What Went Wrong

Okay. I’ve never worked as a tech geek, so I’m speculating from ignorance here. Some of you can probably speak with more authority. Perhaps we’ll hear from the reliably acerbic and insightful Doctor Memory, who knows whereof he speaks on this subject (and several others). But to my uneducated eye, it appears that Arnold Kling has got this pretty much dead-on right: The Obamacare mess “is not a technical screw-up, and it will not be fixed by technical people. It is an organizational screw-up.”

What you had here, among other things (and almost of this is paraphrasing Kling) is:

  • A bunch of people who had never worked in the insurance business appointing themselves executive officers of the world’s largest insurance brokerage.
  • Nobody at the top with the authority to trim features as needed to keep the project manageable.
  • No mechanism for the technical staff to challenge the managers, because all of the management decisions were essentially set in stone before the technical staff — i.e. the outside contractors — came on board.
  • No clear lines of authority and acceptability.

Private enterprises frequently fail, often for one or more of these reasons. But sometimes they succeed, and that’s largely because sometimes they get this stuff right. The government, by contrast, has no mechanism for getting it right. The people at the top are not industry experts, the features are largely determined by the legislative process, which takes place with absolutely no feedback from the tech geeks who are going to have to implement it, the political system pretty much forces you to put the technical part of the project out for bid and to parcel it out among multiple contractors, eliminating any possibility of ongoing negotiation between the managers and the techies, and on top of all that, nobody’s livelihood is on the line.

These are not problems with Obamacare per se; they are problems with large-scale complex government-run enterprises in general. These are the reasons why many writers, including me, were extremely skeptical from the outset that Obamacare could be made to work well. It’s true that none of us (as far as I know) was prescient enough to predict that the website would be a particular point of failure. (Edited to add: Actually, it looks like Megan McArdle was out front on this.) But I think that foreseeing big problems, and foreseeing the reasons for those problems, still counts as prescience.

If you’re tempted to respond that plenty of large-scale complex government-run enterprise work just fine, please provide examples. If your example is Medicare, consider first that an enterprise on the verge of bankruptcy is not typically regarded as “working just fine”. You might want to make a case that government-run programs should be judged not by their financial health, but by their impact on social welfare — and I would agree with you. But I’d still ask for evidence that Medicare works “just fine” by any standard.

Kling has a lot more to say about where it is and isn’t possible to go from here, and it’s all worth reading. Once again, the link is here.

Share/Save

30 Responses to “What Went Wrong”


  1. 1 1 Will A

    Kentucky Kynect, Access Health CT, Washington HealthPlanFinder, and Covered California are apparently very successful implementations of the Affordable Care Act.

    Kentucky Kynect seems to be the most successful.

    These are systems that have just been created.

    However, if your argument is going to be that Kentucky Kynect doesn’t count as a good technical implementation because the affordable care is a bad law, then I guess it wouldn’t count.

    The IRS Web Site works really well, however if you find income taxation to be theft, you would probably say that any system no matter how technically sound that encourages theft doesn’t count.

    Perhaps you could let us know what you think are good laws/policy and then we can mention good technical implementations for these.

  2. 2 2 KS

    “If you’re tempted to respond that plenty of large-scale complex government-run enterprise work just fine, please provide examples. If your example is Medicare, consider first that an enterprise on the verge of bankruptcy is not typically regarded as “working just fine”.”

    What exactly qualifies as the ‘verge’ of bankruptcy?

    I would say social security

  3. 3 3 Navin Kumar

    Before you pat yourself on the back, let me point out that saying “The government will screw up” is a statement that will be right nine out of ten times. There is no great prescience needed.

  4. 4 4 Jens Fiederer

    Megan McArdle was at least ASKING the question back in March, and in February, so was “information technology consultant Edifecs” who polled health care/insurance execs to find 70.7% did NOT believe the exchanges would be ready on time. Hardly a stunning surprise then.

    http://www.thedailybeast.com/articles/2013/03/27/will-obamacare-s-exchanges-be-ready-on-time.html

    http://healthpolicyandmarket.blogspot.com/2013/03/six-months-to-go-will-health-insurance.html

  5. 5 5 Steve Landsburg

    Navin Kumar: There is no great prescience needed.

    That was precisely my point.

  6. 6 6 Alex

    “almost of this” -> “almost all of this”

    (You can delete this comment if you like.)

  7. 7 7 Manfred

    Will A:
    If the government is so successful in running the things you mention, then, perhaps, we should let the government run the whole economy, in big 5 year plans, all driven by bureaucrats sitting in some dungeon somewhere allocating resources by fiat.
    The IRS website works because it is a command economy, it works by compulsion, it has no competitors, and if they screw up, they have a massive, absolutely massive, PR problem at hand (not that the IRS cares, though, as recent experience shows).
    If Kentucky Kinect or whatever really works, my question is, why did not the private market create it on its own long time ago? Did really Leviathan have to come and wave its magic wand and create Kentucky Kinect? Please… again, it is a command economy, that works by fiat, by diktat. Anybody, even me, can allocate resources by diktat.
    Lastly, all you people, here is a suggestion: read the Wall Street Journal editorial of today, Wednesday, Oct 30. It summarizes what is probably the core issue of the ACA.

  8. 8 8 John

    I distinctly recall Megan Mcardle writing about how the website would be screwed up before anyone else.

  9. 9 9 TjD

    Had to respond to

    “No mechanism for the technical staff to challenge the managers, because all of the management decisions were essentially set in stone before the technical staff — i.e. the outside contractors — came on board.”

    First off, CGI has complained severely (after the fact) about every changing requirements, nothing was set in stone.

    Second off(?), what is the incentive for contractors to fight changing requirements if they get paid for time and materials, I think that is why no mechanism was set up to challenge the managers.

    T.

  10. 10 10 Dan

    Well I am something if a tech wiz, so what I can say is that many of the issues are technical.

    I didn’t work on this project, bust as I understand it, the issue relates to verifying and coordinating issues among servers on the backend. Healthcare.gov servers are fine, but the process requires them to check in with all kinds of other servers, from social security administration to private insurers. Healthcare.gov has to wait until these servers return results to its requests. It’s akin to going around your household to figure out what kind of delivery you want. Your process is going to be slower than your slowest link.

    The system supposedly works pretty will at a low request volume. But many of the other servers (particularly other government agencies, but also private businesses) are very slow and buggy. Many of them are old software sitting on old machines. Goverments and businesses alike are really loathe to ever upgrade to new machines, which is penny wise and pound foolish. But look at your own computer…are your running windows 8, or at least windows 7? Or are you running vista or god forbid xp? (In my experience, Mac users are much much better about upgrading, and my business is built on apple products).

  11. 11 11 Alan Wexelblat

    My father worked for Treasury for many years, including a two-year stint at IRS. I’ve heard a good deal about how screwed up government contracting is, in general. It has a lot to do with how budgeting is done and how Congress screws with the budgeting, but just try to imagine planning a multi-year project when you can’t know how much funding you have next year or – in the current paralysis environment – even in the next six weeks. If you tried to operate any business that way you’d go bankrupt, but that’s the way every government project has to be run because Congress will not commit to funding levels.

    Furthermore, healthcare.gov is hardly unique. If you want a truly epic fuck-up, look at the Joint Strike Fighter debacle. Medicare itself took two years after roll-out to get fully functional. The irs.gov site, which is pretty good these days, was something approaching complete disaster for years.

    People forget these things, and because the ACA is controversial, a lot of hay is made. But honestly it’s not unique and it’s not a bigger fuck-up than has happened before or will happen again.

    People also like to wave “Mythical Man-Month” around, and it has tons of relevant wisdom. But people want to pick and choose from it, forgetting that one of Brooks’ insights was that the vast majority of projects he analyzed (all commercial, market-competitive) were late, over budget, and wildly deviant from the understood requirements. That a government-run big software project followed the well-understood behavior of all other big software projects should surprise exactly no one.

    You (and other commenters) are correct that this is a failure of process and organization, not of technology. The notion of a “tech surge” is stupid on the face of it. Someone thought it would sound great in press releases and that was the end of that. That person should be smacked.

  12. 12 12 Al V.

    There were obviously many technical failings in the healthcare.gov effort, but I see one main problem that led to many others. Every large software development effort requires three key roles:
    - A sponsor. This is the ultimate decision-maker, which must make decisions when all other arbitration methods have failed.
    - An architect. The person who keeps the total technical solution in his or her head, and is responsible for resolving technical conflicts.
    - A program manager. The person who oversees all of the individual projects to ensure they are on track, that dependencies between projects are identified and dependent activities will complete on schedule.

    I don’t see any evidence that any of these roles were defined. Zients is now playing the role of the sponsor, but that was a recent change.

  13. 13 13 Kirk

    The IRS website works, if all you need is a quick look at a pub or to check the status of a refund. As The Super Tax Genius, I can tell you that IRS.gov is often the LAST place I go for information. The fact that I look up IRS guidance (written by the IRS) on a private website should tell you enough.

    Also, try doing a search there. You get garbage.

    Also, even Where’s my Refund failed miserably for weeks last year.

    The IRS website is a POS. Sorry to burst your bubble.

    PS – I love that my Tab on Firefox says What Went Wrong at Steven Landsburg The Big Questions!

  14. 14 14 Windypundit

    I don’t know the specifics involved with HealthCare.gov, but your last three points are common problems with failed software development projects.

    Regarding “changing requirements” v.s. “set in stone” there was a bit of both. Detailed requirements were not nailed down until late and some of them still kept changing. However, there was probably no flexibility to do things like, say, simplifying the implementation by postponing subsidized enrollments until next year, or only handling a limited subset of carriers or states in the first year. If they could have deployed a limited site like that, it would have reduced the scope of the system and therefore of the change requests, perhaps to a volume that would have been manageable in the time frame — HealthCare.gov might not have been as ambitious, but it might have worked better.

    The “time and materials” issue is a real problem. If you’re on a T&M contract and you try to save money, people look at you funny. It takes good management to stay on top of the spending.

  15. 15 15 Doctor Memory

    Hm, for a post where I’m being mentioned by name, I should probably try to have something interesting to say, but I’m not sure I have much to add to my previous comments: I think that some if not all of Dr. Kling’s hypotheses for why healthcare.gov shipped broken are likely correct, but the assumption that it is unfixable is simply premature at best. The four failings he describes are endemic in private IT projects as well: what distinguishes the victors from the roadkill is how quickly and diligently they fix the problems.

    (And sometimes more diligence than success: Apple has managed to cram iCloud down an impressive number of developer and user throats despite the fact that the entire project has been a rolling technical debacle for near a decade now under its various incarnations. Apparently you can learn to like the taste of roadkill?)

    It’s certainly a valid point that the fact that the exchanges’ launch date being determined by statute rather than formal Quality Assurance sign-off was pretty much a guarantee for a rough launch. But this, too, is an unsolved problem in private IT as well: imposed deadlines happen, but projects without deadlines sprawl endlessly. If anyone not selling six-sigma snake oil has a solution, I’d love to hear it.

    That said, I personally think that it’s a little rich for advocates of market-based healthcare solutions to point fingers at the feds for having difficulties implementing their own plan, the ACA being basically the 1993 Heritage Foundation proposal with the serial numbers halfway filed off. This rube goldberg contraption is the best that we could get under the circumstances, and a lot of the louder voices complaining about it are in fact the authors of those circumstances.

  16. 16 16 Windypundit

    “But this, too, is an unsolved problem in private IT as well: imposed deadlines happen, but projects without deadlines sprawl endlessly. If anyone not selling six-sigma snake oil has a solution, I’d love to hear it.”

    The solution of late has been Agile development methods: Short deadlines for small, incremental changes. It doesn’t really solve the problem, of course, so much as redefine it, by making scope the independent variable. I.e. you get regular deliveries of a working system, but it starts out really limited and adds features over time. If the feature list isn’t carved in stone, it tends to produce a series of successes.

    This turns out to be a great methodology for the business model of many web sites — release small and add features later — but is of little help if you’re building something that really only makes sense as one big piece, such as the Space Shuttle’s control system (a success) or the Air Force’s Expeditionary Combat Support System (a failure). Or HealthCare.gov, apparently.

  17. 17 17 James Kahn

    To expand on the first bullet point: The notion that in a few months the government could massively intervene in a nationwide decentralized system involving hundreds of millions of people, one that had evolved over decades, was just enormous hubris. The president, after all, evidently believes profits are “overhead,” i.e. that government and non-profits can provide services more cheaply than the private sector.

    We are only just beginning to see the damage, the broken website being the least of it. They could have proposed incremental small changes directly designed to address specific concerns, but they had such belief in large scale social engineering that they couldn’t help going for it.

  18. 18 18 Kirk

    A perfect example of why people who support government intervention to correct problems (which they almost always are correct about) fail to realize that the problem they have identified is actually caused by government intervention and that the likely result of more government intervention is more problems that will result in calls for more government intervention. In twenty years or less we will be introducing Single Payer Health Care to correct the problems caused by Obamacare which was created to fix the problems caused by COPM which was created to solve the problems of Medicaid and Medicare which was introduced to solve the problems caused by…

    http://reason.com/archives/2013/10/30/obamacare-is-the-latest-example-of-faile

  19. 19 19 Daniel

    @ Landsburg,

    Just speaking for the “Medicare on the verge of bankruptcy” scare mongering. et’s consider that the reasons Medicare might not be doing well financially are because we have an aging population and a fixed income tax rate that was implemented under a different age structure. Also the governments hands are tied right now because they aren’t able to negotiate on prices. What would happen to price increases if the government was able to negotiate? Krugman, your most hated enemy, has the answer to this already:

    http://krugman.blogs.nytimes.com/2013/09/21/i-have-seen-the-future-and-it-is-medicaid/?_r=0

    You’ll also see from this chart that Private insurance is doing at least as bad if not worse than Medicare at controlling prices. Medicaid evidently is doing by far the best.

  20. 20 20 Daniel

    @ Kirk,

    Cross country analysis would lead me to believe that government is not the problem.

  21. 21 21 RichardR

    “plenty of large-scale complex government-run enterprise work just fine, please provide examples” – lots of countries have government run health care systems which work fine. One specific example from Wikipedia (and no I haven’t edited the page!): “The French health care system is one of universal health care largely financed by government national health insurance. In its 2000 assessment of world health care systems, the World Health Organization found that France provided the “close to best overall health care” in the world.[1] In 2005, France spent 11.2% of GDP on health care, or US$3,926 per capita, a figure much higher than the average spent by countries in Europe but less than in the US. Approximately 77% of health expenditures are covered by government funded agencies.[2]“http://en.wikipedia.org/wiki/Health_care_in_France

  22. 22 22 Daniel

    Lol Richard,

    And your’s is just one example of many functioning “government run” healthcare system. Apparently though the US is in an alternative universe where all of government is incompetent.

  23. 23 23 nobody.really

    If you’re tempted to respond that plenty of large-scale complex government-run enterprise work just fine, please provide examples.

    Does the Massachusetts health care plan count?

    If your example is Medicare, consider first that an enterprise on the verge of bankruptcy is not typically regarded as “working just fine”.

    Verge of bankruptcy? That’s quite a revelation.

    Not a revelation about Medicare. I could disparage my local police and fire departments for being on the verge of bankruptcy, too — in the sense that they couldn’t pay their bills but for the fact that some legislators make periodic appropriations to maintain the services.

    No, it’s a revelation about the perspective of people who cannot see the world except through a prism of private market transactions.

    You might want to make a case that government-run programs should be judged not by their financial health, but by their impact on social welfare — and I would agree with you. But I’d still ask for evidence that Medicare works “just fine” by any standard.

    The first problem is not finding examples of large-scale complex government-run enterprise working just fine; the first problem is in defining “working just fine,” and explaining what makes that standard relevant.

    So how about we consider an alternative intellectual framework: Compare what the US spends on health to what other nations spend. Compare the outcomes. Ask why we should regard the status quo as “working just fine.” If we don’t, ask why we are comparing Obamacare to some idealized “working just fine” model rather than to the actual status quo alternative.

    I distinctly recall Megan Mcardle writing about how the website would be screwed up before anyone else.

    Wrong again. I was screwed up long before this website was; ask anyone.

  24. 24 24 Patrick R. Sullivan

    ‘…the reasons Medicare might not be doing well financially are because we have an aging population and a fixed income tax rate that was implemented under a different age structure.’

    And this is evidence that government CAN implement a program successfully?

  25. 25 25 Will A

    @ Manfred #7:

    If Karl Marx makes a post that says:

    I’d like anyone to show me a privately implemented computer systems that runs well. Only sites like Google and Amazon don’t count because those companies suppress the proletariat.

    Then my response would probably be something like:

    Yahoo and ESPN.COM seem to work well. But you would probably say that doesn’t count because they are stealing from their workers.

    Perhaps you can let us know what privately run organizations you think are good so we can find a good privately implemented Technical solution that matches what you think isn’t evil.

  26. 26 26 Daniel

    @Patrick

    No it was an alternative explanation for why its costs are increasing. Evidence of government implementing a program well is in the link I provided.

  27. 27 27 Seth

    I’ve witnessed spectacular failures in the private sector as well. I don’t buy that organization has much to do with it.

    I assume most things will fail. I’m surprised when they succeed. Everybody sees the lottery winner on TV, ignoring (nor being able to fathom) the hundreds of millions of fellow ticket buyers who did not win, and thinks ‘that could be me.’

    We see success stories in the private sector and it’s so tempting to try to tell the story of why. We ignore the many others that did something almost exactly the same way, but failed.

  28. 28 28 PE

    For examples of large-scale complex government-run enterprises that worked well, I’d say NASA putting a man on the moon, and the Manhattan Project.

  29. 29 29 Babinich

    “I think that some if not all of Dr. Kling’s hypotheses for why healthcare.gov shipped broken are likely correct, but the assumption that it is unfixable is simply premature at best.”

    The question is not whether something can be fixed. The question is what that fix will cost.

  30. 30 30 Harold

    #23 – it looks clear to me that the USA health system is not delivering value for money, but I also don’t know if people are claiming that “it is working just fine”. There seems to be a particulaly toxic blend of private insurance and govt care that ensures bad value for both.

Comments are currently closed.