An ongoing review of politics and culture

Of course Healthcare.gov calls the liberal project into question

First of all, kudos to liberals for being reality-based. After some equivocation, most liberals have succumbed to facts and admitted that the Healthcare.gov launch is a disaster—not glitchy, but a disaster.

As sure it is that the Sun rises, conservatives have pounced on the Healthcare.gov disaster to argue that it shows that the entire liberal project is made not credible.

And as sure as it is that the Sun sets, liberals have responded with a chorus of “Nuh-huh!”

The representative example is Matt Yglesias, who makes the facile point that bad websites notwithstanding, we find that plenty of countries and plenty of locales have functioning social insurance systems, so it’s silly to say that a bad website discredits the whole liberal enterprise.

Meanwhile Josh Barro notes (sorry for lumping you with the liberals, Josh!) that some DMVs are awful, but some DMVs are not and are pretty good. The implication is that there’s nothing intrinsic to being run by government that makes a system be crappy. Some government systems are badly managed, some are well managed.

There’s an obvious sense in which this is correct, at least superficially (but not unimportantly). Yes, Germany had a social welfare state in the 19th century. Yes, empirically, we do find that reasonably well-run government services can exist. And yes, conservatives should pay particular attention to the latter point, in the Tea Party and post-Bush era, because given that the welfare state won’t go away overnight they ought to pay particular attention to managing it well, and not just pruning it.

But, in the immortal words of Cyrano de Bergerac, “That’s a bit short, young man.”

How do I put this?

My main frustration with moderates and neoliberals is that even though they’ve reconciled themselves to the idea that some sort of capitalistic system is necessary in any worthwhile society, they seem to have expended no effort in understanding why capitalism works (with “works” defined as “being the worst economic system except for all the others that have been tried”).

They seem to take that as an inexplicable fact of life, and asking why it works would be akin to asking a mountain why it decided to sit there and not somewhere else. Okay, fine, the neoliberal seems to sigh, capitalism exists, it’s the least-bad thing there is, and we ought to make our peace with it, and advance with our liberal project anyway as best we can. In a way, this is an admirable posture, made of both pragmatism and fidelity to strongly-held moral instincts. And for obvious reasons I would much rather hand the reins of government to a neoliberal or a social-democrat than to a Communist; in fact I would much rather hand them to a smart neoliberal than to a dumb libertarian.

But to some advocates of free markets, the success of capitalism is not some bizarre, inexplicable fact of nature, it is something that tells us (or ought to tell us) important things about human behavior and patterns of organization that lead to collective success.

To some of us, what the success of capitalism shows, and the reason why capitalism works, is, basically, that humans are pretty much hairless monkeys who are not just greedy and quarrelsome but not very smart. Because us monkeys are dumb and proud, we can only assimilate a little bit of information, and we are terrible at communicating with each other. This, in turn, means that us monkeys are pretty lousy at two important tasks: planning and coordinating. We really are terrible at it.

Now this happens to have some pretty significant consequences for how you might help a tribe of monkeys achieve some progress in material welfare. If your tribe of monkeys works by having one Head Monkey tell the other monkeys what to do, the problem with that is that any monkey, being bad at assimilating information, will give the other monkeys really dumb instructions. And since the head monkey (and the other monkeys) are not only dumb but craven and quarrelsome, you will soon have chaos.

Paradoxically, if you let most of the monkeys mostly just do what they want, actually, they will bumble and stumble and grope their way towards a much more advantageous state of being.

You see where I’m going with this, right?

One problem with current society is that we take many things for granted when we shouldn’t. Everyone “knows” that Communism doesn’t work and capitalism at least sorta-kinda works. But this is a superficial realization. Understanding why is actually profound.

Central planning fails, and spontaneous order arises, because central planners never have enough knowledge, never have enough insight. Meanwhile, because humans are limited in their use and transmission of information, those humans who are closest to the thing they’re doing will have a much better understanding and be able to use these resources much more effectively. And, again, because humans are dumb, we find that decentralized means of aggregating information (whether commodity markets or Wikipedia) actually work much much better than centralized means, and are very precious.

In fact, if we look around at the Universe, we find that for some reason decentralized bottom-up innovation is a very very powerful force.

One of the many lamentable consequences of the Creationist movement has been that among educated elites it has become a kind of shibboleth to not only affirm faith in evolution by natural selection but to sort of scoff at any doubt that evolution by natural selection might not be obviously, self-evidently true.

But it’s worth it for just a minute to take off our cultural blinders and note how shocking to intuition and common sense evolution by natural selection really is. When a Creationist says “How could something as intricate and complex as an eye have evolved?” our first instinct is to get angry because we might be educated monkeys but we’re still monkeys, but we really should appreciate how, on its face, it really seems absurd that you could postulate the evolution of an eye through a totally unguided, bottom-up process. How much more intuitive, how in fact extremely compelling is the idea that something as complex, as awe-inspiring, as intricate, as glorious as an eye, or a tree, or a flower, or a hummingbird just has to be the direct product of an intelligence.

But in fact, we do know that eyes come from this unguided process, because for as much as humans can build rockets to space and stuff, we basically don’t even understand what makes a cell divide. Through our top-down, knowledge-intensive monkey way, we can build a rocket, but in terms of complexity and sophistication, a rocket to Mars is basically as advanced as a flint arrow, compared to an amoeba. And it takes all of human intelligence to build a rocket, but literally zero intelligence to build an amoeba!

That’s some wild shit, man. So wild that even though we actually know this is true, we totally ignore it.

We see this superiority of bottom-up trial-and-error over central planning all over human affairs. Tim Harford wrote a whole book about this, you really should read it.

We see it in the scientific method. Aristotelian science was a kind of scientific central planning, where scientists used reason to acquire “knowledge of the ultimate causes of things.” The problem is that because we’re dumb monkeys, we can’t actually know the ultimate causes of things. And because we’re not only dumb monkeys, we’re proud monkeys, it’s very hard for us to actually convince ourselves that we can’t know these things. And so science was basically just derp for roughly two thousand years; even though, and this is crucial, early scientists were actually really intelligent people! The point isn’t that they were dumb; the point is that they weren’t. But they were still monkeys.

Then we invented the scientific method, which could really be called “knowledge for monkeys.” Don’t actually try to know very much. Just do experiments, and see what works. Keep your experiments as simple as possible. Repeat your experiments as much as possible.

Again, because of the benefit of hindsight, because its benefits are all around us and have been all our lives, we take it for granted that it’s obvious that the scientific method is superior, and it is, but it’s also worth pausing for a second to note how bizarre it is.

You’re telling me the way to know more about the causes of things in the Universe is to actually stop looking for them? That to understand natural phenomena I shouldn’t actually try to understand them? What are you, some cut-rate Zen master or some shit?

(Yes, I know that things are a little bit more complicated than that, because science evolves through experiment within theoretical paradigms and not merely by bottom-up experimentation. But the point that bottom-up experimentation is crucial is true—in fact, it’s only bottom-up experimentation that changes paradigms.)

In fact, this is so hard to understand that we still don’t understand it, or else anyone who sees a headline with the words “studies show” would instantly projectile vomit. Most people on the street think science is this weird machine that somehow produces “truths.” Heck, even scientists think that science creates capital-t Truths instead of limited, small-t, contingent truths subject to experimentation, otherwise Richard Dawkins would have taken a vow of silence. It’s a testament to the power of the scientific method that even people who don’t understand it can use it profitably.

As Feynman said, “science is the belief in the ignorance of experts”, and yet most people today believe that science is the belief in the knowledge of experts. Because we’re monkeys who are dumb and proud, and because we’re proud we refuse to acknowledge that we’re dumb. In the past, we took priests of a religion that proclaimed the value of property and turned them into wealthy princes (and monkeys that they are, they gladly obliged), and now we take priests of a religion that disregards capital-T truths and turn them into purveyors of capital-T Truths (and monkeys that they are, they gladly oblige).

Another point about monkey-knowledge (aka, science) is that the scientific method is only successful on a relative scale. The scientific method is stupendously superior to Aristotelian science, and yet most science experiments fail. Most results are highly contingent, and limited, and even the promising results turn out to be non-replicable.

But that’s the thing about bottom-up, trial-and-error experimentation: most of it fails. In fact, it’s if you don’t have failure that you have a problem. The key thing is that you have a process that identifies success (in science, replication) and that lowers the cost of failure (so that you can run more experiments).

So if you’ve been following me all around this roundabout way, you will see where it leads back to capitalism and to theories of government. It turns out that we’re dumb monkeys who are not only dumb, but too proud to admit to ourselves that we’re dumb. But it turns out that, because we’re dumb, proud monkeys, the least-bad way we’ve got to achieve some process is through bottom-up trial-and-error experimentation, because that’s the thing dumb, proud monkeys do the least bad at. And even though bottom-up experimentation will be very messy, very imperfect, very problematic, it will still be much, much superior to central planning because monkeys are even much much worse at that. But because we’re monkeys who are not only dumb but proud, we refuse to acknowledge our dumbness.

We acknowledge the superiority of the scientific method because we are forced to, but we refuse its epistemological implications. We take the success of science to mean that we can have knowledge after all, but what the success of science actually means is that we can’t.

And we acknowledge the superiority of free-market capitalism because we are forced to, but we refuse to acknowledge its human and social implications. If free markets really work better than central planning, then we really are dumb and proud monkeys and we should really start organizing as such. But because we’re dumb and proud monkeys that’s a really really hard pill to swallow.

In fact, some advocates of free market capitalism make this mistake too. Capitalism works, they say, because capitalists are (somehow) super heroes with super awesome ideas that come fully-formed out of their thigh and that, in their great generosity, they allow us to have for just $9.99. And capitalists are not just awesomer-er, some versions of this argument go, they’re better.

No! Monkeys, the lot of them! Jobs, monkey. Bezos, monkey. Musk, monkey. Edison, monkey. Ford, monkey.

So here’s what I’m driving at. It’s a reasonable generalization, I believe, to state that the liberal project rests on a (largely unspoken) assumption, which is that planning can work. Or, perhaps more accurately and charitably, that to have successful planning is only moderately difficult.

Thus the DMV argument. Sure, some DMVs are bad, but some DMVs are good, and if you put good people in charge of the DMV, then the DMV will work.

The problem with that isn’t that the DMV can’t work.

The problem with that is that the overwhelming record of history shows that it’s highly unlikely that the DMV can work in any sort of sustainable way.

The reason for that is that planning is much, much harder than we are inclined to think.

Planning is very hard because we are dumb monkeys.

And we are strongly inclined despite all evidence to the contrary to believe that planning can work because we’re not only dumb but proud monkeys and we don’t want to admit to ourselves that we’re dumb monkeys.

And when I say planning I mean planning.

I don’t mean “the market” versus “the gummit.” I mean planning.

All planning is extremely difficult.

In some sense, it’s utterly baffling that CEOs of big companies think the private sector is more efficient than the government, because boy, are big companies utterly awful. Big companies are notoriously awful at planning. If big companies were good at planning, no startup would ever succeed, and yet not only do startups succeed, they succeed all the time.

Conservatives like to grouse at the perverse incentives of tenured public school teachers, but any big corporation is a tangled mess of awful, destructive incentives.

It’s monkeys all the way down.

99.9% of science experiments fail, and yet that is still so much better than anything else.

This understanding of the role of limited information and our own limitedness is really, really, really, deeply counter-intuitive, but it is also really important. Because we are dumb, proud monkeys, we think we know a lot more than we actually know. And so we make big mistakes. We invade Middle Eastern countries. We think we can centrally plan an industry that’s 20% of GDP. And when the car ends up in the ditch, we say “Guess the plan was wrong!” No, you chimp! The plan wasn’t wrong. A plan is impossible. We’re willing to admit our plan was wrong. It is much, much, much harder to admit that planning itself is impossible.

Donald Rumsfeld is actually an extremely talented and competent guy. Just like Bob McNamara before him. Just like Cass Sunstein. Just like you-name-it. The point isn’t that these people are dumb and if we got smart people instead they would do better. The point is the opposite. The point is that these people are very smart, but you’re still asking them to do something that is impossible.

When we see a company fail, we reassure ourselves that we aren’t monkeys by declaring that the people who run the company are stupid. It’s much harder to admit the truth, which is that in most cases they’re actually very smart.

The point isn’t that a well-run DMV is a physical impossibility. The point is that if you assume a well-run DMV as a crucial part of your plan, then you are basically rowing against the laws of history and human nature. You are raising a middle finger to God in an Old Testament story where God routinely smites people who tell him to f off. But hey, you might get lucky! But “This extremely unlikely thing that basically only ever happens intermittently and in very specific circumstances just needs to happen for this extremely ambitious and far-reaching initiative to work” is a very very poor rationale for public policy. And yet it’s the rationale we always employ! We don’t put it that way of course, least of all to ourselves.

The reason why Healthcare.gov puts in question the liberal project is because you do not build something like Healthcare.gov unless you have a delusional confidence in the possibility of planning.

As Yuval Levin pointed out, Healthcare.gov is not a bad website selling a good product, Healthcare.gov is the product. There is now a non-trivial chance, acknowledged even by liberals now, that Healthcare.gov’s malfunctions could put the entire American individual insurance market on a death spiral. Whoops. If Healthcare.gov doesn’t work, all of Obamacare could implode, and with it the US healthcare system. Whoops. The overwhelming record of history shows that websites such as these are extremely extremely hard to pull off (and if you think that’s hindsight talking, talk to anyone involved in large-scale IT projects, they could’ve told you). The point isn’t that it’s impossible for any government to make a website that works. The point is that the kind of people who are so confident in their ability to make a good website that it’s a crucial part of their plan against all good judgement should not be trusted with the levers of public policy, because they are dangerously delusional. And we have a word for that kind of people, and that word is “liberal.” A prerequisite of being a liberal is having a faith in the possibility of planning that, when put together with the evidence of monkeyness, is exposed as a delusion. A highly understandable, a highly natural, delusion, precisely because our monkeyness makes us delusional. But a delusion nonetheless.

Ok, so when does planning work?

Answer A: Almost never.

Answer B: When it does, it almost only ever works in very limited and partial ways that involve a lot of bottom-up experimentation.

Ok, but what about World War II/the Manhattan Project/the Apollo Project/DARPA/that DMV that runs good?

Yglesias points to the fact that NASA can put a robot on Mars as evidence that the government can do stuff. Of course, for a country that sent a probe to the outer Solar System 45 years ago, putting a roomba on Mars is laughably pathetic.

It’s not the space equivalent of Healthcare.gov, because the roomba does get to Mars instead of crashing on the White House lawn, but it’s very much the space equivalent of right-wing-talking-point-DMV. Compare the evolution of space technology to the evolution of information technology over the past 50 years. Which one was driven by government? Riiiiight. Now read the essay A Rocket to Nowhere and have a good cry. When the space industry really gets into gear, the wretched state of what passes for space exploration will become as glaring as an internet.

But, ok, ok, the government does do cool stuff once in a while.

That being said, when you understand, actually understand, deep-down understand our monkeyness, and understand the difficulty of planning, you understand that the instances of successful government planning actually prove the case for monkeyness, just like the failures of private-sector corporations do. The point isn’t that there’s some awesome magic thing that makes the private sector efficient and gummit bad, the point is that we’re monkeys. If you think private sector good, gummit bad, you still think that there’s a place where there’s no monkeys. Monkeys everywhere, you monkey. Repeat it every time you go to bed.

As I wrote about DARPA inventing the internet, the way the government invented the internet was through bottom-up innovation. DARPA hired a bunch of hippies and gave them free rein to screw up tons of things. It was totally unbureaucratic. It was, in fact, all more than a little bit insane—the way most startups are insane.

I don’t think I’m descending into Rush Limbaugh-ism when I say that this is quite far from being the “default setting” of government. The success of DARPA calls into question the liberal project for the same reason as the failure of Healthcare.gov does, because it shows the futility of planning, and the virtues of bottom-up experimentation. It shows the futility of “markets good, gummit bad” which is another version of rejecting monkeyness. It shows the validness of embracing monkeyness. DARPA is the exception that proves the rule—not in a facile idiomatic way, but in the profound way that it stands out precisely because it is so unlike the way government normally operates.

Again, the point isn’t that there is some inherent property to government that makes it sucky, the point is that the way government is set up makes extremely unlikely that our monkey natures will be oriented towards good outcomes.

When writing about the scientific method I showed that one thing that makes these systems work, besides bottom-up experimentation, is clarity of a definition of success (in the case of science, reproducibility) and a possibility of failure leading to re-experimentation.

The Manhattan Project and the Apollo Project exhibited those traits. First, the definition of success was abundantly clear. Make a bomb make a really big boom. Put a man on the Moon and return him safely. (Compare with, say, education and healthcare, where it is quite literally impossible to define success adequately.) Second of all, the possibility of experimentation was there. Really really smart people were taken and written blank checks. (Note, again, how unlike the normal operation of government this is.) Finally, there was enough political will that failure was taken in strides. When some of the Apollo rockets blew up, Congress didn’t all of a sudden decide that NASA would be run by lawyers, because everyone was scared shitless by the Soviets’ early lead in space. Again, contrast this with the mess that NASA is today.

Tom Ricks convincingly described how, during World War II, most US Army generals failed, and were very quickly removed when they did. For all the top-down-ness that is inherent in the US Army, necessity being the mother of invention led them to, in spite of themselves, create some bottom-up innovation. Individual corps commanders who performed were rewarded, those who didn’t were culled. There was much unfairness to the system—not infrequently, genuinely talented commanders were unlucky and removed nonetheless. Just like many talented entrepreneurs experience failure. But it still made the system much healthier overall. Ricks aptly contrasts this with the ineptly bureaucratic way the US military now functions.

True fact: which was the first large organization to experiment with making employees rate their superiors? The Wehrmacht. Yep. (Source: Wikipedia, so it’s gotta be true.)

What we find is that top-down planning can kinda-sorta occasionally work under basically the following conditions: 1) clarity of objective (invade Europe; make plutonium go boom; land on the Moon; make money); 2) effective urgency, by which is meant that either necessity (kill the Germans or they’ll kill us; release a great new personal computer or go bankrupt) or a great political will (“we choose to go to the Moon”; the iPhone will be ready to ship by June) which seems to get the monkeys to be somewhat less awful at grokking and sharing information and letting their pride get into the way of collaboration; 3) extensive bottom-up experimentation within the top-down framework (and sometimes causing change/abolition of the top-down framework).

We find that these conditions quite rarely obtain, and that these conditions seem to be necessary but not sufficient. Top-down planning fails plenty even when these conditions are met.

We also find that while these conditions obtain rarely overall, they obtain vanishingly rarely in the government sector. Meanwhile they obtain less rarely in the private sector, because the private sector is by its nature more (not completely, not intrinsically, but slightly more) oriented towards bottom-up innovation, creating failure and bounceback, and providing clarity of objective.

I think this is also why founder-driven companies consistently outperform professional manager-driven companies. It’s not that they’re genuises who have this great vision, or at least it’s not just that. Yes, having a better top-down plan helps. But in the end they’re monkeys too and they screw up all the time. It’s that they provide the (2), the effective urgency which breaks down our innate, monkey tendency towards bureaucracy, and waste, and stupidity, and focuses us.

Again, I want to emphasize the extent to which the private sector is by and large an enormous engine of bureaucracy. Most large companies are only slightly more efficient than the proverbial DMV, and many may be less.

But as a prudential rule, assuming that the government can’t plan things makes a lot of sense, exceptions notwithstanding.

It’s a prudential rule, and it’s a good one. Just like when we prevent banks from having almost no capital, we’re not saying it’s impossible for a bank to have almost no capital and still succeed. But we still don’t allow that because we recognize that a) it’s still very unlikely; and b) when it fails the consequences are catastrophic. It’s possible for a financial institution to be totally reckless and succeed, even without bailouts, but the grain of the Universe goes against it. It’s possible in the same way that kayaking upstream of a torrent is I guess maybe possible sometimes? I know nothing about kayaking. It’s possible only when a bunch of stars that almost never align align. And even when it’s possible, history teaches us that it’s only possible for a time. Again, the analogy with government planning works. NASA once pushed the frontier of human exploration and knowledge, and now it’s building rockets to nowhere. And when you say we should stop building rockets to nowhere and unleash some entrepreneurial innovation, you’re accused of hating America and pissing on everything. But the Apollo Project! Precisely. Monkeyness tells us the Apollo Project is possible, and also not repeatable as such.

The law of monkeyness is to human affairs as the law of entropy is to the Universe. Once/if Healthcare.gov is up and running again, we will see plenty of longread post-mortems in our nation’s greatest journals, and plenty of earnest wonks will inweigh on the specific mistakes that were made and say: “See? If we hadn’t made specific mistakes X, Y and Z, it would have been fine!” The wiser, monkeyness-embracing realizes that by the nature of the thing mistakes will always be made. And in some complicated way that is impossible to forecast a priori, the attempt to prevent mistakes X, Y and Z will lead to mistakes A, B and C next time around. Such is the way of dumb, proud monkeys.

If you add monkeyness to the nature of modern democratic government, you quickly realize why monkeyness makes successful government planning highly unlikely not just in abstract, but now. As we’ve said, two necessary conditions for functioning central planning include tremendous political will and a tolerance for failure. These are the things that are absent from modern democratic system. And for good reason. It’s great for humanity that Steve Jobs terrorized Apple employees into making gorgeous phones, but the idea of Steve Jobs in control of the FBI makes Nixon look like a Quaker.

Democratic government is a government for monkeys, and that’s why it works. It’s a government for monkeys because monkeys get arbitrarily fired and replaced by other monkeys at regular intervals, which means there is some learning-through-failure that goes on. But it’s a government for monkeys because it’s full of checks and balances. The whole idea of checks and balances is that politicians are craven and proud and dumb and are going to spend time fighting each other instead of fighting the people. And it’s wonderful! It embraces monkeyness.

But the checks on government, particularly in the modern era, aren’t just checks of the constitutional variety. Popular opinion is also a great (perhaps the most important) check on politicians. And it’s easy to see how the influence of popular opinion makes planning impossible. Absent Sputnik scaring the crap out of the Nation, if you’re the monkey who decides to make NASA go to the Moon and your rocket blows up killing a photogenic astronaut, a Congressmonkey will give you a proctology exam on national TV. As I wrote, bottom-up experimentation is glorious over the long haul but incredibly messy up close, and the nature of politics ensures that all of the messy experimentation and failure is going to look awful and be a huge political hit. Which explains why the successful examples of government planning happened as a result of World War II or the Cold War, where the existential threat upon the nation briefly and partially suspended the normal rules of politics.

That’s the impact of broad popular opinion, but there is also the impact of narrow popular opinion, ie special interests. NASA is probably, net-net, a waste of resources at this point, but seriously reforming it is politically impossible at this point because too many entrenched interests are at play.

Now if you’re a conservative, this is okay, because monkeys are monkeys. A democracy is a regime for monkeys and this is good (or, less bad than the alternatives). A market is a regime for monkeys and this is good (or, less bad than the alternatives).

But plenty of liberals sometimes realize that the (or, more accurately, a reason) reason why it’s so hard to have Government-Planned Projects is because of this messy business of politics and checks and balances. Or, in other words, democracies. And so plenty of liberals, with the best of intentions, find themselves really really craving more power for the central government, so that we can finally get those big projects gone. Now I’m not saying all liberals are closet fascists, or even some of them (well, maybe some of them), but, again, we are proud monkeys. Proud monkeys crave power, and because they are proud, they need to reassure themselves that they are craving power for other reasons than the love of power. And I don’t think it’s unfair to say that liberals have an impulse towards increasing the power of central government.

And this is where conservatives start thinking about Second Amendment remedies. Because, of course, it’s not only deeply unwise (politics is only the proximate cause of the failure of planning; see: Soviet Union. No, really. See it. ), it’s also very dangerous and deeply morally wrong. It would be good if we could give lots of power to a benevolent non-monkey, but, again, we’re monkeys. And monkeyness means not only that we’re very awful when we get power but also that we’ll never learn to not crave it anyway, even with the best of intentions.

Meanwhile, markets, because they encourage bottom-up experimentation, because they have mechanisms for aggregating information which are necessary for our stupid monkey brains, because they provide failure-bounceback mechanisms, because they provide (some) clarity of objective, tend to be preferable to central planning whenever they can’t work. Not because they don’t suck. They do. Because they’re just an organization of monkeys. But they just suck slightly less. And that slightly less happens to make a very big difference.

…

Okay, so where does that leave us?

To the complicated and counterintuitive point that the overwhelming record of history is that bottom-up innovation produces the best outcomes, particularly for groups of monkeys such as ourselves. That monkeyness makes planning extremely hard. That planning should be resisted not only because it’s very unlikely to work but because proud monkeys will always overestimate their own skills at planning. That government planning in particular should be resisted because not only is it even more unlikely to work than other forms of planning, but because the consequences of failure of public sector planning are much more likely to be catastrophic than the consequences of private sector planning failure.

And if you want an example of that, look no further than Healthcare.gov.