President Obama today addressed the problems with the rollout of the Affordable Care Act's website... kind of. He tried refocusing media attention from just the website's problems, and he promised that the problems would be fixed. But he didn't actually admit what those problems were. Today, to be fair and evenhanded, we're going to take a look at the president's statement extolling the good things his signature healthcare reform law has accomplished, as well as the bad things he couldn't quite bring himself to talk about in detail.
President Obama spoke for roughly a half an hour on the subject, by the official transcript. He fully admitted that the federal website has been plagued with problems. He tried to refocus attention on the benefits of "Obamacare" itself (rather than the problems), saying repeatedly "the Affordable Care Act is not just a website." It is likely that little of this will get through to the American public, because the media will focus on Obama's admission of website problems, perhaps best exemplified by a letter he read at the end of his address, which contained the line: "Yes, the website really stank for the first week."
Obama's approach today was to admit the website problems, promise they would be fixed as fast as humanly possibly, but also to make some valid points that the media has been ignoring for years. The first was to remind people of the benefits of Obamacare which have nothing to do with the website -- the end of "pre-existing conditions," better prescription drug coverage for seniors, young adults allowed to stay on their parents' insurance, preventative care, and all the rest. The president also focused on the savings many Americans would be seeing -- the "really stank" letter was from a self-employed guy who would see a reduction in his monthly insurance costs from $1,600 to $692 -- a savings of over ten thousand dollars per year. This is why Obama kept trying to separate the affordability of the insurance itself from the website problems, obviously. Obama pointed out that there is a phone number to call to sign up on the exchanges if the website is having problems, and the website was reportedly changed over the weekend to prominently feature this call number on the site itself, which may help. Obama promised that people who have tried to apply through the website and gotten stuck will be personally contacted to help complete their application process -- which is also an important message for a frustrated public to hear right now.
Obama's most important point -- again, one that will likely be ignored by the media -- was to compare the new exchanges with what they were designed to replace:
[W]e've actually made the overall process of buying insurance through the marketplace a lot smoother and easier than the old way of buying insurance on your own. Part of the challenge here is that a lot of people may not remember what it's like to buy insurance the traditional way. The way we've set it up, there are no more absurdly long application forms. There's no medical history questionnaire that goes on for pages and pages. There's no more getting denied because you've had a pre-existing condition. Instead of contacting a bunch of different insurers one at a time, ... there's one single place you can go shop and compare plans that have to compete for your business. There's one single phone number you can call for help.
Those were the informative things Obama's address today covered. But what was missing from his speech were any specifics on why the website stank so badly. The first part of fixing any problem is identifying what is going wrong, after all. Obama did pledge that some talented people were now joining in the effort to fix the website:
[W]e are doing everything we can possibly do to get the websites working better, faster, sooner. We've got people working overtime, 24/7, to boost capacity and address the problems. Experts from some of America's top private-sector tech companies -- who, by the way, have seen things like this happen before -- they want it to work. They're reaching out. They're offering to send help. We've had some of the best I.T. talent in the entire country join the team. And we're well into a "tech surge" to fix the problem. And we are confident that we will get all the problems fixed.
All fine and good. But, once again, the problems were not really specified, beyond the obvious problem of the extremely high interest from the public which has been slowing the site down with traffic overloads. Other problems were only hinted at -- the closest Obama came to admitting them was "the number of people who have visited the site has been overwhelming, which has aggravated some of these underlying problems." The underlying problems were never named, though.
Technologically speaking, the "sheer volume" problem should have been the easiest to fix. Any web-oriented software system like the exchange website is designed to run on multiple servers. Servers are the machines you communicate with when you read any website. And all but the smallest websites rely on what is called "load balancing" so they can plug in two, five, a dozen, or hundreds of servers to handle whatever demand comes in. But the software doesn't really need to change (beyond updating a few files) to add more and more servers to the mix. It's a hardware problem, not a software problem, to put it another way. Just buy more servers, and then hook them up and get them up and running. Problem solved!
But then there are problems which have nothing to do with the load. If the website's only problem was too much demand, it likely would have been fixed by now. With nothing but leaks and rumors to go on currently (since Obama obviously didn't want to get bogged down in describing software bugs today), we have no idea of how critical these problems are to getting the website up to speed. All we can do at this point is make a few educated guesses.
From what I've heard in the tech community, there may be a design flaw that is exacerbating the traffic problem. If the system's design requires too much data to be sent back and forth to the user's website by inelegant use of code or just bad design, then the individual user probably wouldn't be affected too badly (it would contribute to slow browser response, but it wouldn't freeze the browser). The problem is that the bloat in the bandwidth necessary for one single user is then multiplied by the total number of current users -- which could indeed bring the entire system to a crawl, if badly designed. If the software is sending ten times the amount of data back and forth per user than it really needs to, then the system as a whole will only be able to handle ten percent of the load it optimally could. This may be one of those "underlying problems" Obama spoke of, and a rather large one.
There was one report that the insurance companies themselves are getting "bad data." The insurance companies are at the end of the signup process -- the Obamacare exchange website is nothing more than an intermediary or "marketplace"-- and if you buy insurance through it, eventually your data is sent to the insurance company you have selected (at least, that's the way it is supposed to work). But the "bad data" seems to all fall into a couple of categories. The first is repeat applications from the same person (or family). This is likely caused by people clicking the final "submit" button over and over again, in the hopes of getting through on a slow and unresponsive website. This should be a fairly easy problem to solve, though, by running checks on the incoming data to look for duplicates, before the final submittal. The second category of bad data is incomplete submissions, which, again, could be solved by automatically double-checking the application before it is submitted. The third category is more troublesome -- some insurers reported the wrong data in the wrong field of the form. Children listed as multiple spouses, for instance. This could either be a serious bug in the software (swapping fields around at random in the database, somehow), or it could be what is politely called "cockpit error" or "user error" by the software industry (you don't want to know the impolite terms they use, trust me...). Software bugs can be fixed, but you can only do so much, programmatically, to fix a user typing the wrong thing in the wrong field (to put it bluntly).
Most of these programmatic problems should not exist, of course. Software is supposed to be designed to double-check for multiple entries in the database. With all the personal information entered, this shouldn't have been too hard to design in to the system. The fact these problems exist shows that there were major failures in either the design, the coding, or the testing of the product. They can be fixed, but doing so is harder than just adding a few dozen servers.
The biggest problem in the system as designed, though, is that to get a quote on the Obamacare website, you have to fully sign up. You can't just go and browse and see what things cost, without entering in your own personal data. There was one troubling rumor that this may have been a political problem rather than a software problem.
Comparing the Obamacare exchange to any other online retailer -- as just about everyone has done -- isn't really a fair comparison. How many websites need to know your estimated income for next year, after all? But this shouldn't excuse what is a serious design flaw. And the flaw may exist because so few Americans understood how the exchange actually is supposed to work, and someone on some government committee decided that the only way to fix this was to force everyone to pre-enter their data before shopping around.
The Obamacare exchange is supposed to do one thing: show you prices for available insurance plans. This is already a complicated thing, though, because there are four levels of coverage available (bronze, silver, gold, and platinum). So, already, the user is faced with four categories -- with multiple entries per category (for each insurer's plan under each category). That's complex enough to begin with.
But if users just jumped onto the site, entered what state they live in, and were presented with these numbers, then many of them would not be seeing their actual cost. They would be seeing the total price of each plan, not what they would have to pay. The prices would not take into account the subsidies. And the subsidies are based upon income.
To state the problem more clearly: if Joe Sixpack could go to the Obamacare exchange site and start browsing after entering just what state he lived in, he might see prices which are a lot higher than what he'll actually have to pay out of his own pocket. This could lead to a monstrous amount of public confusion, and much bad press over "sticker shock." The news would be full of stories about Joe Sixpack seeing monthly prices of $500 and up for his health insurance. But Joe never entered his income before seeing this number -- if he had, he would have seen that he was eligible for a $425 monthly subsidy.
So the Obama administration decided to make it foolproof -- Joe would not start by seeing the price as "$500 per month," he instead would see the correct out-of-pocket price of only "$75 per month." This seems reasonable, on the face of it, because it avoids a whopping amount of confusion and false media stories about outrageous prices.
There are two things seriously wrong with the way the Obama folks went about this, though, at least if initial reports are to be believed. The first is that it shouldn't have been that hard to force the user to enter only the absolute minimum of information before being allowed to browse. This might include "scratch" (or "ballpark") fields for: number (and ages) of people in family to be covered, rough family income estimate, and the state of residence. That really should be enough to browse, and all prices could be clearly marked "these are estimates only -- final prices can only be shown when the registration form is completed, with specific income figures."
The second problem is the most worrisome, though. I heard an expert who seemed to have inside information about the website's technological woes speak on the news last week (which has to be considered "unconfirmed rumor" status, it bears pointing out) and he said an astounding thing. The decision to force people to fully register before shopping around was apparently made in late September, when the website's rollout was planned for the first of October.
That's just flat-out insane, if true. Ask anyone in Silicon Valley, they'll tell you.
This is a major part of the design of such a website -- how the user is allowed to navigate through the site's pages. If such a major redesign was demanded mere days from its rollout, then no freakin' wonder it has such problems. The other shocking claim made seemed to reinforce this jaw-dropping piece of idiocy, too -- that while the Health and Human Services Department contracted out the programming of the site, they kept in-house the product management of the entire program. It's really the only explanation for why they were even able to make such a stunning last-minute design change.
If true, this could be the core explanation for the failures of the website as a whole. Software project management is a very specialized field. When done by amateurs, chaos of one sort or another should be the only expected result, really. It would be like hiring the manager of Grand Central Station to be the captain of an aircraft carrier. Sure, both have to do with "transportation management" in a loose way, but the two are simply not equivalent.
If the problems the Obamacare site is experiencing were due to a lowest-bid contractor doing some awfully shoddy computer programming, then the federal government should announce it is suing the contractor for non-performance of the contract's specifications. But if the heart of the problem was that government bureaucrats had too much power over the project management of a gigantic and brand-new software system, then someone should be fired over at H.H.S. for not hiring an expert to guide the process.
The Obamacare website will eventually be fixed. Who, after all, remembers all the confusion when the Medicare prescription drug program was rolled out? Eventually, as Obama was clearly hoping for today, the problems will be solved and the Affordable Care Act will have to stand or fall on its own merits -- whether affordable care is indeed available to all Americans or not. No matter how big a bump in the road this seems now, it will in time be forgotten if the program itself is ultimately successful. But that shouldn't excuse finding out who was really responsible for such a disastrous rollout. Up to and including having some heads "roll out" the door as a result. If this had been a few minor problems in the first few days, it might have been excusable, but it's been three weeks now, folks, and someone really should be held accountable for the poor performance so far. Even if it was only the person who made the decision: "We'll just keep the project management itself in-house, what could possibly go wrong with that?"
-- Chris Weigant
Cross-posted at The Huffington Post
Follow Chris on Twitter: @ChrisWeigant