It's no mystery to figure out what happened, if the original story is accurate (no reason to believe it's not)... the developers reportedly ignored the plan and built decorated sprawl. On the one hand, I'd like to see Andrés go back and straighten things out. On the other hand, where's the confidence the developers will follow the plan this time when they didn't before? Building a real town requires more commitment and intelligence than building sprawl... That's why there is so much sprawl and so few real new towns. Hopefully the meltdown will change that equation.

Certainly the developers developed sprawl, but that's a non-explanation. Why would they do that?

Christopher Alexander wrote an entire book about the reasons plans are ignored, The Oregon Experiment.

Most likely they developed sprawl (I haven't seen it) because it's their default setting. It takes an intelligent AND committed developer to change their status quo.

In Avalon Park the developer subsidized new businesses with profits from housing so that his development would look like a mixed used neighborhood. That may be committed but it's certainly not intelligent.

The developers in Cornell decided that it wasn't intelligent to stay committed to a plan that made no sense. They switched back to sprawl because that's what is most natural in that system.

What about the transit service? If it was supposed to be a transit-centered community, did the transit agency support the vision by providing service to the community center? If the transit never materialized, you can hardly blame the residents for defaulting back to their cars.

Let's not "blame" anyone. New Urbanism is not a pact that you enter in. It's supposed to be a better way to grow a town. If it doesn't work, then it doesn't work. It's not the developers who were not committed enough, or the residents who refused to take the bus. It's just the plan that didn't work. Let's learn from it and try something else.

New Urbanists want to make another Savannah. If you look at the original plan of Savannah, it was just a bunch of identical rectangular lots with the same tiny shacks on them. It grew into Savannah by feeding back on itself. (All historic cities were built that way.) The problem with the plans the New Urbanists have been making is that they try to get everything right at once - assembling the right mix of uses, the right transportation, and the right density - when in fact the problem may be that there's too much density and not enough room to grow over time. They are stitching up a Frankenstein monster, and trying to persuade a developer to shock it to life with a high-voltage dose of speculative capital, instead of trying to grow a new town.

"If it doesn’t work, then it doesn’t work" ~And if you don't follow the plan, then you'll never know whether the plan would have worked.
With that being said, you raise an intriguing point about the way we attempt to build finished cities today, rather than growing a town. I personally thing that this was a necessary intermediate step to get us started back to real town-building. If you allow links in comments, then go to http://www.originalgreen.org/OG/Resources.html and click on The Sky Method. I'd love to know what you think.

Sorry, I hit Submit too soon... I meant to add that the Sky Method is my best take on using New Urbanist tools such as the Transect to provide a framework for growing a town. It was developed for a corner of DPZ's Sky in the Florida Panhandle that was so remote from the rest of the development that it couldn't be a normal neighborhood... so for today, it could only be conceived as large lots. But when you take away the limitation of time, then it's able to grow into a full-fledged neighborhood.

I saw the PDF for the sky method floating around the Pro-Urb list a few weeks ago. It's fairly close to my vision of an emergent urbanism, although I don't understand why you complicate things with transect zoning. An emergent process will create a transect all by itself, as seen from the image of Tultepec in my previous update.

What we've learned from these New Urbanist experiments is that only time can make a normal neighborhood.

Did people move there wanting to live in a village? If that were the case there is no lack of villages looking for people to come back. In fact, people moved there looking for a suburb within the greater metropolitan area. Nothing wrong with that, if people are looking for a suburb might as well make it a better one.

Can a plan ensure that businesses will set up shop there? Can it control economic forces that historically have taken decades to generate the current conditions?

New Urbanism is an intelligent design, creationist view of communities, that their genesis does not require evolution. In the same way, new urbanists come up with transects, which achieve through infallible directives what land economics would achieve in a much longer period, a density gradient from a designated centre outward. This is something that has always puzzled me. Why would you want to re-create in the lab the feature that is primarily responsible for sprawl, the density gradient from the centre to the expanding edge?

Rather than growing sprawl in the test tube using artificial land economics, why not grow medieval walled towns, using artificial brigands and wolves?

Judging from Cornell's layout, it seems that Cornell's failure is due to its location of retail spaces inside the neighbourhood, away from the distributor roads that form the perimeter of the development. This is a common failure of almost all suburbs, villages and neighbourhoods developed around the same time period - it fails because retail cannot simply survive on the sole catchment of its walkable district in a low-density environment; it relies on passing traffic.

Planning documents now reflect this, and support retail development on the distributor roads (which can be easily laid out in a new urbanist manner - nil street setbacks, on-street parking with rear service access, etc).

Distributor roads are also excellent and direct carriers of transit routes, and retail development provides a good buffer to residential uses behind, as well as significantly increasing opportunities for a permeable road layout. And it provides a fantastic treatment of what would otherwise be a traffic sewer, which is how Cornell's perimeter roads are treated.

Some retail has succeeded in the space for which it was planned, so it's not true that retail cannot survive on simple catchment of walkable sectors. (And isn't walking the point of a New Urbanist development?) There just isn't as much of it as was planned. Planning retail for some other location will just replace one error for a different error.

In a traditional city there is no strict demarcation of "retail" and "non-retail" sectors. Commerce sets up where it makes the most sense for it to be. So a few very local services will build in between houses, while large shops will build along major integrator roads. There will be other uses randomly sandwiched between shops, with the frequency inversely proportional to propensity for commerce, and particular categories of commerce, of the space.

The mistake of the plan is defining a land use in the first place and building out the blocks too densely so that there remains no space to fill in late-stage additions to the neighborhood. Traditional cities don't grow in solid blocks but in gradients of increasing density. That's the source of their complexity.

Why do we plan villages/neighborhoods which historically evolved on their own? Because the conditions under which they evolved no longer exist. The 'feedback' is simply not there to create an element such as a town square because the demand for a local market and gathering place is not as high as it needs to be to make that happen. Elements which used to evolve as necessities are no longer necessary, and while they may be used and even desired, the grassroots support needed to create them from scratch simply does not exist. It is only after those elements already exist, that modern uses are applied to them and the community utilizes them. In other words, people will use what is there, but are not (in this modern climate) disposed to going to the lengths necessary to have elements created from nothing. Current cultural patterns in North America support the exploitation of the most easily accessible elements, not the more time and labour intensive ones. This is one of the reasons why, as a general rule, people will build new subdivisions from inferior materials rather than restore older residences - it's simply easier; easier to find funding for, easier to customize, easier in terms of time consumption. If we build villages with features people want, and they are easily utilized, people will use them. If we build subdivisions with empty lots to be filled in later, people will think "it sure would be nice to do something with that lot" and then just grumble a bit when it's simply used to build more houses on - which is yet another reason to support the transect zones of the sky method (in my opinion). Obviously, there are no universals, and there are some locations in North America with dynamic, involved populations which do provide feedback and actively restructure their environment - I am generalizing here to capture the overall climate and issue.

If more houses are built in a neighborhood it may be that there is a real need for more houses. In Cornell the problem wasn't only that there were too many houses but that shops that were put there by the developer to respect the plan failed. I saw the same thing happen in mixed-used neighborhoods in France. Why would anyone risk building another shop?

What we've found since modernist planning processes were widely adopted in the mid-20 century was that any space that somehow escaped the control of the authorities would become a spontaneous town, so there is nothing inevitable about sprawl. It is a choice by the authorities. It is a choice to give developers and property owners certain incentives. It is a choice whose consequences are not very well understood, and this is why this website exists, because it's possible to make a different choice if we understand the consequences.

What's not possible is to attempt to replicate spontaneous cities without changing any of the planning process. Cornell is only one failure amongst many others in that experiment.

So what do we do... throw up our hands and just do sprawl, or try to re-learn how to build real incremental places again? This "building the finished town" thing has been a necessary re-learning step, IMO. And also the only thing we could do, given municipalities' current insistence that developers show the finished product in excruciating detail before getting an approval to turn a spade of dirt. It is only after 30 years of the New Urbanism that we've built a big enough body of work that we're even BEGINNING to get a tiny fraction of municipalities to even listen to proposals for building incrementally. I believe that you and I both, Mathieu, would like to see this be the default condition. But the planning disasters of post-WWII have so screwed things up that any form of unpredictability has been rendered strictly illegal. Try to do otherwise in most places and they'll send the constables out and shut you down if you're a developer. So building the completed town has been a necessary intermediate step to regain trust.

Predictability is the greatest difficulty because predictability has been the characteristic of a "scientific" process since Descartes. That is the revolutionary insight of Wolfram, that by insisting on predictability when doing scientific research scientists are only selecting problems to be solved based on predictability, and in so doing are avoiding a scientific study of the enormous body of physical phenomena that isn't predictable.

Why shouldn't authorities insist on predictability? That is what has been considered scientific for centuries. Predictability means progress. It certainly isn't by creating failed TNDs that they will be convinced to give up on scientific methods. However if we can redefine what is scientific and show them another way to understand the processes of urban development, they may well be willing to switch to other processes in order to jump on board a whole new, forward-thinking scientific paradigm.

But predictability means doing precisely what you don't want to do, which is to design the finished town from the beginning.
Unless, of course, an area can "escape the control of the authorities," as you've noted. Those are the only rare instances where anything, TND or sprawl, can avoid the fate of having to design the completed town before a spade of earth is turned.

There are ways to describe and explain processes that are unpredictable, so it is possible to design a town without it being "finished." From there it is possible to show how any random configuration of that town out of the infinity of possible configurations meets the standards of the authorities. Each spade of earth can be turned with full confidence that things will work out.

A TND is sprawl. It's enlightened sprawl, but it has the same unsustainable features. It's not going to be persuasive enough to roll back planning to the past. The new science doesn't roll back planning, it drives it forward. That's why it will work.