Can government learn how to ‘fail fast’?

Can government experiment?

Here’s a common pattern: An agency spends a few years working on a project that could improve residents’ lives. Procurement delays and construction issues take extra time. The project opens, there’s controversy and people call for changes or say the project was a waste. Public employees get the message. Next time, they spend even more time designing the project.

Are we on a cycle in which everything government does happens slower and slower?

Let’s look at a field in which things aren’t slowing but speeding up: software. The people making Web sites and apps are innovating at a frenetic pace. In recent years, a new management philosophy called “lean startups” has taken hold. One of the basic principles, according to guru Eric Ries, is to build something quickly, measure how well it works and improve it. The faster you run through the “build-measure-learn” cycle, the better.

Now, government can’t do this with, say, a bridge. You can’t just throw up a bridge, see if it works, then put up a better bridge. It has to work right the first time. But there are a lot of ways government programs could work like lean startups. Capital Bikeshare, for instance, has stations that are easy to move. All it takes is appropriate space and enough sun to power the solar panels. If one location isn’t working well, officials could try another.

We can do this with streets as well. After a pedestrian was killed at the intersection of 15th and W streets and Florida and New Hampshire avenues NW, the District Department of Transportation (DDOT) installed temporary curbs and bollards to prevent drivers from taking turns too quickly. It has a nice long-term design for the intersection in the works, but for the years in between, we still get a safer, if slightly ugly, intersection.

Of course, experimentation isn’t new to many areas of government. The police try out crime-fighting techniques and look at whether the crime rate goes down. Educators test better ways to teach. Even capital projects, such as in transportation, have long involved pilot programs designed to test an innovation.

Classically, however, these pilots could last for years or even drag on indefinitely; the District has numerous parking “pilots” that never really conclude and generate no useful data about whether they are helping or hurting. Lean startups, in contrast, demand trying something fast, measuring its success quickly, and learning and revising fast.

Often the obstacles to quick government innovation are political and social. Bureaucratic processes, and expectations from the public and elected officials, slow things down.

That’s because trying things means, inevitably, that sometimes they will fail. And nobody likes to see money wasted, least of all public money. Residents, and reporters, understandably recoil at any sign that funds might have been squandered on a project that turned out to not be a good idea. But it’s actually more fiscally responsible to try a lot of cheap ideas than to spend years on staff and consultants to pick a single idea, which still might not work out; what’s most important is to make the attempt quickly and cheaply (“fail fast,” as the lingo in the tech world goes). That’s one reason small tech startups, which might only have the capital needed to operate for a short time, are the biggest adherents of this approach.

Unfortunately, scandals around misspent public money have led governments in the opposite direction, adding layer upon layer of review to every effort. That can stop real fraud such as an official steering a contract to a family member, but it can also keep honest and worthy projects bottled up.

Residents and elected officials also have to be willing to accept experimentation. In Glover Park, DDOT installed “traffic-calming” measures to make the area safer for pedestrians. The project isn’t yet complete, but already some residents who preferred being able to drive quickly through the neighborhood are calling for its removal, and D.C. Council members want answers.

Maybe the project will be a net positive, maybe a net negative, but DDOT hasn’t finished setting up the roads, so it’s too early to know. The agency should finish it up as quickly as possible and measure the results as quickly as possible, and we should give them some time to do so.

Unfortunately, this project took so many years between the study and construction that many residents don’t remember the planning process. Experimentation requires getting from idea to test with less delay.

Take Arlington’s bus-and-streetcar “super stops,” which have drawn criticism for being overly expensive. Building something that will get a lot of wear and tear and has to last a long time is more expensive than most people think. Still, there are legitimate questions about whether the super stop is really as much of a step forward as it could be.

One Arlington resident pointed out that Arlington considers this a “prototype” and suggested that, rather than building a stainless-steel prototype, in the future it could set up a bare-bones one and let residents try it out. As it happens, this resident, David Panarelli, is a designer for LivingSocial, a technology startup, so he’s familiar with this way of thinking. A prototype is a good idea, but residents, reporters and political opponents of those in power need to acknowledge that it’s experimental and react accordingly.

Too often, if government tries to move quickly, officials just come in for criticism for not planning enough, spending money on temporary measures or doing things that look slipshod. If they move slowly, they’re attacked for taking too long, spending too much or “gold-plating” everything.

In management, there’s a principle called “Fast/Cheap/Good”: You can only get two. Without realizing it, the way people react to government programs pushes them away from fast and implicitly away from cheap as well. The lessons from startups are clear: We’re better off with a lot more fast and cheap, and by experimenting, we can get something that’s even better than slow. Can we allow our public servants to make this happen?

David Alpert is editor of the blogs Greater Greater Washington and Greater Greater Education.

 
Read what others are saying