AVOID, THIS.
MakerOfErrors
Karma: 130
Practical, actionable advice ideas:
Seriously, if bookmarking and remembering a tab can change your life expectency by years, that’s one hell of a cost/benefit ratio. I’d put absurdly high odds on it being worth picking out a default hospital beforehand, so you don’t have to make a split-second decision in an emergency. Like, you just never see low hanging fruit with cost/benefit ratios that high in day-to-day life.
Like, maybe print off the info for a couple choices, and magnet them to the fridge with the closest one on top and the best one underneath? Add closest address to your GPS, and drive the route once so you know that you know it? I’m pretty sure this could be optimized further with more thought.
Also, I can’t find it now, but I think I recall Robin Hanson commenting on proximity of home to the nearest hospital being a major determinent in life expectency, along with things like rural living. I think someone was musing about whether property values were any higher near hospitals, but I don’t remember if they got an answer or what it was.
It just occurred to me that your link is likely to be an Incredibly Important(tm) tool for the weird sort of person who might actually be interested in themselves/friends/family not dying during a procedure.
(In opposed to just being interested in signaling how caring we are, or seeking medicine to feel cared for.)
Nice catch!
Googling the term brought me to the Vendor Lock-In Wikipedia page, but no page just for “lock in”, even though that’s a common term for this sort of thing. However, the “see also” section mentions Path Dependence, which mentions the Bandwagon Effect, which is the perfect term for the Craigslist phenomenon.
These aren’t all quite the same thing, but they all seem related. They all highlight different aspects or special cases of similar phenomena.
More generally, could we fight all such problems of this class by claiming to believe in Moloch, as a vengeful god? Then ask for religious exemption from all coordination problems where exemption is legally possible.
Why not organize a religion to spite its god, rather than worship it?
EDIT: The concrete benefits could come from a single commandment to defy Moloch whenever possible. All shoes must be velcro and exempt from dress codes, doctors must be specialists when available, and you can sue for religious discrimination if someone makes hiring decisions based on autodidact/community college attendance over Ivy League, etc.
I know the IRS’s definition of a religion is deliberately fuzzy, and allows anyone with a “sincerely held belief” to call their thing a religion. If other religious exemptions are similarly open, then it would be hilarious way to fight Moloch. After all, coordination problems are a real, empirically verifiable thing, which economists at least sincerely believe in.
If the Flying Spaghetti Monster is considered a valid belief in some legal contexts, then so should Moloch. And, personifications of natural forces were some of the first historical gods, so there’s a precedent. Egregores may not be physical things, but believing in the processes they are names for should qualify if people believing in separate “Non-Overlapping Magisteria” qualifies.
There’s a big gap between “should work” and “works in practice”, though. Anyone know how big this particular gap is?
Two things: an expansion on the “employers optimizing for IQ” model, and a defense of regulations as critical tools for *solving* coordination problems.
suppose that there’s a magical tower that only people with IQs of at least 100 and some amount of conscientiousness can enter, and this magical tower slices four years off your lifespan. The natural next thing that happens is that employers start to prefer prospective employees who have proved they can enter the tower.
I think most companies are sufficiently broken that they aren’t even capable of optimizing for the metrics which would earn them the most money. (Before anyone cries “EMH”, It’s a Principal-Agent Problem, and probably unexploitable unless you unilaterally control a whole company. :p)
For example, you’d think that law firms would be as ruthlessly motivated by money as anyone. They should always use the best available metrics to hire the most cost-effective lawyers. But, as Robin Hanson points out, they ignore track records and hire based on fuzzy personal impressions, even when this leads to demonstrably worse outcomes. And, this is pattern repeats in media pundits, teachers, etc. Why? What on earth are we actually optimizing for, if not expected revenue generation by the new hire?
Robin’s answer is that hiring managers are optimizing for looking good to their bosses, and to powerful elites more generally. In short, hiring managers are optimizing for prestige in the eyes of everyone who’s opinion they care about, not money. (At least, not money for the firm.) Maybe they do care about IQ to the degree that it gives them prestige, but not as much as you would expect from assuming they’re maximizing expected profits.
We like to see ourselves as egalitarian, resisting any overt dominance by our supposed betters. But in fact, unconsciously, we have elites and we bow to them. We give lip service to rebelling against them, and they pretend to be beaten back. But in fact we constantly watch out for any actions of ours that might seem to threaten elites, and we avoid them like the plague. Which explains our instinctive aversion to objective metrics in people choice, when such metrics compete with elite advice.
Ok, onto my second point:
I absolutely love the extended Tower metaphor for College burning 4 years of life and money on runaway signaling competitions. (Although it’s never stated explicitly, it screams it once the thought strikes you, and footnote 6 corroborates this. Not sure if footnote 5 was supposed to include a link to the original SSC comment, so I can’t verify that the tower metaphor started as a college metaphor.) At least, I loved everything but this:
simplicio: I agree that trying to build a cheaper Tower Two is solving the wrong problem. The interior of Tower One boasts some truly exquisite architecture and decor. It just makes sense that someone should pay a lot to allow people entry to Tower One. What we really need is for the government to subsidize the entry fees on Tower One, so that more people can fit inside.
That’s a bit of an unfair straw-man of tuition subsidies, conditional on the college metaphor being intentional. If we wanted to read people bashing the outgroup, we’d all go to /r/atheism. Like, I got a strong “childish kicking someone when they can’t kick back” reaction, which I found hard to put aside to read the rest. It’s thinly veiled, so not all liberals will realize they’re being kicked, but that doesn’t help much.
Obligatory defense which I don’t want to give but feel obligated to anyway: tuition subsidies are mostly for community colleges, or are only large enough to cover just what community college would have cost if the student decides to burn the cash on the added prestige of another school. No one is suggesting everyone attend Harvard, or hand out Harvard-sized subsidies.
And, college does offer some real education along with the signaling. The art degrees are a form of countersignaling by the upper class, but if you look at the degrees that lower class people get, I would bet that they skew much more toward trade schools and more practical information. (Also, observation-selection effect may be incredibly strong here.) So, tuition assistance programs likely do far more good than harm. Sure, it would help more if some of it wasn’t going to zero-sum signaling competitions, but the subsidies themselves are hardly as big a driver as they would be in the Tower model.
</end obligatory defense of defenseless punching bag.>
And, more generally, I see this as a symptom of a larger problem. I Agree Denotationally But Object Connotationally with a lot of the libertarian undercurrents.
Maybe future chapters will feture more of the Dilbert-scale inadequacies in industry, like Robin Hanson’s point on law firms. But I’m worried not, based on quips like this:
visitor: Do they not have markets on your planet? Because on my planet, when you manufacture your product in a crazy, elaborate, expensive way that produces an inferior product, someone else will come along and rationalize the process and take away your customers.
Although, while scanning through looking for that quote, I noticed the Craigslist thing, which removed about half my worry. The remainder would be solved by explicitly stating that privatizing and/or deregulating everything won’t magically make everything better, as they appear to have on the Visitor’s planet.
(And, for those who need evidence for that claim:
1) Cost Disease is just as bad in for-profit hospitals as not-for-profit ones.
They’re minimally regulated. There’s no credentialing process or anything. There are many different kinds, each privately led, and low entry costs to creating a new one. They can be very profitable – pretty much any rehab will cost thousands of dollars, and the big-name ones cost much more. This should be a perfect setup for a hundred different models blooming, experimenting, and then selecting for excellence as consumers drift towards the most effective centers. Instead, we get rampant abuse, charlatanry, and uselessness.
On the other hand, when the government rode in on a white horse to try to fix things, all they did was take the one effective treatment [which the rehab clinics also strongly discourage], regulate it practically out of existence, then ride right back out again. So I would be ashamed to be taking either the market’s or the state’s side here.”
So, it looks to me that any given regulation has about 1:1 odds of either hurting or helping. I would love to have a density-map of inadequate equilibria by things like different industries, regions, scientific/academic disciplines, section of government, type of organization (publicly traded, privately owned, government owned, charity, church, co-op, hobby, club, sport, and anything else humans do in groups), size of organization, age of organization, organizational structure, etc. Not as a way to choose designs, since honestly most human institutions need to be redesigned from the ground up using game theory, but as a map of what needs fixing.)
So, using all these politicized examples is probably a Bad Idea.This is difficult to avoid, given the subject mater, but your usual habit of using historical examples whenever making inherently political points was a good one. Maybe that could be done for the most controversial ones? (Doing it for all would leave the reader thinking “yeah, but is that even still a problem today?”)
And, I especially don’t want a “yay-markets! Boo-government!” message to dissuade EAs from interacting with politicians or trying to push for legislation. It can be extremely influential to pull sideways in policy tug-of-war. Both parties may be pulling in opposite directions on some issues, and have no spare time (read: no free energy in the system) to invest in researching possible policies on an orthogonal axis. If you do all that for them though, and just hand them a Pareto-improvement, then both sides will likely support it, or at least not oppose it bitterly with every fiber of their being, like with partisan issues.
After all, what politician wants to be the guy to kill a important-sounding thing with no downsides? They might support the regulation or whatever it is, just to avoid the bad press.
Government’s main benefit to humanity is the ability to unilaterally solve coordination problems. It can unilaterally discourage pollutants, or build roads, or regulate monopolistic utilities. Sure, things like unions can fight monopolies too, but nothing else has quite the versatility of government. It’s not the only tool we have in the fight against Moloch, but it’s the biggest single one. Maybe Moloch is corrupting it too, but it hasn’t fully lost that battle yet.
Related data:
A couple years ago, there were a bunch of sensational headlines along the lines of “US is an Oligarchy, not a Democracy, New Study Finds”. The actual study is an interesting read: Testing Theories of American Politics: Elites, Interest Groups, and Average Citizens
It correlated the average preferences of average citizens, economic elites, and special interest groups (divided into “mass public interest groups” and “business interest groups”). Interestingly, Table 2 shows that economic elite preferences actually correlate with average citizen preferences at .78. Also, business interest groups correlate with mass public interest group preferences at −0.05, but all interest groups overall at 0.96, so I guess the vast majority of interest groups are business interest groups? Aside from “all interest groups”, business interests seem to correlate negatively with everything else, although not as negatively as I would have thought.
They then looked at the probability of various legislation passing, given that various categories were for or against it on average. They focus on cases where, for example, economic elites favor something but average citizens are against it, and compare it to actual policy outcomes passing or failing. They build several models for power and influence, and do a bunch of statistics I haven’t read the details of to try to figure out how much influence various groups have on actual policy outcomes.
Eyeballing figure 1, if looks like if 10% of average americans support something, it has maybe a 30% chance of passing, but if 90% of average citizens support something, it has maybe a 31% chance of passing. Conversely, if 10% of economic elites favor a piece of legislation, it has only a 10% chance of passing, but if 90% of them favor it, it has a 60% chance of passing.
Both of those graphs were fairly linear, and had a good spread, with lots of legislation with both high support and high opposition. They also tried to graph the net number of interest groups in support or opposition to legislation, vs. the probability of it passing. However, in almost all cases, there were less than 4 more either supporting or opposing each bill, so all their data points are clustered in the center. I’m not sure I believed their model projections outside of that narrow range, but if you believe that weird-shaped s-curve, then more net interest groups in support of a bill does cause it to be much more likely to pass.
Take all this with a grain of salt, though. There have been a couple follow up studies which stepped this back substantially. As Vox puts it “When the rich and middle class disagree, each wins about half the time”:
And also:
That makes you wonder: If only 7.4% of the variance is due to these groups, then what on earth is determining policy??!!
Right. Obvious answer is obvious.
Or maybe the middle class, since they are conspicuously absent from “rich, the poor, and interest groups put together”? I think “rich” is still being defined as top 10% here, and “poor” as bottom 10%.
Or, some interest groups are much more effective than others, so their preferences will be the main driver of everything, but only weekly correlated with the average interest group preference. The same could be true of certain influential elites, if their opinions diverged from those of the rest of the elites. But, that’s a wild guess on my part. This whole thing is a mess I haven’t even begun to sort out in my head.
But it’s probably Moloch.