I work for the Federal Government (health) and this resonates with what I see. Government staff are buried under dozens of process requirements (like Small business requirements) and leadership measures success as no one yelled at us, not “did we objectively improve things.”
> And importantly, telling agencies to try to hit a half-dozen different goals … doesn’t ensure that all those things will be done simultaneously. What it ensures is that, with no principled way to evaluate projects or say no to things, costs explode.
The cynic in me wants to believe that agency leaders would like to avoid a simple and objective measure of performance. If they are solely evaluated by ridership/dollar-spent then they could fail to improve on that metric and be charged with bad decision making. In contrast, if they claim to be balancing numerous unquantifiable objectives, then it’s much harder to criticize their performance. For example, if we point out the low ridership volume then they can point to their progress in expanding access and creating good paying jobs.
And I think there is a general tendency for all leaders and workers to resist objective evaluation. For example, I've seen this criticism applied to the anti-testing movement in education where it is alleged that poor teachers simply want to evade detection.
I’ve also seen this aversion to performance evaluation within the tech industry, particularly within discussions on the tech forum Hacker News. While there are many poor approaches to measuring software developer performance (e.g., lines of code count), there still need to be some objective evaluation. Quality measures can include annual peer review using a rubric of dimensions.
Yet there is commonly fierce objection to any evaluation of software engineers within such discussions. Numerous excuses are given for any proposal, while no alternatives are offered. It seems software engineering is conceived as some impalpable art that cannot be evaluated, even by fellow artists. And this extends to strategies and criteria for interviewing prospective engineers. All commonly used methods, such as whiterboarding code exercises, are castigated as inadequate to measure the intangible brilliance and abilities of engineers.
I just think we all have some natural aversion to being evaluated for fear of failure. Therefore our institutions, both public and private, need to be designed to counteract that tendency.
I agree with everything you've written here, but you've focused, perhaps, on the wrong anecdotal examples in the NY Subway expansion. The choice to tunnel instead of cut-and-cover wasn't the main reason for cost-overruns. Nor was opting for the double-wide station model. These things certainly added costs, but compare them to similar projects in other countries and see that you could have all the things and still avoid the cost-disease in American-style transit.
Case-in-point: the MASSIVE, MULTIPLE-tunnel-digging project that's unfolding to expand several lines of the Stockholm metro as we speak. (One right next to my home and I don't notice it at all). You've already mentioned, I believe, how Swedish rail is roughly 1/10th per km the cost of American rail construction. It's a similar story for the very difficult/expensive work of digging tunnels DEEP underground in the archipelago that is Stockholm, an old city with many of the same issues as NYC, plus far less tolerance for disruptive surface noise. Also, all of these expansion projects have included stations and auxiliary offices/maintenance facilities that are generally larger and nicer than the grubby and cramped NYC Subway. So, how can Sweden extend three separate, really nice and spacious subway lines via tunneling in its biggest city under budget and ahead of schedule?
Clearly there are other reasons that are more salient than the decision to tunnel or not. Chief among them, seemingly, are all the things you've written about previously: entrenched NIMBYism, weaponizing of environmental reviews, multiple-juristiction governance, lack of economies of scale in one-off projects (Sweden is constantly rolling out these projects in modular fashion, leaving an existing production base and experienced labor force), and a budgetary process that is really short-termist and volatile, with constantly shifting priorities and budgets.
Maximizing ridership is what *should* happen, but two thoughts on what *is* happening:
1. The main cause of low ridership routes is not ‘planners not wanting to be yelled at in a public meeting,’ though of course that does happen. It’s that agencies have formal policies, approved by their governing bodies – which in most places means elected officials – setting out minimums for service coverage and frequency. So the planners have to first cover their entire service area at this minimum level, and only then go back with whatever resources remain and add more service in high-ridership corridors.
2. There’s a real status quo bias in the federal Title VI analysis that any agency has to do before implementing a major service change. Even if the existing network is sub-optimal from a total ridership perspective, reallocating those resources without causing a disparate impact on a particular group of riders can be tough.
Many sectors could be improved through focusing on simple, utilitarian goals. Take personal injury law.
People who are injured in accidents that are not their fault need medical care and replacement of lost income. Yet people need these same things regardless of who caused the accident. Any person with bad enough morbidities needs these things! The tort system spends a lot of money trying to distinguish between deserving and undeserving victims. Yet the results are completely arbitrary and don’t provide much security. Get hit by a UPS truck and you will get a significant payout, though you’ll only get 60 cents on the dollar after you pay your lawyer and experts. Get paralyzed by a driver with minimum coverage and you’ll only get $25k even if you lost $3M in future wages. Get screwed by a congenital condition, and you are left to the tender mercies of the welfare state.
The project of distinguishing “deserving” from undeserving victims does have some benefits. It encourages safe driving, discourages corporations from hazardous activities, etc. However, it doesn’t do these things very well. If the tort system were a highly effective way of incentivizing safety, we wouldn’t need DUI prosecutions or speeding tickets or even drivers licenses, an insurance mandate would be sufficient and the market could operate. Lawyers don’t often question the system because we are trained that victims should be made whole and the system is lucrative for established lawyers. My point is that intuitive goals like “make victims whole” are often a) not served very well by the status quo and b) subordinated to the interests of insiders.
>>The carless Americans who also don’t live near good transit are a very marginalized group<<
This one thousand percent.
And the situation brings to mind one of the worst takes of my own political tribe, namely the one that villfires Uber and Lyft.
They're not perfect, but Uber/Lyft are a godsend for folks who, for whatever reason, don't own cars. I remember talking to an Uber driver a few years back in Silicon Valley who informed me (I'd guess truthfully, but who knows?) that a lot of folks in that region were eschewing car ownership, because a car-free life featuring a lot of Uber fares saved money on net once the costs of car ownership were eliminated. Also, as an expat who is contemplating a return to the States, the existence of these services does rather expand the footprint of plausible destinations (how are normal people buying cars these days in the USA? Talk about sticker shock!).
>>But slow, expensive projects don’t become unbearably slow and unbearably expensive thanks to one big bad call <<
I'm a frequent reader of well-known California-based pundit Kevin Drum (I believe he and Matt are acquainted). IIRC, Drum is more or less of the opinion that there was indeed "one big bad call" that doomed California's HSR from the get-go: the project, because of horrendous engineering challenges flowing from the region's geography (something about a mountain pass in the path of the only route between LA and SF that really makes economic sense), was simply never very viable.
I don't possess the engineering chops (to say the least) to assess this claim, but in full disclosure I'm a big fan of bullet trains, and every time I ride one (vastly more comfortable and fun than flying; and they're miraculously conducive to napping) I think to myself "My fellow Americans don't know what they're missing."
But even the HSR-obsessed Chinese mostly don't build them where the engineering problems are excessive, I reckon.
One really easy way to improve ridership on transit would be to go back to not letting people smoke crack on the train. I was on the train at about 6:45 this morning and, in addition to the piles of garbage, sticky floors and seats, and abundant suicide-related ads, there were people smoking, doing drugs, and drinking on the train. For the first time in ages, transit police actually did hop on and made them get off. I would guess that they got off and then got on the next train. This isn’t rocket science. We don’t need to hire legions of consultants or create a whole management layer at the transit agency to create a pilot program for crack smoking strategy innovation. We had this figured out ten years ago, and decided we didn’t want to know how to deal with it anymore.
I partially buy it. and speaking here in my personal capacity as an advocate for MD's Purple Line. I'd say right now transit advocates are largely incentivized to make arguments based on the multifaceted political benefits of a projects, e.g. jobs including during construction, small business benefits, etc., with ridership as perhaps the most important factor but typically one part of a three part argument.
Meanwhile, the transit skeptical, if not looking to just kill projects, are looking to save money in absolute terms, not in per rider. For example, a cut to the MD Purple Line cost engineering was going from 6 minute headways to 7.5. Thankfully this is highly fixable, but would definitely fail the ridership test. Similarly, I think Matt's cafe car example is incomplete. A ridership oriented answer would be thinking about how to more cost effectively deliver something riders clearly want. Maybe that's a cart or better setting up to sell bento box equivalents in the stations. NARP and I wouldn't love it, but that seems to be the solution that other countries have settled on.
But I feel like the organizational and permitting part to this is probably an equal part of the problem. Like a huge cost driver on the Purple Line was a lawsuit exploiting the NEPA process and an activist transit skeptical judge forcing a year of delays (also probably underbidding re: stormwater management, CSX being uncompromising, and allegedly slow property acquisition on MD's part). On top of that, the P3 model meant to save costs didn't do enough to resolve conflicts and resulted in an aggressively bidding builder exploiting a poorly chosen clause to walk off the job.
Gov. Hogan made some dubious choices, but he would argue that the P3 would have helped the cost/rider. I think emphasizing cost/rider would greatly help on the incentives front but it will still be a highly painful process with any number of mistakes made in pursuing it. I think we need a handbook of what effective implementing approaches look like from comparative studies. The Eno center does some interesting work here https://www.enotrans.org/article/eno-releases-major-report-on-u-s-transit-costs-and-project-delivery/
The only difference between non-existent public transport and crap public transport is that the latter will cost a lot of money to deliver and serve only slightly more people than the former. The sigmoid function usefully describes this phenomenon, and is used to model expected public transport usage at a given level of service. You need high quality provision to deliver decent ridership.
The tragedy of bad public transit projects is that it ruins the idea that the government can provide good quality public services. When you see the government sending vast resources to stupid, failed projects like California HSR you realise this. I mean, I read they want to try running battery operated trains pending construction of new electrical infrastructure, which is laughable (https://calmatters.org/politics/2022/05/california-high-speed-rail-standoff/).
From a somewhat more internal perspective, I will say the concern is less about getting yelled at in meetings (though that sucks quite a bit) than it is about trying to complete the project and actually build it, rather than finish a proposal and then being stuck in litigation for a million years (or even just briefly, litigation sucks for the people involved, even the lawyers, usually).
If you've been working on a project long enough to get to litigation, you probably believe in it a fair amount and want it to happen, even if only so the last X months/years of your life haven't been wasted. And annoying the neighbors is an absolutely certain way to ensure that instead of building a project, you spend the next X months/years of your life justifying to a lawyer and then to judges why you annoyed the neighbors.
Amtrak runs the long-distance trains because they are congressionally mandated to. They run the regional corridors (except NEC) in partnership with states. The money from the infrastructure bill for short-distance, high-ridership routes *must* be applied for by states, not Amtrak.
Amtrak does a lot of things wrong, but this is not really a decision they're making.
The more I read MY, the more I settle in to an opinion that an active and well funded state is a good idea in countries where public agencies are well run, like in many countries in Europe and some in Asia.
But at the same time, I get more and more skeptical of giving money to US government agencies. I give MY huge credit for actually calling out and grappling with the problems, but I am not at all persuaded that these issues are fixable.
Is there an example where some set of reforms have turned around a wasteful and unfocused US government effort into an effective one? The only examples I can think of involve the military or wartime mobilization efforts, which don't seem to provide much guidance.
I work for the Federal Government (health) and this resonates with what I see. Government staff are buried under dozens of process requirements (like Small business requirements) and leadership measures success as no one yelled at us, not “did we objectively improve things.”
> And importantly, telling agencies to try to hit a half-dozen different goals … doesn’t ensure that all those things will be done simultaneously. What it ensures is that, with no principled way to evaluate projects or say no to things, costs explode.
The cynic in me wants to believe that agency leaders would like to avoid a simple and objective measure of performance. If they are solely evaluated by ridership/dollar-spent then they could fail to improve on that metric and be charged with bad decision making. In contrast, if they claim to be balancing numerous unquantifiable objectives, then it’s much harder to criticize their performance. For example, if we point out the low ridership volume then they can point to their progress in expanding access and creating good paying jobs.
And I think there is a general tendency for all leaders and workers to resist objective evaluation. For example, I've seen this criticism applied to the anti-testing movement in education where it is alleged that poor teachers simply want to evade detection.
I’ve also seen this aversion to performance evaluation within the tech industry, particularly within discussions on the tech forum Hacker News. While there are many poor approaches to measuring software developer performance (e.g., lines of code count), there still need to be some objective evaluation. Quality measures can include annual peer review using a rubric of dimensions.
Yet there is commonly fierce objection to any evaluation of software engineers within such discussions. Numerous excuses are given for any proposal, while no alternatives are offered. It seems software engineering is conceived as some impalpable art that cannot be evaluated, even by fellow artists. And this extends to strategies and criteria for interviewing prospective engineers. All commonly used methods, such as whiterboarding code exercises, are castigated as inadequate to measure the intangible brilliance and abilities of engineers.
I just think we all have some natural aversion to being evaluated for fear of failure. Therefore our institutions, both public and private, need to be designed to counteract that tendency.
Reducing antisocial behavior on public transit would probably increase ridership.
I agree with everything you've written here, but you've focused, perhaps, on the wrong anecdotal examples in the NY Subway expansion. The choice to tunnel instead of cut-and-cover wasn't the main reason for cost-overruns. Nor was opting for the double-wide station model. These things certainly added costs, but compare them to similar projects in other countries and see that you could have all the things and still avoid the cost-disease in American-style transit.
Case-in-point: the MASSIVE, MULTIPLE-tunnel-digging project that's unfolding to expand several lines of the Stockholm metro as we speak. (One right next to my home and I don't notice it at all). You've already mentioned, I believe, how Swedish rail is roughly 1/10th per km the cost of American rail construction. It's a similar story for the very difficult/expensive work of digging tunnels DEEP underground in the archipelago that is Stockholm, an old city with many of the same issues as NYC, plus far less tolerance for disruptive surface noise. Also, all of these expansion projects have included stations and auxiliary offices/maintenance facilities that are generally larger and nicer than the grubby and cramped NYC Subway. So, how can Sweden extend three separate, really nice and spacious subway lines via tunneling in its biggest city under budget and ahead of schedule?
Clearly there are other reasons that are more salient than the decision to tunnel or not. Chief among them, seemingly, are all the things you've written about previously: entrenched NIMBYism, weaponizing of environmental reviews, multiple-juristiction governance, lack of economies of scale in one-off projects (Sweden is constantly rolling out these projects in modular fashion, leaving an existing production base and experienced labor force), and a budgetary process that is really short-termist and volatile, with constantly shifting priorities and budgets.
Maximizing ridership is what *should* happen, but two thoughts on what *is* happening:
1. The main cause of low ridership routes is not ‘planners not wanting to be yelled at in a public meeting,’ though of course that does happen. It’s that agencies have formal policies, approved by their governing bodies – which in most places means elected officials – setting out minimums for service coverage and frequency. So the planners have to first cover their entire service area at this minimum level, and only then go back with whatever resources remain and add more service in high-ridership corridors.
2. There’s a real status quo bias in the federal Title VI analysis that any agency has to do before implementing a major service change. Even if the existing network is sub-optimal from a total ridership perspective, reallocating those resources without causing a disparate impact on a particular group of riders can be tough.
Many sectors could be improved through focusing on simple, utilitarian goals. Take personal injury law.
People who are injured in accidents that are not their fault need medical care and replacement of lost income. Yet people need these same things regardless of who caused the accident. Any person with bad enough morbidities needs these things! The tort system spends a lot of money trying to distinguish between deserving and undeserving victims. Yet the results are completely arbitrary and don’t provide much security. Get hit by a UPS truck and you will get a significant payout, though you’ll only get 60 cents on the dollar after you pay your lawyer and experts. Get paralyzed by a driver with minimum coverage and you’ll only get $25k even if you lost $3M in future wages. Get screwed by a congenital condition, and you are left to the tender mercies of the welfare state.
The project of distinguishing “deserving” from undeserving victims does have some benefits. It encourages safe driving, discourages corporations from hazardous activities, etc. However, it doesn’t do these things very well. If the tort system were a highly effective way of incentivizing safety, we wouldn’t need DUI prosecutions or speeding tickets or even drivers licenses, an insurance mandate would be sufficient and the market could operate. Lawyers don’t often question the system because we are trained that victims should be made whole and the system is lucrative for established lawyers. My point is that intuitive goals like “make victims whole” are often a) not served very well by the status quo and b) subordinated to the interests of insiders.
>>The carless Americans who also don’t live near good transit are a very marginalized group<<
This one thousand percent.
And the situation brings to mind one of the worst takes of my own political tribe, namely the one that villfires Uber and Lyft.
They're not perfect, but Uber/Lyft are a godsend for folks who, for whatever reason, don't own cars. I remember talking to an Uber driver a few years back in Silicon Valley who informed me (I'd guess truthfully, but who knows?) that a lot of folks in that region were eschewing car ownership, because a car-free life featuring a lot of Uber fares saved money on net once the costs of car ownership were eliminated. Also, as an expat who is contemplating a return to the States, the existence of these services does rather expand the footprint of plausible destinations (how are normal people buying cars these days in the USA? Talk about sticker shock!).
Great points, but really bad for my mood.
It was tremendously frustrating to see Washington Metro spend $45 million on free bus service when frequency and route availably are more important.
>>But slow, expensive projects don’t become unbearably slow and unbearably expensive thanks to one big bad call <<
I'm a frequent reader of well-known California-based pundit Kevin Drum (I believe he and Matt are acquainted). IIRC, Drum is more or less of the opinion that there was indeed "one big bad call" that doomed California's HSR from the get-go: the project, because of horrendous engineering challenges flowing from the region's geography (something about a mountain pass in the path of the only route between LA and SF that really makes economic sense), was simply never very viable.
I don't possess the engineering chops (to say the least) to assess this claim, but in full disclosure I'm a big fan of bullet trains, and every time I ride one (vastly more comfortable and fun than flying; and they're miraculously conducive to napping) I think to myself "My fellow Americans don't know what they're missing."
But even the HSR-obsessed Chinese mostly don't build them where the engineering problems are excessive, I reckon.
One really easy way to improve ridership on transit would be to go back to not letting people smoke crack on the train. I was on the train at about 6:45 this morning and, in addition to the piles of garbage, sticky floors and seats, and abundant suicide-related ads, there were people smoking, doing drugs, and drinking on the train. For the first time in ages, transit police actually did hop on and made them get off. I would guess that they got off and then got on the next train. This isn’t rocket science. We don’t need to hire legions of consultants or create a whole management layer at the transit agency to create a pilot program for crack smoking strategy innovation. We had this figured out ten years ago, and decided we didn’t want to know how to deal with it anymore.
I partially buy it. and speaking here in my personal capacity as an advocate for MD's Purple Line. I'd say right now transit advocates are largely incentivized to make arguments based on the multifaceted political benefits of a projects, e.g. jobs including during construction, small business benefits, etc., with ridership as perhaps the most important factor but typically one part of a three part argument.
Meanwhile, the transit skeptical, if not looking to just kill projects, are looking to save money in absolute terms, not in per rider. For example, a cut to the MD Purple Line cost engineering was going from 6 minute headways to 7.5. Thankfully this is highly fixable, but would definitely fail the ridership test. Similarly, I think Matt's cafe car example is incomplete. A ridership oriented answer would be thinking about how to more cost effectively deliver something riders clearly want. Maybe that's a cart or better setting up to sell bento box equivalents in the stations. NARP and I wouldn't love it, but that seems to be the solution that other countries have settled on.
But I feel like the organizational and permitting part to this is probably an equal part of the problem. Like a huge cost driver on the Purple Line was a lawsuit exploiting the NEPA process and an activist transit skeptical judge forcing a year of delays (also probably underbidding re: stormwater management, CSX being uncompromising, and allegedly slow property acquisition on MD's part). On top of that, the P3 model meant to save costs didn't do enough to resolve conflicts and resulted in an aggressively bidding builder exploiting a poorly chosen clause to walk off the job.
Gov. Hogan made some dubious choices, but he would argue that the P3 would have helped the cost/rider. I think emphasizing cost/rider would greatly help on the incentives front but it will still be a highly painful process with any number of mistakes made in pursuing it. I think we need a handbook of what effective implementing approaches look like from comparative studies. The Eno center does some interesting work here https://www.enotrans.org/article/eno-releases-major-report-on-u-s-transit-costs-and-project-delivery/
The only difference between non-existent public transport and crap public transport is that the latter will cost a lot of money to deliver and serve only slightly more people than the former. The sigmoid function usefully describes this phenomenon, and is used to model expected public transport usage at a given level of service. You need high quality provision to deliver decent ridership.
The tragedy of bad public transit projects is that it ruins the idea that the government can provide good quality public services. When you see the government sending vast resources to stupid, failed projects like California HSR you realise this. I mean, I read they want to try running battery operated trains pending construction of new electrical infrastructure, which is laughable (https://calmatters.org/politics/2022/05/california-high-speed-rail-standoff/).
From a somewhat more internal perspective, I will say the concern is less about getting yelled at in meetings (though that sucks quite a bit) than it is about trying to complete the project and actually build it, rather than finish a proposal and then being stuck in litigation for a million years (or even just briefly, litigation sucks for the people involved, even the lawyers, usually).
If you've been working on a project long enough to get to litigation, you probably believe in it a fair amount and want it to happen, even if only so the last X months/years of your life haven't been wasted. And annoying the neighbors is an absolutely certain way to ensure that instead of building a project, you spend the next X months/years of your life justifying to a lawyer and then to judges why you annoyed the neighbors.
Matt, "Amtrak staff" doesn't really make the decisions about route selection. The map that you posted is made up of routes suggested by states, not Amtrak: https://www.trains.com/trn/news-reviews/news-wire/amtraks-gardner-asserts-mechanical-ranks-not-thinned-by-furloughs/
Amtrak runs the long-distance trains because they are congressionally mandated to. They run the regional corridors (except NEC) in partnership with states. The money from the infrastructure bill for short-distance, high-ridership routes *must* be applied for by states, not Amtrak.
Amtrak does a lot of things wrong, but this is not really a decision they're making.
The more I read MY, the more I settle in to an opinion that an active and well funded state is a good idea in countries where public agencies are well run, like in many countries in Europe and some in Asia.
But at the same time, I get more and more skeptical of giving money to US government agencies. I give MY huge credit for actually calling out and grappling with the problems, but I am not at all persuaded that these issues are fixable.
Is there an example where some set of reforms have turned around a wasteful and unfocused US government effort into an effective one? The only examples I can think of involve the military or wartime mobilization efforts, which don't seem to provide much guidance.