Caltrain expresses. The things labelled as Caltrain expresses start in SF, make the Caltrain express stops along the line (Transbay, Mission Bay, Millbrae, Hillsdale, Redwood City, Palo Alto, Mountain View, San Jose) and then continue to Los Banos and points south.

    Caltrain tickets don't work south of San Jose. HSR tickets do. Empty seats are occupied by commuters and gradually turn over to HS passengers, finally departing SJ full with the many hundred people an hour who seek to visit Visalia, Bakersfield, or Fresno.

    The second approach is to terminate some HS trains in San Jose and offer a cross-platform transfer (either to a Caltrain express or to a HS train following two minutes later) to that subset of passengers who wish to continue further north.

    Oh oh oh oooooh but transfers kill ridership. I read it on a blog somewhere!

    Well, BFD. Blowing out the project budget by a few billion and pissing off everybody between SJ and SF and pissing off most Caltrain riders until the end of time also hurts ridership.

    The fact is that with 8tph coming into the Bay Area from Los Banos that a same-direction transfer in San Jose would be a matter of minutes of delay.

    The way this works in your average first world advanced industrialized democracy (I'm scared to give any examples, like Arth-Goldau, or I'll tick off the New Jersey Amtrak contingent...) is that your all-stops Central Valley milk run trains is scheduled to arrive in San Jose 2 minutes before the super-kewl ultra-express which has been catching up to it all the way up the state. The CV train pulls into Track 2 of Diridon Intergalactic, opens its doors, and the Amtrak Conductor dude (with his cute little blue hat, tobacco stained handlebar moustache, union seniority, and infinite disdain for passengers) announces "This train terminates here. All out, all change. For those of you continuing to points north of San Jose, Capital of Silicon Valley, home of Didiron Memorial BART Station, please step across to the other side of the platform and catch the train that is pulling in as I speak."

    Southbound, a similar deal. A SF to LA few-stops train pulls into SJ station, the Central Valley all-stopper starting from SJ is positioned across the platform, doors open, passengers with destinations such as Fresno transfer, the fewer-stops train leaves, and 120 seconds later the more-stops train follows it.

    End result: billions saved in capital costs, fewer empty seats shuttled back and forth all day (= tens of millions annually saved in operating costs), nearly zero travel time impact, much less infrastructure and disruptive construction required (because fewer and better-used trains run SJ-SF-SJ.)

    Chances of it happening in our corrupt, incompetent little backwater: zero. Because higher cost and lower utility is always is somebody's interest.

  50. Instead of stopping the North local in San Jose or starting the South local in San Jose, could not those trains continue to or come from the east bay or even Sacramento and be timed to arrive at San Jose. If this was done with a simultaneous cross-platform transfer then HS passengers could switch to the east bay/Sacto train and locals could switch to the HS train to SF.

    On a related note it would be nice if the coast line were upgraded to provide faster local service down the coast.

  51. Richard, the way you describe the San Jose transfer is how things work in Switzerland, which is just one industrialized democracy. If you look at most other industrialized democracies, they do things differently.

    In Korea, they don't make you transfer. In fact they electrify and double-track legacy lines so that the KTX could run on them, giving people a one-seat ride to Seoul.

    In Japan, they do the same thing - and that's a country where the Shinkansen doesn't even run on the same gauge as the legacy network. JR East converted two legacy lines to standard gauge, and now runs through-trains on them from Tokyo. The cars are towed behind a real Shinkansen train, and are then split off and sent along those converted lines at lower speed. No transfer necessary.

    In France, the entire LGV system was built to allow this through-running. On some lines, like Paris-Nice, the TGVs spend as much time running on legacy track at low speed as on LGVs running at high speed. Like Korail and the JR companies, SNCF doesn't make people transfer. SNCF is so neurotic about one-seat rides it builds dedicated bypasses to make its flight level zero TGV connect pairs of cities on different lines without needing to transfer.

    I don't think they even force transfers in Germany. The ICE is run like the TGV, only with more legacy track and less high-speed track. If you want to get from Frankfurt to Essen, you don't need to transfer in Köln.

  52. Alon,

    You just don't get it.


    What planet are they (and you) living on?

    Why should I (and everybody else in California, and probably people in New Jersey through their generous federal tax donations) have to pay to run shockingly large numbers of empty seats back and forth?

    Why should our use of Caltrain -- whose passengers will remain 75% of the USERS of the public transpotration corridor -- be fucked and and fatally compromised forever by a bunch of kindergartners playing choo choo as often as possible with the shiny pointy-nosed toys?

    I'm pleased you've visited Japan. And France. That puts you two steps ahead of many others. But so have I.

    Now, next step: try to consider the real world of COSTS and benefits, and real world engineering trade offs of operations and infrastructure, not just some lah-lah-lah wouldn't if be wonderful if tooth fairies left a cute little Shinkansen under everybody's pillow land.

    What PUBLIC BENEFIT comes from planning to run 8 high speed trains per hour from Los Banos to San Francisco? Is it in any way commensurate with the PUBLIC COST and OPPORTUNITY COST of doing so?

    This isn't a zero sum game. Sorry to break it to you.

  53. What PUBLIC BENEFIT comes from planning to run 8 high speed trains per hour from Los Banos to San Francisco? Is it in any way commensurate with the PUBLIC COST and OPPORTUNITY COST of doing so?

    Based on BART-Milbrae, VTA light rail, etc, we can predict the following:

    Step 1.
    Publish EIR and business plan showing ludicrous passenger demand.

    Step 2.
    Build ludicrous amount of infrastructure to support the purely theoretical passenger prediction.

    Step 3. Profit!

    Step 4. When system goes operational and actual passenger count is just a fraction of original prediction, come up with cover story. Blame 9/11, or the dot-com bust, or say the system just needs more expansion to reach originally projected goals.


    What planet are they (and you) living on?

    The same one SNCF is living on. Look, it's one thing to bash predictions made by engineering consultants who have never worked on a rail project that was completed. It's another to bash predictions made by one of the most profitable HSR operators in the world.