Comments on Instead of projecting traffic, reduce our dependence on it - Greater Greater Washington

Comment by Charles Marohn

Mon, 30 Jul 2012 14:39:46 EDT

I'm going to be addressing all these questions/comments in depth on the Strong Towns Blog this week, starting today:

http://www.strongtowns.org/journal/2012/7/30/build-a-better-model.html

GGW, as always, is welcome to link to or reprint any of these posts.

I appreciate the conversation here. The feedback is helpful and is giving me good guidance on how to better present these insights. Thank you GGW readers.

Comment by xmal

Sun, 29 Jul 2012 22:09:17 EDT

@AWalkerInTheCity
> But not investing in infrastructure is itself a choice, with massive opportunity cost.

Fair point!

Comment by AWalkerInTheCity

Sun, 29 Jul 2012 20:37:15 EDT

"Finally, another point of contention: AWalkerInTheCity, a ‘recovery’ is generally returning to the prior average rate of growth, not exceeding that rate until catching up to the long term trend."

no, you are incorrect. Recoveries from recessions usually do involve above trend line growth.

"What we as a society need to do is remove the government incentives for building this settlement pattern and tie in the true cost of living in low-density areas with hierarchical road systems by making the residents pay the true cost of all the infrastructure they need to have the suburban lifestyle. Unfortunately, this would and will be a long and painful process resulting in eventually only high-income residents having the modern suburban lifestyle (with all the amenities) and low-income residents either hopefully getting into cities or else becoming endemically poor. This is the natural dichotomy of areas that don’t have subsidies supporting them "

most of us here would like an end to those subsidies (and to limits on urban density) I do NOT think that would leave only the wealthy living a suburban lifestyle. Perhaps we disagree on the magnitude of those subsidies, or the extent of preference on the part of many middle class for that lifestyle.

"since, by nature, black swans are impossible to predict (he also refers to them as the “highly improbable”), it is impossible to know the point at which using a bell curve becomes worthless."

in which case using them may well be worthwhile.

"Therefore, we shouldn’t bet on massive infrastructure projects based on inaccurate projections that, when all tallied together, leave our society indebted for smaller and smaller returns on the investments."

But not investing in infrastructure is itself a choice, with massive opportunity cost.

Comment by Skyler Yost

Sun, 29 Jul 2012 20:14:24 EDT

@AWalkerInTheCity and @Jasper:

It seems that both of you are taken with statistics and dismiss Chuck's approach based on your love of the bell curve. Jasper, you counter xmal by pointing to the discovery of the Higgs-Boson using advanced statistics as evidence that we could theoretically predict the winner of Wimbledon in 2025 or 2035, which isn’t even related to xmal’s point that the player that would win Wimbeldon in 13 or 23 years is likely in the earliest stages of his career or else hasn’t even begun playing yet, and projections of what young individuals will become world class players (let alone what world class player will win a knockout tournament) are almost entirely inaccurate and essentially futile. AWalkerInTheCity, you counter Chuck by pointing to using projections of gas prices and macroecon conditions 20 years from now in order to predict demand in 20 years as how projections can be done better (!) even though, as you know, those initial projections of gas prices and macroecon conditions will not pan out themselves, exacerbating how poorly the traffic projection will perform. Both of you seem to be scoffing at Chuck’s thesis that “We don't need better projections, we need a system that is robust to modeling error. We need a system of growth and development where we don't need to project correctly in order to succeed” based on your underlying belief in bell curves without considering the basis of Chuck’s whole argument.

Chuck is looking at the problem from the perspective of Nassim Taleb’s Black Swan. Taleb explains that bell curves are useful only between cataclysmic ‘black swan’ events. Since, by nature, black swans are impossible to predict (he also refers to them as the “highly improbable”), it is impossible to know the point at which using a bell curve becomes worthless. Therefore, we shouldn’t bet on massive infrastructure projects based on inaccurate projections that, when all tallied together, leave our society indebted for smaller and smaller returns on the investments.

Personally, I think projections are still useful for betting on short-term growth that will pay for itself quickly and easily cover its maintenance costs (even after factoring in the length of its lifecycle and the associated replacement costs), but only for smaller-scale projects. Leave the large-scale stuff to either supporting current land-use patterns or satisfying projections through highly flexible infrastructure like putting a single track of rail in place to satisfy future needs because it can accommodate a-little-more-than-the-current-land-use all the way up to heavy-future-demand-projections. For another example, building an ‘urban extension’ satisfies growth projections while minimizing the amount of infrastructure needed to be maintained when compared to the same population increase in a suburban pattern, while also reducing individual costs like transportation and increasing the likelihood of small, independent businesses due to a larger amount of viable commercial space and smaller store sizes. This is the kind of robustness that helps protect cities and towns from suffering when Wall Street torpedoes the macro-economy.

On another note, yes there is still a huge amount of demand for single family homes in suburban areas. Currently. What we as a society need to do is remove the government incentives for building this settlement pattern and tie in the true cost of living in low-density areas with hierarchical road systems by making the residents pay the true cost of all the infrastructure they need to have the suburban lifestyle. Unfortunately, this would and will be a long and painful process resulting in eventually only high-income residents having the modern suburban lifestyle (with all the amenities) and low-income residents either hopefully getting into cities or else becoming endemically poor. This is the natural dichotomy of areas that don’t have subsidies supporting them and don't have the density and/or number of residents required to provide per person cost savers like public transportation, municipal sewer, and the like. It is also one of the major reasons that cities have driven human development throughout recorded history.

Finally, another point of contention: AWalkerInTheCity, a ‘recovery’ is generally returning to the prior average rate of growth, not exceeding that rate until catching up to the long term trend.

Comment by Froggie

Fri, 27 Jul 2012 09:38:44 EDT

goldfish earlier wrote: "In my travels to Minnesota, I am awed by how much they overbuild roads,"

As a native Minnesotan, I had to laugh at this. Within the Twin Cities metro, the only road that was "overbuilt" was I-94 north of downtown Minneapolis. Most of the rest are congested, some heavily and outside of normal rush hour.

I could perhaps see where some of the rural 4-lane highways seemed "overbuilt". For better or worse, there comes a traffic volume point where it is too much for a 2-lane road, but is light enough to where a 4-lane highway appears overbuilt.

Comment by AWalkerInTheCity

Thu, 26 Jul 2012 16:45:59 EDT

"@AWalkerInTheCity, it actually says that they expect not just a recovery, but a catch up.
And there is no long term trend, unless you consider a couple decades "long term".

Catch up after a recession IS recovery - thats not uncommon business cycle behavior.

Its not unreasonable to think that the trend line for econ growth hasnt changed - but you easily model a scenario of long term (ie beyond one business cycle, please dont quibble) growth slowdown.

"You yourself are citing shifts in usage in your argument, the desire to go "car lite" and use rail for all but the last few miles. I agree, but traffic projections don't consider human "

note, lots of folks wont go carlite, and lots of freight will continue to go by truck origing to destination. do not misstate what I said.

"desires, emotions or reactions to outside forces like recession, high gas prices or sustained unemployment."

Any good forecast should include a forecast of gas prices. And it could include macroecon conditions as well - (in another life, when I did ROI analysis for a RR, we always modeled a periodic dip in demand)

You can of course model changes in desires - those you can predict, or those you want to do sensitivity analysis on. The staff report arl Co did on Col Pike street car basically did that - they explicitly stated that the mode preference FTA allows for rail is too low based on recent experience, and implied their recommendation was heavily based on that. Now can you convince ALL policy makers that there will be further shifts away from autocentrism - I doubt it - but thats a problem in making your case - not a limit of models per se. If you can make that case, you can build that into your model.

Comment by Charles Marohn

Thu, 26 Jul 2012 16:17:35 EDT

@AWalkerInTheCity, it actually says that they expect not just a recovery, but a catch up.

And there is no long term trend, unless you consider a couple decades "long term".

You yourself are citing shifts in usage in your argument, the desire to go "car lite" and use rail for all but the last few miles. I agree, but traffic projections don't consider human desires, emotions or reactions to outside forces like recession, high gas prices or sustained unemployment.

We should not be spending tens a billions a year on approaches we know are consistently wrong. We have to think differently, not just accept that it is the best we can do.

Comment by AWalkerInTheCity

Thu, 26 Jul 2012 15:57:11 EDT

"The Ports re-evaluated their growth projections after the recession and made adjustments. The revised projections show slower growth in Port activity in the next decade, but the 2035 total TEU estimate remains the same: 42.7 Million. In other words, the Ports anticipate to handle 42.7 Million TEUs by 2035, even if growth is slower in the next few years.

In other words, it might slow down for a while, but we're confident it will go back up ultimately because our projections say it will."

thats not what that says - it says they expect recovery from recession by 2035 back to the long term trend (which is not unreasonable) and they ignored the panama canal issue.

Comment by AWalkerInTheCity

Thu, 26 Jul 2012 15:53:52 EDT

"The premise that most seem to have here is that our world always has to be built around the automobile."

the Pew survey that many of cite that showed high demand for urban walkable, still showed over half of Americans wanting traditional low density auto centric suburbs. And many people who want to be less auto reliant, want to be car lite, and still use cars to some degree. leaving aside trucks (where rail is important, but truck almost always provides the last mile, and truck dominates some commodities) So yes, I think motor vehicles will play a very important part of transportation for the forseeable future. which is quite compatible with wanting to see them play a smaller role.

Comment by AWalkerInTheCity

Thu, 26 Jul 2012 15:48:14 EDT

"@jasper, For the most part, Washington DC has a productive land use pattern. I don't believe the metro is being run to a greenfield in anticipation of new growth. Any extension is in support of current land use patterns. "

it still requires projections.

"That is not how traffic modeling is generally being used, as I cited in my post."

Much of it is - for example the beltway hot lanes, the I95 hotlanes.

Comment by Charles Marohn

Thu, 26 Jul 2012 15:44:04 EDT

@jasper, For the most part, Washington DC has a productive land use pattern. I don't believe the metro is being run to a greenfield in anticipation of new growth. Any extension is in support of current land use patterns.

That is not how traffic modeling is generally being used, as I cited in my post.

Comment by Jasper

Thu, 26 Jul 2012 15:35:21 EDT

@ Charles:The premise that most seem to have here is that our world always has to be built around the automobile. If you free yourself from that constraint and instead imagine the world you would like to live in, projections become much less important.

Euhm, no. Are you suggesting the Silver Line is being built without any projections of its use? Are you suggesting, the new Rush minus service from WMATA was not implemented after a model showed that more people would gain from it than would loose?

Everything can be modeled, except certainty.

Comment by Charles Marohn

Thu, 26 Jul 2012 14:34:21 EDT

xmal is the smartest person in the room, IMO

The premise that most seem to have here is that our world always has to be built around the automobile. If you free yourself from that constraint and instead imagine the world you would like to live in, projections become much less important.

Does that mean get rid of the automobile? No, it simply means that enormous infrastructure projects are the eventual outcome of productive land use patterns, not something we do in the hopes of creating growth.

Comment by David C

Thu, 26 Jul 2012 12:04:11 EDT

xmal, I think you misunderstood my comment. I was not accusing the author of being bad at math. I was replying to your comment that there are no "better projections". That is clearly a fallacy. Whether or not they are good enough to draw any conclusions is another question, but clearly having biased people create the projections is not the best path.

Comment by Jasper

Thu, 26 Jul 2012 11:43:27 EDT

@ xmal:The author is not trying to say "Math is hard, let's go shopping" but rather that there are things for which math cannot be used reliably and therefore should not be used because it leads to bad decisions.

Nothing personal, but your lack of distinction between math and modelling proves my earlier point about numerical illiteracy.

The challenge is not predicting the winner of Wimbledon in 2015, but in 2025 or 2035. There is no model that you could develop that would be able to do this.

Funny you say that. The Higgs-Boson was just found using rather advanced statistics. With your attitude, it would not have.

You are right that models can not be used to predict the future. However, models can be used to describe likely events in the future. That's the whole point of modeling.

Modeling does not give certainty, but likelihood (or probability). The difference between certainty and likelihood is equal to the unknown influences. So by talking in likelihoods, you are in fact taking unknown things into account.

If you're about to make an investment that will last 50 or 100 years, you have to have some idea of what's gonna happen, because otherwise, you can not make a decision. And that's where modeling comes in.

If you want to build a large infrastructural project, that benefits a lot of people, you can better do so in New York than in Moore, MT. Why? Because there are more people in NYC than in Moore. Can you know for sure that NYC is not gonna collapse with everybody moving to Moore? No. But it sure is unlikely.

@ Greenbelt:Are we letting dubious projections and doomsday scenarios force us to keep widening roads forever, at great expense in livability, economic vitality etc.?

I think everybody here agrees that current models suck. The question then is what to do. Get rid of models and just take a swing, or invest in better models? I think the latter. Taking a swing has not proven to be a good model to move forward.

Comment by AWalkerInTheCity

Thu, 26 Jul 2012 11:05:20 EDT

"So it's far better to think about the future that you WANT and build towards that rather than chasing after what you think you NEED, as shown in projections."

if what yuo WANT is a walkable community, thats easy. If what yuo WANT is more economic development in general, or reduction in traveler delay time, with no a priori modal or urban form preference, I dont think you can make a case without projections. I think this approach leans very heavily towards folks with explicit preference for particular urban forms and modes, and against those who are more neutural on those things.

In a different context, some people want a future of high GDP and minimal regulation - if the climate cannot be predicted, then they ask why sacrifice what they want for something that may not come to be. The answer of course, is that while we can't predict the future with exactitude, we can do so well enough, including with sensitivity analysis for the known unknowns, that we can know that not making the sacrifice is a very bad bet.

"Or vice-versa, people are so excited by the idea of the new comfortable stadium and they overwhelm it, requiring you to expand further. Sound familiar?"

but it does NOT require further expansion. You can still choose not to expand further. If you built a facility and it got utilized, I dont see how thats an argument for not building the facility. Thats the extreme version of the induced demand argument - the assumption that the goal of capacity expansion is purely to create an uncongested facility, and that the induced trips have zero incremental value. I think that approach leads to the silent majority fearing some kinds of urbanist arguments - folks LIKE being able to use their cars to have a wider choice of shopping, or to make it easier to live in a SFH on a quarter acre. We should change policies that subsidize those choices - but to discount ANY value to those choices, seems to me to isolate us from real stakeholders conversations.

Comment by Greenbelt

Thu, 26 Jul 2012 10:58:04 EDT

Straight line projections are fine for exogenous factors that are, in fact, growing in straight lines.

I think the author's points are two:

1. Exogenous factors, like traffic in an area, don't necessarily grow according to straight line projections -- it's much more complicated than that usually, and

2. Traffic isn't exogenous. When we build car infrastructure to meet straight line projections, we encourage more driving (unless the lanes are tolled). The only significant marginal costs of most non-tolled driving are gas costs and congestion costs (cost in time delayed). Holding gas costs aside, when we build more "free" lanes for driving, we encourage more driving until a new equilibrium of congestion costs is achieved. This why new lanes tend to fill up to roughly the same congestion level and why it's so hard to out-build the traffic.

The author is essentially asking whether it makes sense to continue at this rat race. Shouldn't we be making decisions about how to plan transportation based on other considerations too? Are we letting dubious projections and doomsday scenarios force us to keep widening roads forever, at great expense in livability, economic vitality etc.?

Comment by Geof Gee

Thu, 26 Jul 2012 10:35:09 EDT

@Alex who wrote

"So, why not explicitly account for the uncertainty and lack of precision?"

One certainly can try and there are techniques for doing so. Assuming we all agree that modeling the expectation is hard, modeling higher-order moments is much more difficult. So depending on the problem, one should seriously consider what types of uncertainty -- consider a situation where your estimates are based on a data set with sampling error ... that error might be realistically modeled whereas the error associated with the model might not -- one is trying to capture. I'm no expert on the types of problems that Charles discusses here, but I have a feeling that modeling the uncertainty of the model is a wildly difficult problem. Note that if one is doing some sort of simulation, one can always check the variation due to parameter assumptions and so on.

Broadly, I would not expect naively applied models to have a high degree of precision over the lifespan of these long term infrastructure investments. If we're building a road, railway, waterworks, whatever that will last 20, 30, 50 or more years, you should understand that things can be wildly different from the projections. Nonetheless, number crunching can give an otherwise fuzzy discussion a much better context.

Comment by xmal

Thu, 26 Jul 2012 10:25:59 EDT

@David C and others thinking that the author is bad at math:

The author is not trying to say "Math is hard, let's go shopping" but rather that there are things for which math cannot be used reliably and therefore should not be used because it leads to bad decisions.

The challenge is not predicting the winner of Wimbledon in 2015, but in 2025 or 2035. There is no model that you could develop that would be able to do this.

Or to use a more germane example, predicting Wimbledon attendance in 2025 or 2035 so you can, say, rebuild the stadium. The best you can do is look at the past and draw a line or a curve that continues the trend, and adjust for things like expected population in 2025 or 2035, which themselves are projections just as tenuous.

But then you have a curious problem: you're locked in to that projection; you now have a bigger stadium to fill, so you're going to do what you need to fill those seats. Or vice-versa, people are so excited by the idea of the new comfortable stadium and they overwhelm it, requiring you to expand further. Sound familiar?

So the conclusion is that there are things that one cannot predict, "unknown unknowns" such as changes in fashion, technology, world events, induced demand, path dependence, network effects that cannot be modeled. And these things are crucial, overwhelming what you CAN model (think of the invention of the automobile: just years prior, people were modeling how New York would drown in horse manure). So it's far better to think about the future that you WANT and build towards that rather than chasing after what you think you NEED, as shown in projections.

Comment by AWalkerInTheCity

Thu, 26 Jul 2012 09:40:09 EDT

@jasper

I agree with both your points.

Comment by Jasper

Thu, 26 Jul 2012 09:21:41 EDT

@ AWalkerIntheCity:I think educating decision makers about the uncertainty factors in projections, standard deviations of models, sensitivity analysis to changes in assumptions is great

But it's rather a large task. This goes back to basic education and numerical literacy.

Try to explain to anyone that getting the same amount for 50% of the price is a much better deal than getting 50% more for the original price. And then try to explain why the latter deal is 1/6th more expensive than the first. Then explain them that that 1/6th equals 16%.

And this is simple fractions math, which is a few steps down from understanding statistics, errors bars, and uncertainties, the last being a pretty toxic word in political discourse. 'You mean to say you are not 100% certain this is going to happen?'

It was more "projections are uncertain, so lets build projects that are less dependent on projections"

Yeah. That's how I read it: Projections suck, so let's get rid of them, and build an urban utopia without projections. Not the way to go. Total lack of accountability.

Comment by AWalkerInTheCity

Thu, 26 Jul 2012 09:07:08 EDT

"The core of what this post is talking about isn't really about the projections - or the content of those projections. It's about the decision making process and the weight assigned to those projections without taking into account their accuracy or precision."

I think educating decision makers about the uncertainty factors in projections, standard deviations of models, sensitivity analysis to changes in assumptions is great - thats not what I read the OP as being about. It was more "projections are uncertain, so lets build projects that are less dependent on projections"

Comment by Jasper

Wed, 25 Jul 2012 20:57:17 EDT

@ Jack Love: I think we basically agree. I just wanted to make the point that the word 'manipulation' does not have a negative connotation in the world of modelling. You are right that models yield what you put in. Put in junk, get out junk. Put in good stuff, get back good stuff. The former is a lot easier than the latter.

@ Alex B:The core of what this post is talking about isn't really about the projections - or the content of those projections. It's about the decision making process and the weight assigned to those projections without taking into account their accuracy or precision.

I understand that perfectly. This post says: Projections don't work, so let's get rid of them. I disagree, because that will make traffic planning only more political. The solution to this is to get better data.

I like David C's idea:they'll get a bonus based on how well they project outcomes

Problem is that even that can be fudged by setting enormous error bars. And worse, it would allow politicians to defer partial payment till way after their term.

Comment by Alex B.

Wed, 25 Jul 2012 18:34:30 EDT

No, not irrelevant at all. Better projections would be a welcome development. However, I don't know that better, more accurate projections would solve the core issues that Charles is talking about here, since those issues are really matters of process and how we make decisions.

Comment by David C

Wed, 25 Jul 2012 18:10:22 EDT

but let's not lose sight of the core issue here.

Thank you for taking the time to make sure that I know that what I've talked about is irrelevant to you.

Comment by Alex B.

Wed, 25 Jul 2012 17:18:12 EDT

That's how you get the best projections.

The core of what this post is talking about isn't really about the projections - or the content of those projections. It's about the decision making process and the weight assigned to those projections without taking into account their accuracy or precision.

The real issue is decision making and the process we use. Since these kinds of projections are enshrined into the process (often by law), the projections matter a great deal - but let's not lose sight of the core issue here.

Comment by David C

Wed, 25 Jul 2012 16:55:00 EDT

there is no such thing as "better projections" when dealing with complex systems.

Really? If you asked me to guess the winner of the 2015 Men's Wimbledon, I would probably guess Agassi, because I know nothing about tennis and he's the only player I can think of. An expert would probably have a better projection. Hiring knowledgeable people and incentivizing the right things (accuracy) will probably give you the best possible projection on average.

Jasper and selxic hit it on the head. The way to get better projections is to choose independent projectors and tie their pay to how good the projection is. Create error bars and let them know that they'll get a bonus based on how well they project outcomes. That's how you get the best projections.

Comment by Jack Love

Wed, 25 Jul 2012 15:03:51 EDT

Jasper, I think we're saying two slightly different things regarding the point at which the manipulation is introduced. The distinction is important in that a manipulated model, to which I referred, carries the cachet of authority for having passed scientific rigor (ha!), but carries the risk of garbage-in/garbage-out.

I take your meaning to be that the model itself provides the tools by which the real world is reduced and described, and then manipulated through policy, which is a more transparent process. (tempting to say "ha!" again here)

I don't necessarily think models alone got us to our poor capacity planning, but they sure enabled some addictive behavior.

Comment by Jasper

Wed, 25 Jul 2012 14:31:09 EDT

@ Jack Love:Models are pretty much like surveys: you can manipulate the inputs and factors to the point where you can produce any output.

The point of a model is to reduce a complex system to a point that can be understood and therefore manipulated. Manipulation is not bad in this case, it is the point.

The necessary simplification is based on certain assumptions, which are crucial to the validity of the model. Those assumptions have to be valid, as well as explicitly stated. The assumptions also limit the validity of a model, which is fine, as long as those limitations are understood.

It is the conditions described in the last paragraph, where most models get in trouble. Not just traffic models, but also scientific models.

Comment by Bertie

Wed, 25 Jul 2012 14:27:52 EDT

We don't need better projections, we need a system that is robust to modeling error.

So what is this system that is robust to modeling error? How can we make rational decisions about large-scale, long-term transportation infrastructure without projections?

Comment by Jack Love

Wed, 25 Jul 2012 14:08:07 EDT

Models are pretty much like surveys: you can manipulate the inputs and factors to the point where you can produce any output. I am not convinced they are at all reliable, as we constantly seem to be catching up with real-world experience. By 2050, in other words, we will be able to accurately model the 2020 transportation network.

If that is analogous to the demand side, then what about the supply side? A modern approach to dealing with capacity is to provide a second hand that manipulates facility usage via user fees. See the HOT lanes on the Virginia Beltway, for example: the projection is that there will be 'x' VMTs in the HOT lanes, but that can be capped with an adjustable toll.

Comment by Alex B.

Wed, 25 Jul 2012 14:01:20 EDT

Honest people know how to interpret these simulations and, at least implicitly, account for the uncertainty associated with estimates.

So, why not explicitly account for the uncertainty and lack of precision?

Comment by YIMBY

Wed, 25 Jul 2012 13:58:27 EDT

"The problem is that many traffic projections are not honest studies. They are studies by stake-holders, who basically set the outcome from the start."

The above could not be any more valid... it's unfortunate that sprawl and traffic drove the US economy from 1945 til 2008, but the damage that was done will take some time to undo...

Comment by Geof Gee

Wed, 25 Jul 2012 13:53:09 EDT

This has little to do with model quality, IMO. Honest people know how to interpret these simulations and, at least implicitly, account for the uncertainty associated with estimates. Jasper above describes the issue succinctly ...

"The problem is that many traffic projections are not honest studies. They are studies by stake-holders, who basically set the outcome from the start."

Comment by selxic

Wed, 25 Jul 2012 12:57:48 EDT

Thanks Tyler. Ironically, I think that actually helped me out a lot.

Comment by AWalkerInTheCity

Wed, 25 Jul 2012 12:54:45 EDT

Oboe,

thats fine - it may well be that there are better alternatives to our current modal mix of investments (I wont even start on the debate about the extent of and meaning of induced trips). But lets realisze that that - the costs and benefits of different modes - is what we are talking about - not about the accuracy of projections in general

Comment by oboe

Wed, 25 Jul 2012 12:41:39 EDT

Considering 1 through 4, ignore the traffic that does in fact exist and pretend it will eventually go away.

I don't think anyone's denying that traffic exists--it's just that more and more people are coming to the understanding that it's not going to get any better. So enjoy your commute today; it's the best it's ever going to be.

The question is, how many hundreds of billions of dollars a year are we going to throw down the toilet in order to fail to alleviate the problem? And what else could we be spending the money on?

Comment by goldfish

Wed, 25 Jul 2012 12:31:21 EDT

In my travels to Minnesota, I am awed by how much they overbuild roads, and how easy and manageable the traffic is there. Compared to DC, it is suburban paradise. So how seriously can I take the complaints about traffic projections?

Comment by ceefer66

Wed, 25 Jul 2012 12:23:13 EDT

So, we should:

1. Ignore traffic projections - because they're "done and skewed and by those with a vested interest".

2. Dismiss traffic projections as "innacurate" or, better yet, "irrelevant" (see 1).

3. Do nothing about the traffic. After all, the numbers mean nothing and any traffic that does in fact exist is
the result of "human behavior" caused by "induced demand". Which, of course, can be always changed by simply not building roads.

4. Spend all available resource on transit. After all, we can ALWAYS believe transit use projections, can't we?

5 Considering 1 through 4, ignore the traffic that does in fact exist and pretend it will eventually go away.

Sounds like a plan!

Comment by AWalkerInTheCity

Wed, 25 Jul 2012 12:17:19 EDT

"We need a strong towns approach which, if you stop and look, is a lot like the pre-automobile approach that served us well for thousands of years. I'm not saying we get rid of the automobile, but when we build our entire environment around its propagation, we are slave to our own hubris and lack of understanding."

pre automobile in the USA intercity freight and passengers moved by rail. Private railroads (and govts offering subsidies) certainly projected traffic, even if not using formal models. Often that worked. Sometimes (like the panic of 1837) it did not.

Prior to that canals were built. Again using informal projections. And bridges. And aquaducts.

Infrastructure is lumpy, and its long lived. Unless you are going back to unpaves foot paths, and neolithic approaches to production and services, there is no way around that.

Comment by AWalkerInTheCity

Wed, 25 Jul 2012 12:12:43 EDT

"we need a system that is robust to modeling error. We need a system of growth and development where we don't need to project correctly in order to succeed. We need a system where we build incrementally in a replicable and evolving pattern, one where each fractal evolves continually and naturally over time. We need a system where we're not required to place huge bets on the future, oversizing infrastructure in service of projections, but instead can invest in high return endeavors where the likelihood of success is great."

this seems unrealistic to me. Infrastructure has natural economies of scale. Even if you were to never build another interstate highway, what about bridges, tunnels, and heavy rail metro lines? You can't build them in tiny increments, and you can't justify them without an estimate of usage in the out years.

he is right that it would be great to build in ways that made projections unnecessary. I doubt that you can do that - or at least to do that, you would have to impose limits (no heavy rail, no bridges and tunnels, etc) that would be more costly the mistaken projections

note, making us LESS dependent on projections is something else

Comment by TransitDork

Wed, 25 Jul 2012 12:10:04 EDT

I don't have the title of the textbook with me today (rode my bike instead of Metro to work) but I think it it "Urban Transportation Planning". In it they detail the methodologies for traffic studies and then basically admit that these studies have a very dismal success rate and are very, very open to being manipulated to providing the answers those who commission the study want to see.

I never trusted the studies much before this but now I have good reason to not trust them. But, that the studies must be done is progress and over time it seems they are slowly (SLOWLY!) becoming better. Rome wasn't built in day...

Comment by Tyler

Wed, 25 Jul 2012 12:04:52 EDT

Traffic models are akin to the ubiquitous minimum parking requirements set in zoning ordinances. As Shoup puts it, "traffic engineers use very precise numbers to report extremely uncertain estimates, and urban planners [and public officials]rely on these precise numbers to make bad decisions."

Comment by Tyler

Wed, 25 Jul 2012 11:59:11 EDT

@selxic It's Strong Towns, not Smart Towns.

Comment by xmal

Wed, 25 Jul 2012 11:58:07 EDT

So far the comments are missing the point, which is there is no such thing as "better projections" when dealing with complex systems. Just like it was impossible to predict the flood of retirees in the author's example, so it will be impossible to predict, say, the impact of driverless cars (will they make traffic flow faster, or will people liberated from the wheel drive more, negating the savings?). Repeat this for any of a million other factors and tell me where that leaves the "projections".

The solution is making our transportation network resilient so that even if in 10 years the projections prove wrong, the whole thing doesn't collapse.

Comment by selxic

Wed, 25 Jul 2012 11:39:09 EDT

Would someone explain Smart Towns to me? It reads like creating sprawl in areas historically inhabited instead of focusing on current urban areas.

Comment by selxic

Wed, 25 Jul 2012 11:24:18 EDT

Begins post. Hit refresh. Jasper beats me to it.

A lot of the initial premise of this piece seems to be off. Plus, as an engineer, how often does the initial idea meet what is actually built?

Comment by Jasper

Wed, 25 Jul 2012 11:18:05 EDT

Instead of projecting traffic, reduce our dependence on it

Or get better projections.

The problem is that many traffic projections are not honest studies. They are studies by stake-holders, who basically set the outcome from the start.