Last Friday, Jeffrey Zients, the fellow who’s been tasked with overseeing the fixes to HealthCare.gov, gave a telephone briefing to reporters. Regretfully, I was at a conference and unable to sit in on that call. But I’ve corresponded with people who were, and it’s been thoroughly written up in the Washington Post and elsewhere. The gist: They have a list of what needs to be fixed, and Zients says that by Nov. 30, he expects the website to be working for the vast majority of users. Jonathan Chait remarks:
The administration is obviously putting its neck on the line here. If it fails to hit the deadline, all political hell will break loose. (There is a little wiggle room, as the promise applies to "the vast majority of users.") Therefore, presumably, the administration is extremely confident it can hit this deadline. On the other hand, it was also extremely confident it could have the site working reasonably well by October 1. So Obama apparently believes not only that his administration can fix the technical problem, but also that it has already fixed the managerial problem that caused it to underestimate its technical problem.
There's also the third possibility: The administration has learned that a large meteor will destroy the world on or before November 30, and wants to live out its remaining time on the planet in relative peace, rather than dodging "are we there yet?" questions about the website every day. So basically the possibilities are:
1) They know what they're doing.
2) They have fooled themselves into thinking they know what they're doing, but don't.
That list is amusing, but it's not exhaustive; there is also a fourth possibility, which is that they are buying themselves time. Think of it this way: If the website doesn’t work by Nov. 30, all hell will break loose. How much more hellacious will it get if the president’s tech guy promised that the website would work and it didn’t? Not all that much more hellacious, really.
On the other hand, if they admit that it might not work on Nov. 30, the national discussion will immediately start focusing on delaying all or part of the law. This way, they buy themselves a month’s respite from demands for a delay, and they also make that delay much harder to implement, because more people will have bought through the state exchanges or had their existing insurance policies canceled by their insurers. Sure, that might seriously screw up the insurance markets -- but as far as I can tell, the administration has decided, possibly correctly, that any backward movement, however small, puts the whole law in grave danger. So damn the torpedoes, full speed ahead.
And perhaps they have already bought themselves a little bit of that time they need. Sunday has been, for the last three weeks, the day when we get big features from major newspapers explaining how bad things are with the website. But this Sunday, the Wall Street Journal and the New York Times instead had features exploring two related questions: How did things get so bad and why didn’t the White House seem to know about it?
Both paint a picture of a near-total disconnect between the political, policy and technology-implementation staffs. The White House frequently weighed in on items such as the user interface of the website or various policy details, but it didn’t appear much interested in the information-technology portion. The policy staff sat in their office in Bethesda drawing up a logical, incredibly detailed road map for the rollout, which they then mailed off to the IT folks. The IT folks subcontracted out whatever orders they were given, but they apparently did not do a good job of communicating their growing sense of dread to the rest of the people involved in this massive project. And so the White House merrily went around giving presentations -- consisting of some screenshots enlivened with interactive features -- that implied the new website would work just like Travelocity for health insurance.
The yawning gap between what the IT people knew and what everyone else seems to have realized is staggering. Now, I’ve worked on some projects in which the business units seemed to have some sort of selective deafness that only materialized when we tried to tell them that they couldn’t have their magic fairy computer system that did everything they could imagine, only better, in the three months they wanted it to take, or for the paltry sum that they were willing to spend. And I learned the hard way not to assume that the business units, or even the chief information officer, had heard and understood what you said. That is how I became gifted in the art of writing CYA memos when I was directed to do the unwise or the impossible. So I do have some sympathy for the IT folks.
But I’ve never seen a gap this complete -- one in which the entire IT organization seems to have been panicking about the impossibility of their task, and then the inevitable failures, while the folks issuing the orders were blithely issuing last-minute change orders and telling everyone they could find how swell this was all going to be. Usually, when things are going this wrong, you do more than casually mention it; you sit the folks on the business side down and explain that unless the project is pushed back, it’s going to be an unmitigated disaster.
Someone, somewhere pretty high up in the food chain must have understood that the website could not be ready on time and would not do what the political folks were promising. That person failed to communicate not only the unlikelihood of making the Oct. 1 start date, but also the extent of the problems that would follow if they opened anyway. The policy and political staffs genuinely seem to have been expecting a few weird errors and a little bit of downtime, not a computer system that failed at pretty much every step.
There’s been some talk in the blogosphere recently about whether this calls the liberal project into question. I think that’s going a bit far, but I do think it calls into question the competency of President Barack Obama’s technocrats. This is the first big program they had to implement from scratch: They had to actually make something happen, unlike the stimulus, which mostly consisted of writing checks, usually through programs that already existed. And so far, it’s a complete failure. The only parts that are even arguably working are some minor tweaks to coverage rules and the Medicaid expansion, which consists of, yes, writing checks through a program that already existed.
You occasionally hear conservatives sneer about Obama’s socialist tendencies, but if anything, I think it is the lessons of free-market economics that have undone him. Obama surrounded himself with policy folks who thought a great deal about incentives; the whole idea of the insurance-market reforms is that if you line up all the insurance incentives just so, the market will, like a well-tuned machine, produce exactly the results we want. They were open to discussing which levers might be incorrectly aligned. But it doesn’t seem to have occurred to them that there is much more to markets, and policy, than just twiddling levers. It’s no accident that their greatest policy successes came from Treasury banking programs for which twiddling the financial levers really is pretty much all you need to do . . . or that some of their greatest failures came when they tried to transfer those programs into industries such as cars, where success or failure relies on a lot more than simple math.
Don’t get me wrong: I believe that the White House understood that implementation matters -- intellectually. But I do not think they understood it viscerally, because the president prefers to surround himself with talkers and designers rather than doers. That’s not a slam on talkers and designers; you will notice what profession I myself chose. But before I became a journalist, I worked in private firms for a decade, which gave me a lot of training in all the ways that something that looks good in the design lab can go wrong when you try to roll it out across a big company.
I spent last weekend at a seminar on spontaneous order led by Vernon Smith and Bart Wilson of Chapman University. The way it worked was this: We read philosophers on such topics as virtue and property. Then we read papers on economics experiments in which Smith and Wilson and others had teased out some aspect of how and when various spontaneous orders emerge. All of the experiments were either game-theory constructs or little virtual markets where we had to buy and sell.
The seminar attendees -- mostly academics, plus a few journalists -- were all extremely smart people. We all knew a lot about markets. And we had one advantage that the original experiment subjects had not had: We had read the papers and knew what the optimal strategy was.
We were terrible.
In fact, we often performed worse than the original experiment subjects. Reading the papers didn’t help at all. In fact, we often outsmarted ourselves. For example, in one experiment, I was a retailer, and my supplier had priced slightly too high, so we were losing business. But because my supplier didn’t understand that I had fixed costs for the store, they thought I should lower my price even further, but because we couldn’t talk, they had no way to tell me that. Eventually, they tried to send me a signal by raising prices, which just looked crazy and greedy to us. So we raised ours even more, to signal that they were acting crazy. So they raised theirs even further . . . and anyway, we didn’t do so well.
Knowing the blueprint and theory behind the experiment didn’t help. In fact, we outsmarted ourselves with game theory. If we hadn’t been so involved in playing elaborate signaling games, we probably would have paid more attention to how much money we were losing.
That was extreme, but in every game, one thing was true: Knowing the theory did almost no good in helping you establish a market and a set of rules for interacting with two or six or 12 other people. You could develop this elaborate theory of how people ought to act, but that wouldn’t prevent them from surprising the hell out of you by thinking and acting in entirely different ways from your simple mental model. Nor would it prevent you from missing some key element -- like the store cost -- that made your neat plan go wildly awry.
And if you talk to the folks who actually implement programs, whether they be software developers or entrepreneurs or government bureaucrats, they’ll tell you the same thing: Clean models last about three seconds in the face of reality. Not because you’re stupid, but because the world is too complicated for you to foresee every contingency.
Looking back on all the conversations I had with the health-care law’s supporters in and out of government in 2009, I do not think they had that gut-level understanding that the universe was under no obligation to comply with their elegant system. They thought that if they just got all the incentives right, everything else would fall into place. They didn’t reckon with the nuts-and-bolts details -- like computer systems -- that could easily destroy their carefully laid plans. They figured that if you wanted it badly enough, and had enough money, and put your smartest and most competent and most dedicated people in charge, it would have to work. All the focus was on the levers they would be pulling; no one was brought in to think about the engine of the machine until much too late.
And perhaps Jeff Zients will fix the website, and everything will work smoothly, and Obamacare will turn out to be a smashing success. On the other hand, you do have to wonder what other nuts-and-bolts problems might have gotten insufficient attention. Because if the websites do start working, that’s when we’ll find out if there are other missing pieces in this vast machine.