I would like to compile an as-comprehensive-as-possible list of known Inadequate Equilibria / large scale coordination problems.

An Inadequate Equilibrium is a situation in which a community, an institution, or society at large is in a bad Nash equilibrium. The group as a whole has some sub-optimal set of norms and it would be better off with a different set of norms, but there's no individual actor who has both the power and the incentive to change the norms for the group. So the bad equilibrium persists.

Eliezer offers the following more specific criteria:

  1. Cases where the decision lies in the hands of people who would gain little personally, or lose out personally, if they did what was necessary to help someone else;
  2. Cases where decision-makers can’t reliably learn the information they need to make decisions, even though someone else has that information; and
  3. Systems that are broken in multiple places so that no one actor can make them better, even though, in principle, some magically coordinated action could move to a new stable state.

I want to generate as many real-life examples of this phenomenon as possible. Help me generate some?

The goal here is quantity. Originality is not required. 

As an additional incentive, I'll award $100 to the user who generates the most examples. (In order to count, each example has to have enough detail, or link to a detailed enough explanation, that I can understand how it is an inadequate equilibrium, and not just an unfortunate state of the world.) And I'll give $50 to the user who produces the second largest number. 

I'll prime the pump with a few (LessWrong-centric) examples.

Widespread adoption of prediction markets

It seems like our world would make saner decisions were prediction markets commonly used and commonly consulted. Making the switch from not relying on prediction markets to relying on prediction markets is fraught, because it might embarrass the leadership of existing institutions by revealing that their professed estimates are not very credible. 

As Robin Hanson said in a recent interview,

I’d say if you look at the example of cost accounting, you can imagine a world where nobody does cost accounting. You say of your organization, “Let’s do cost accounting here.”

That’s a problem because you’d be heard as saying, “Somebody around here is stealing and we need to find out who.” So that might be discouraged.

In a world where everybody else does cost accounting, you say, “Let’s not do cost accounting here.” That will be heard as saying, “Could we steal and just not talk about it?” which will also seem negative.

Similarly, with prediction markets, you could imagine a world like ours where nobody does them, and then your proposing to do it will send a bad signal. You’re basically saying, “People are bullshitting around here. We need to find out who and get to the truth.”

But in a world where everybody was doing it, it would be similarly hard not to do it. If every project with a deadline had a betting market and you say, “Let’s not have a betting market on our project deadline,” you’d be basically saying, “We’re not going to make the deadline, folks. Can we just set that aside and not even talk about it?”

Moving from proprietary journals to open-access journals

There's broad agreement that it is better for science to be open, and for anyone to be able to access scientific papers. 

Unfortunately, scientific publishing is currently dominated by a cabal of journals that can gatekeep access to most scientific papers, charging people (or institutions) for the right to access them.

Individual scientists might prefer to publish in open-access journals instead, but to do so unilaterally means taking a career-hit, because the most prestigious journals are not-open access, and publishing in prestigious journals is an important competent of the academic signaling game. So, modulo coordinated action, scientists in most fields are incentivized to publish in closed journals instead of open ones.

Using Bayesian Methods instead of Frequentist stats

The standard statistical method of hypothesis testing, or testing for statistical significance is fraught with problems. Many of these problems could be avoided if scientists switched to reporting "likelyhoods" instead of p-values. [See the arbital page on this topic.]

But again, for any individual scientist, using a non-standard statistical methodology that is unfamiliar to others in their field is damaging to their career prospects: their papers are less likely to be accepted by journals and less likely to be cited. So no individual scientist benefits from unilaterally switching, even if the any given field would benefit if everyone switched.


Do you guys have more?


[Edit 6:45 PM]: 
To be counted, every entry should include:

  1. How things are currently, and why that's bad.
  2. How they could be instead, and why that's better.
  3. What's blocking the transition from 1 to 2.

It's not sufficient to point a place where things are a mess, if there isn't a clear, stable, alternative. And it isn't sufficient to point out a way that things could be better without an explanation for why the change hasn't already taken place.

New Answer
New Comment

18 Answers sorted by

  1. The FDA (and to a lesser extent regulatory agencies generally) being extremely over-reluctant to approve things, because of the misaligned incentive that heavily punishes approving something that ends up being bad, but doesn't generally punish failing to approve something that would have been good. For the greater public good, individuals within the organization would have to take on substantially more personal risk, with little to no corresponding personal gain.
  2. Much lumber and other treated wood is treated with formaldehyde, a carcinogen, which then vaporizes back out of the wood over time, causing a health risk. The EPA regulates the allowed quantity of formaldehyde emission, but the limits, while they prevent acute poisoning, leave a substantial carcinogenic effect. The costs (an increased level of cancer in some people, likely dozens of years later) are diffuse and hard to notice, so there is no great incentive for the EPA to reduce the requirement, and formaldehyde is a useful treatment, so the companies themselves have no incentive to unilaterally reduce their use of formaldehyde. Thus, people who work with wood and those in newly constructed wooden structures bear an increased risk of cancer.
  3. Schools starting at times much earlier than the window of wakefulness for teenagers. It's likely that the knowledge diffusion problem has some impact (I do not have good information about how many school administrators don't know the science, vs how many are just ignoring it), but even with appropriate knowledge diffusion, a degree of incentive misalignment remains, where the people who suffer the most (the students) have nearly no power to enact the policies, while the adult administrators, with their earlier circadian rhythm, bear no personal cost from the schedule. Likely the combination of both leads to the current state of affairs persisting, as there is some incentive for the administrators to use better schedules, as students would be able to perform better on tests when properly awake.
  4. Ongoing overfishing of ocean fish. Each individual fishery (and, at a higher level, each country) would prefer a world where everyone fishes a sustainable amount, rather than overfishing and crashing the fish populations that they all rely upon, but without a centralized enforcement mechanism, they have no way of ensuring that the other fisheries (or countries) go along with them in cutting back on fishing, so unilaterally doing so would simply make them get out-competed by others.
  5. The lemon problem. Someone is selling their used car, and they know that it works fine, but they have no real way to cheaply, credibly give that information to prospective buyers. Pretty much any assurances that the seller could give the buyer could also be given by someone with a lemon, as long as that lemon works at least some of the time, so the information can't reliably propagate, and the used-car market remain inefficient, with the uncertainty effectively being a tax on all transactions.
  6. Countries building up their militaries. Most of the use of sizable militiaries is fighting against other militaries (and as a deterrent against such), so they are overall a negative-sum game. If countries all agreed to cut back their militaries, they would (for the most part) all benefit, but due to the competitive nature, there is a strong incentive to not cut back. Even if a cutting-back deal were arranged, there would be a strong incentive to merely produce the appearance of cutting back, while maintaining as much capacity as possible, as doing so would lead to an advantage against their newly-weakened rivals. Thus, the combination of the multipolar nature (hindering cooperation in the first place) and the difficulty of credibly demilitarizing puts an upward pressure on military expenditure.
  7. The "race to the bottom" problem. Using companies producing widgets as an example, each company might wish to fairly pay their workers, maintain a safe work environment, and not pollute the environment. However, other companies can gain an edge by sacrificing things in favor of producing more widgets (e.g. hiring more workers at cheaper wages). Thus, the principled company must make similar changes, or get outcompeted. This can continue until the companies have all sacrificed everything they can in favor of more productivity, even if all of them would have preferred to peacefully coexist with comfortable work conditions.
  8. Doctors being overly cautious in treatment. Similarly to the first example, the incentives punish positive mistakes much more heavily than negative ones. In this case, any deviation from what is considered to be the "proper" way of dealing with a case subjects the doctor to risk of being sued for malpractice in a way that sticking to the "proper" method does not, even if the deviation would have been a net positive in expectation for the patient. On a similar note, severe problems as a result of treatment gain much more than proportional negative attention relative to minor problems, so doctors have an incentive to avoid even small chances of severe negative results, in the process causing much larger amounts of harm to patients through large amounts of small harms (e.g., the doctor's more at risk if one patient gets a stroke than if a thousand get headaches).

EDIT: For some of these, the end-state is not especially stable by itself. In those cases, central enforcement would likely be the most realistic way to stabilize it.

A broad range of examples, lots of variety! That's perfect for gesturing at the overarching idea, lest we blow up conversations focusing on specifics in a narrow set.

3. pretty sure the biggest blocker to changing school times is parents' work schedules.  If the schedules diverge too much parents would have a harder time providing transportation to school and mandating adult supervision at all times.
6. it's not entirely about relative armament. there is base benefit to being able to get rid of neighbors you don't like or who are sitting on top of resources you want, independent of whether they pose any military threat to you.

Parents' work schedules are relevant at elementary school, but not at high school, yet high schools keep the same schedule (as far as I know; probably depends on country and school).
Another few reasons that I've heard for what's opposing later high school start times are 1) due to limited numbers of buses, doing high school later would require the lower schools to be earlier, and parents don't want their elementary schoolers out before sunrise, and 2) after-school activities like sports would be disrupted, both in an absolute sense (they already sometimes run pretty close to sunset) and a relative sense (a school that moved to a later schedule would either not be able to do sports games with other schools, or would have to have the athletes miss much more school than they already do in order to match the schedules of the other schools). To be clear, I do still think that the cost-benefit is clearly in favor of later starting.

Apparently (link to tweet), most countries have straightforward automated systems for tax collection, requiring minimal user input. Obviously, this setup saves everyone the pain of filling out confusing tax forms.

But the US has confusing tax forms because intuit (the company behind turbo tax), successfully lobbies to keep the current system in place, so that they can charge people for the service of helping them fill out their confusing taxes.

It's true. We usually have a house tax night, and one time this fell when a Latvian guest was present, and we were all crying and tearing our hair out and drinking heavily, and he was like "wtf is this? in Latvia they just send you a bill"

I've heard rumors (and I think a planet money episode) that it's not just them, but also small-government conservatives who think taxes will be easier to raise (less opposition) if they're easier to file (people are less aware of them).

Here's transcript of the episode This focuses on a pilot attempt to introduce this for California state taxes. Some highlights * In the 2004 tax season, they sent ReadyReturns out to about 11,000 taxpayers along with this little survey. * 99 percent of the people said that they would like to use it next year And indeed * Norquist thought ReadyReturn is tantamount to a tax hike because * Norquist put out the word to California Republicans - if you back ReadyReturn, you have broken the pledge [to not rise taxes] * "so and so - I've even forgotten the legislator's name - is not with us. It looks like we're going to be one vote short."

There are two kinds of problems with filling out tax forms:

One is understanding the meaning of inputs, like if they say "income of type X goes here, income of type Y goes there", you need to know whether your income was of type X or Y.

Another is following the processing instructions, such as "line 20 is line 17 plus line 19, assuming that line 19 is not greater than $500 + 0.15 × line 18, in which case line 20 is zero".

In Slovakia, we recently have an online system, where you enter the data, the processing instructions are automatically validated, if all v... (read more)

the bottom 50% of taxpayers also contribute 3% of tax receipts. So after accounting for frictional costs it is almost certainly a deadweight loss.

People being held against their will and drugged at long-stay psychiatric hospitals. Psychiatric hospitals profit from keeping their patients looked up instead of them getting their freedom back. Drugs given to calm the patients can make it easier to interact with the patient for the hospital even when the drugs aren't increasing the likelihood of recovery for the patients.

Whichever private or public organisation is paying is not incentivised to pay for long stays.

It's not like there's a different viable choice, is there.

Prediction-based medicine would allow psychiatric hospitals to compete on actually provide good treatments. You could have experts at multiple mental hospitals give predictions for the various outcomes of their treatment before admitting a new patient and then let the hospital with the best treatment prognosis take the patient. If you want to take a more conventional approach if you consider the person in question to be unable to consent because of mental illness you could make it an option for their spouse/family to overrule doctors the way you allow them to make decisions for other medical decisions unless a court decided about admitting the person to a mental institution. 
1Mary Chernyshenko3y
Sounds kinda utopian even in a not-the-least-convenient-world.

We're still teaching children arithmetic in base ten. The objectively correct base is six.

I'm serious. This is not just my opinion. The base we choose had better be a whole number, but the size of the multiplication table we have to memorize goes up roughly quadradically in the size of our base (although there are regularities that can sometimes make this easier), so it has to be a fairly small whole number. There just aren't that many choices, and we've checked them all. Base six is optimal.

  • The smallest four primes {2, 3, 5, 7} are either divisors or nearest-neighbors of six, which give the system a lot of nice properties, like easy divisibility tests.
  • The single-digit addition and multiplication tables are much smaller. The 0's and 1's are trivial, so that leaves two 4x4 tables, but because these operations are commutative, there are only ten "facts" to memorize for each table, instead of the 36 facts per table required for decimal.
  • More small-denominator fractions have a short or short-repeating seximal representation than in any other small base.
  • You can easily count up to five-x five on your fingers, instead of just to ten.

So why aren't we all using seximal? The decimal system is, to put it mildly, extremely entrenched. We developed it before we knew better. Path dependency. Almost all of our books and devices show numbers in decimal. Our money is decimal. All of our metric prefixes are based on powers of ten. Everybody who's learned arithmetic already did so in decimal. If you want to talk to others about numbers, you pretty much have to do it in decimal. We've calibrated our numeric intuitions in decimal. Learning decimal is easier than learning decimal and seximal, so almost nobody learns seximal. Having two different bases in wide use seems like it would make us worse off, but that's what we'd have to go through, since there's no emperor of the world to accomplish it by fiat.

If we wrote numbers in two different bases for a while, we'd have to distinguish them somehow, but if we only tag the seximal, then it feels second-class. I think we'd be better off writing seximal using completely different symbols to minimize interference, like ٤٢ instead of or 0s42.

This video makes a compelling case for base two as optimal (not just for computers, but for humans) which I had dismissed out of hand as unworkable due to the number of digits required. The more compact notation with digit groupings demonstrated gives binary all the advantages of quarternary, octal, or hexadecimal, while binary's extreme simplicity makes many manual calculation algorithms much easier than even seximal. I'm not convinced the spoken variant is adequate, but perhaps it could be improved upon.

And while we're at it, I think that when we teach arithmetic, we should be representing negative numbers using compliments like computers do. It makes negation a little more difficult, but still pretty easy, and eliminates subtraction as a distinct concept from negation and addition, along with the entire subtraction table.

So, for example, a -2 is like saying . We're used to having a weird unnecessary special case when going below zero, but if we mechanically use the normal subtraction rules, realizing that there are an infinite number of implicit leadi... (read more)

This seems super annoying when you start dealing with more abstract math: while it's plausibly more intuitive as a transition into finite fields (thinking specifically of quadratic residues, for example), it would really really suck for graphing, functions, calculus, or any sort of coefficent-based work. It also sounds tremendously annoying for conceptualizing bases/field-adjoins/sigma notation.
Maybe it's not better. I could be wrong. My opinion is weakly held. But I'm talking about eliminating the arithmetic of subtraction, not eliminating the algebra of negation. You'd still have a minus sign you can do algebra with, but it would be strictly unary. I don't see high-school algebra changing very much with that. We'd have some more compact notation to represent the ...¯¯¯9, maybe I'll use ^ for now. So you can still write −2 for algebra, it just simplifies to ^8 for when you need to do arithmetic on it. And instead of writing x−y, you write x+−y. Algebra seems about the same to me. Maybe a little easier since we lost a superfluous non-commutative operator. In base six, in complement form, the ^ now represents ...¯¯¯5, so a number line would look like ... ^43 ^44 ^45 ^0 ^1 ^2 ^3 ^4 ^5 0 1 2 3 4 5 10 ... i.e. all the numbers increment forwards instead of flipping at zero. You can plug these x-values into a y= formula for graphing, and it would seem to work the same. Multiplication still works on complements. Computers do integer multiplies using two's complement binary numbers. Maybe a concrete example of why you think graphing would be more difficult would help me understand where you're coming from.
Sorry for the lack of clarity-I’m not talking about high school algebra, I’m talking about abstract algebra. I guess if we’re writing -2 as a simplification, that’s fine, but seems to introduce a kind of meaningless extra step-I don’t quite understand the “special cases” you’re talking about, because it seems to me that you can eliminate subtraction without doing this? In fact, for anything more abstract than calculus, that’s standard-groups, for example, don’t have subtraction defined (usually) other than as the addition of the inverse.
...¯¯¯97 is the simplified form of −3, not the other way around, in the same sense that 0.¯¯¯3... is the simplified form of 3−1. Why do we think it's consistent that we can express a multiplicative inverse without an operator, but we can't do the same for an additive inverse? A number system with compliments can express a negative number on its own rather than requiring you to express it in terms of a positive number and an inversion operator, but you still need the operator for other reasons. ^8 seems no more superfluous of an additive inverse of 2 than 0.5 is as its multiplicative inverse. Either both are superfluous, or neither is. That was kind of my point, as far as the algebra is concerned—subtraction, fundamentally, is a negate and add, not a primitive. But I was talking about children doing arithmetic, and they can do it the same way. Teach them how to do negation (using complements, not tacking on a sign) instead of subtraction, and you're done. You never have to memorize the subtraction table.
Ok, I think I understand our crux here. In the fields of math I’m talking about, 3^(-1) is a far better way to express the multiplicative inverse of 3, simply because it’s not dependent on any specific representation scheme and immediately carries the relevant meaning. I don’t know enough about the pedagogy of elementary school math to opine on that.

First past the post voting often leads to undesirable outcomes, as explained by Jameson Quinn in his voting theory primer. There are several newly designed voting methods which are likely to be improvements over the current system, but most have seen limited, if any, uptake.

Some factors blocking such a transition include:

  • it's difficult to change political systems from the outside (e.g. few and infrequent opportunities to place referendum questions on election ballots)
  • within a two-party system, both benefit from first past the post voting, as they know they have a ~50% chance of winning each election, so there is no incentive for them to change from within
  • proponents of voting reform have not yet been able to coordinate on which method they recommend (i.e. establish a Schelling point)
  • individual voters will have to be able to intuitively understand how the new method works and why it should lead to better outcomes for them to consider supporting it

I'm sure there are more that others might think of.

within a two-party system, both benefit from first past the post voting, as they know they have a ~50% chance of winning each election, so there is no incentive for them to change from within

Some politicians inside major parties benefit from first-past-the-post voting, other don't. Centrist politicians who have to fear primary challenges from more extremist members of their party could benefit from a switch to approval voting that disempowers the extremists of their party. A politican like Joe Liberman who had to run outside of the Democrat party after loosing a primary would be more secure under approval voting in his reelection.

Agree that some individual moderate politicians could benefit from a change in voting methods (Romney is another that comes to mind). But I don't see how they could convince the majority of their party to support changes that go against their personal best interests.
A majority of the Democratic congressman are corporate Democrats and those have to fear getting primaried by people like Alexandria Ocasio-Cortez. This is especially the case as the power of far-left activsts to primary corporate Democrats rises. At the same time people like Alexandria Ocasio-Cortez would likely have a hard time publically openly arguing that far-left ideas that wouldn't win under approval voting should win out.
But each individual "corporate Democrat" congressperson also has huge incentives to cooperate with the rest of the party to maintain the current system. Their party has a ~50% chance of winning each election, and incumbents have a substantial advantage when facing re-election. Any individual (or small group) who proposes to defect in favor of a new, untested voting system would likely face a backlash for not being "true Democrats" and for making it more likely that non-Democrats would win more elections going forward. Also, I notice you mentioned approval voting several times as an alternative voting method. Is the voting theory community unified behind approval voting at this point? I feel 3-2-1 voting has a greater likelihood of becoming widely supported by the general population, for reasons I outlined in this comment.
Congressional seats are not given out by the rest of the party. The decision for them is made by the primary voters in a given  I think "there's too much polarization between the political parties" and it would be good to have a voting system isn't a position for which you are likely going to be attacked by people on the fringes who don't like politics as usual.  You have centrist Democratic strategists who worry that Democrats don't win certain elections because the primary voters chose a candidate that's too extreme to win the general election. A lot has happened since that discussion. OpenPhil funds the The Center for Election Science which managed to run a successful referenda to get Approval Voting into Fargo, ND.  While 3-2-1 voting might be easier then ranked choice voting Approval Voting is even easier to understand (there are not multiple rounds). 
Hmm. I feel like we are talking past each other to some extent, or not using terms the same way due to inferential differences, or something like that. Sorry if I've been unclear, I'll try to explain my position better. Very true. I was thinking more of the question, "Which actual senator or group of senators would propose an actual bill to effect change to a new voting system?" I still don't see how any have an incentive to do that. Re polarization: I also think there's too much polarization, which is why I support any efforts to get away from FPTP. I agree approval voting is a huge improvement over FPTP, and I'm glad Fargo has adopted it. I don't think getting people to understand [how the rules work] is the key variable to optimize for here. I think it's getting them to understand [why the heck changing to this system will make any difference] in an intuitive way. I think that if I explained approval voting to my friends, they would understand how the rules work in ten seconds. But when they ask why we should switch, I would have to say something that boils down to "well a bunch of nerds ran some simulations, and this one worked better" and they would pretend to fall asleep. I think that if I explained 3-2-1 voting to my friends, they would understand how the rules work in twenty seconds. And when they ask why we should switch, there is the obvious reason that the candidate they most fear winning the election gets knocked out in round 2. Democrats will be relieved that Trump would never win under such a system, and Republicans will be relieved that Bernie/AOC types would never win either. And if it's established as a clear anti-both-party or anti-the-current-two-party-system alternative, it may have a better chance of gaining support from voters from a wide variety of backgrounds.
I think the main issue with that is that senators don't vote on bills that decide how senators get elected. It's state laws that decide how the senators of a given state are elected (and whether they are elected at all or appointed). Change would more likely to happen by a senator saying: "Our state is grid-looked because we don't have enough bipartisanship. We should change our voting system in our state to allow for more bipartisanship."  You likely need some grassroots movement for voting reform for a politician to make such a move and be able to portray it as being a noble reformer but it seems possible to me to happen without the politician acting against their interests.  Bernie and AOC are very different here. Bernie wins election where he is on a third-party ticket (he got 67.44% as an independent). AOC on the other hand got on the Justice Democrats board which is an organization that promized to support primary challenges of Democrats by candidates that aren't in the board of it. Then Justice Democrats poored all their resources into getting AOC to win her primary challenge and unseat a powerful Democrat.  
Agree with how senators are elected. I was thinking of senators proposing a bill to change how federal elections are held, especially for President. As I understand it (90% confident?) such a change would require a constitutional amendment ratified by Congress and the states. Most of my considerations have been around Presidential elections, since they are the focal point of the US political system. As you mention, state laws decide how their senators are elected and it's quite likely that some states (preceded by cities such as Fargo) will change to an alternate voting method before the federal government does. This provides an additional incentive for individual senators to remain unattached to any alternate voting method proposals, as they will be able to watch how public opinion reacts to the experiments on state and local levels. Or a newcomer. Is there a popular grassroots movement taking place right now? For a newcomer, being a noble reformer, and actually being a noble reformer, could squarely align with their interests. Re Bernie/AOC: I was again referring to the chance of them being elected President, which is what I think drives most people's fears. It could make sense they would remain popular locally and maintain their current seats in Congress. I was thinking more of a typical Republican in a red state, who probably doesn't have to worry about losing their own congressional seats, but would be very worried about losing the Presidency. And especially worried that a future Democrat nominee would be more extreme than the current President. Many can probably live with weird Bernie up there in Vermont, at least he's different for a change. Since I'm getting it all out, a couple more advantages of 3-2-1 voting: Assuming people understand how 3-2-1 voting works, they automatically understand how approval voting works, since it's just the first half of 3-2-1. So then you can say things like, "If you understand this one voting method, you'll understand ev
Practically, that means not caring about moves to get alternative voting systems adopted.  From a principled reason it's stupid to experiment with new voting systems at the place that's most central instead of other less central places. Practically, senators and congressman from places with an alternative voting system can reasonable make proposals like "Our voting system is much better, everyone should adopt it" without getting punished as being unserious.  If you both have established politicians speaking in favor of changing the voting system and a bunch of grassroots reformers it's easier for the proposal to succeed in a state.  Most people's fear is a different topic then what's good politically for senators and congressmen. Senators and congressmen fear getting primaried.  A typical Republican in a red state has to fear getting successfully primaried by some extreme tea party candidate or Qanon believer under the current system. On the other hand they are unlikely to lose their seat to a Democrat in either the current system or approval voting (and 3-2-1). Approval voting (and 3-2-1) make it less likely that a tea party candidate or Qanon believer manages to unseat them.
Sorry, again I realize I didn't explain some of my thoughts clearly enough. I think we are discussing two different but related questions: 1. How do we convince the average voter to support alternate voting systems, vs. 2. How do we convince senators, state and local governments, local political activists, etc. to support alternate voting systems, get them on the ballot, and ultimately passed into law. Most of my thinking and comments in this thread have been more related to question 1, but it feels like you interpreted some as related to question 2. Both are important, but I think the incentives and strategies required are different for the two questions. When I said Presidential elections are the focal point of the political system, I was thinking about how I would try to convince the average voter to support alternate voting systems. In such a conversation, I know I have very limited time to make my argument, and my conversation partner is likely predisposed to doubt me, since I'm sending clear signals that I'm not "on their team" (whichever "team" they are on). Therefore I need to be able to explain very quickly how the alternate system would work and how it would improve outcomes. Since President is the most important office in US politics and most average voters have a decent understanding of how it currently gets elected, my strategy would be to use Presidential elections as an example, and point out how 3-2-1 can help prevent scary extremists from getting elected as President. I think ranked choice voting failed to capture much public support (as seen by its failure in recent state ballot questions) because it's too complicated to explain quickly, and too hard for the average voter to quickly understand how it would improve outcomes. 3-2-1 is substantially better by both measures. Switching over to question 2 now, I agree it would be foolish to start by trying to change federal elections before state and local elections. And I agree that once an alternate
One last comment/reminder to myself: I read nostalgebrist's summary of Weyl's "Why I am not a technocrat" argument (haven't read the original yet), and his last few points seem very relevant to my argument:

anoter big point is that people who got voted and benefited from first past the post have incentive not to change it

Just to point out that there are consequentialist arguments for first past the post, namely that it leads to stable majority governments and a small number of well-defined political parties.

I don't think that's the case. The US doesn't have well-defined parties but different wings of the same party that frequently vote differently on issues in parliament. That's different then German parties where usually the party first finds an internal position and then uniformly votes according to that position.  That happens because in the US a politican only needs the support of the constituents in his constiency and donor support (which can be gotten independently of party infrastracture even if the party sometimes helps with that). On the other hand German polictians need the support of their party beyond their voting district to get a place on the party list.  The US is politically a very gridlocked country.

Passive investors should want companies they hold to not engage in zero sum games with each other (like advertising), but in practice best corporate governance practices spreading is often limited by the bandwidth of shareholder activists.

Make More Land.

—American policing. There's tons of obvious problems with this, and the proposed solutions range from obvious and nearly universally agreed upon (reduce SWAT-style raids by >90%, end/drastically reform qualified immunity, abolish civil asset forfeiture, etc) to speculative and contentious. Such reforms generally don't match the incentives of the police themselves, and civil oversight is currently too diffuse and ineffective to impose reform. National authorities are incapable of any reform, and while a few state and cities have made small changes in the past year, the core issues will not be addressed anywhere in the near future.

—COVID vaccine distribution. Our society can reach better equilibria, as shown by the normal flu vaccine distribution and by the occasional stories where a freezer breaks down and all the shots are administered before they expire. Nevertheless our current system is a mess, no one feels they have responsibility for the mess, and there is no sign it will be fixed before the pandemic ends.

—Whatever made development of the F-35 fighter plane so much worse than the development of previous fighter planes.

—There are several cases in America (and presumably elsewhere, although I'm less familiar) where a relatively small, coordinated group successfully lobbies for laws which benefit themselves by imposing small costs on the country as a whole. Because the costs are so widely distributed, no one has a strong incentive to overturn these, and lawmakers have a moderate incentive to cooperate with the special interest group. The canonical examples include the artificial complexity of the tax code held in place by tax-preparation firms; ethanol subsidies supported by corn growers; and comically long copyright for artistic works, supported by media conglomerates (IIRC these expire 70 years after the death of the author!).

Hypnosis + anesthetics is likely better then just anesthetics when doing surgery. It means you need less anesthetics and there's decent evidence that it provides better wound healing. 

As a paper in a reputable anesthetics journal suggests:

Faster wound healing, earlier postoperative gastrointestinal recovery, and less nausea have been reported when hypnosis or positive suggestions were part of the perioperative management 

I have a friend who I knew from having attended hypnosis seminars together who worked as an anesthetic in a hospital who still didn't use hypnosis. It would have meant spending 15-30 minutes more per patient and the hospital was organized in a way to pressure him to do anesthetics as fast as possible. 

Instead of being able to do it, his job demoralized him enough to quit never wanting to work in a hospital again. 

The hospital doesn't profit from the faster wound healing enough to do the nonstandard thing of using hypnosis even when it has skilled personal. 

Scott mentioned another one recently:

I’ve had Uber drivers who were doctors back in their home countries. They’re stuck driving Ubers because the US insists that nobody - not British doctors, not Canadian doctors, nobody - can work in the US unless they repeat their last four years of training in an American hospital. American hospitals don’t have many training positions, and reserve the ones they do have for American med students, so most of them can’t do it. Meanwhile, my patients complain of having to wait months to get a desperately-needed appointment, and the hospitals just say “well, there aren’t enough doctors”. There are, they’re just all driving Ubers because of regulatory failure!

Drug expiration dates are probably far too conservative in most cases, with strict regulations forcing hospitals to discard medication that is likely still effective, and to go to the expense of replacing it, which makes medical care that much more expensive than necessary, and worse, they are still required to do so even during a shortage when replacements are not available, potentially costing lives.

The FDA allows manufacturers to extend expiration dates with further studies, but big pharma has little incentive to do so. Studies are expensive, and shorter expiration dates mean they can sell more drugs.

Stockpiles of medicines are important to the military, and the government also maintains stockpiles for emergency civilian use, but keeping these up to date given short expiration times is very expensive, so at the behest of the Department of Defense, the FDA undertook the Shelf Life Extension Program, which proved that many medications remain effective for years past their expiration date when properly stored, saving the government a lot of money on maintaining stockpiles. But hospitals are still required to throw these away.

One wonders why the expiration dates were set so short to begin with, or not extended automatically through continued testing.

Sell medical care based on the calibrated likelihood of achieving the wished for medical outcomes.

This would allow treatments that can't be easily standardized (and thus studied with controlled studies) be brought based on their impact.

We should be teaching trigonometry and the unit circle in terms of tau instead of pi, that is, since the mathematically natural angle measure is radians (not diameters), we should express radians in terms of the ratio of the radius to the circumference, instead of the ratio of the diameter to the circumference, which adds a spurious factor of 2 everywhere.

The choice of pi in the first place was due to path dependency, and is now extremely entrenched. (Nearly?) all of our textbooks use pi and all our school graphing calculators have a pi button. Python, however, is more enlightened.

One is blame games. Many problems don’t have a clear cause. Pointing the finger creates stress and defensiveness that’s counterproductive to solving the problem and emotionally harmful, and also doesn’t help identify how to avoid similar problems in the future. It would be better not to even start. But the fact that each person anticipates being on the receiving end of a lot of unfair blame prevents them from not engaging.

From a utilitarian standpoint doing human challenge trials for the covid vaccines would have been preferable to ecological / uncontrolled trials (for instance, we would have clearer data, today about the optimal dosage, which translates into maximizing the number of successful vaccinations from a given batch of the vaccine). 

My current understanding for why the US didn't do human challenge trials comes down to the incentives of the decision makers: they are likely to experience some backlash, especially if something goes wrong, but they don't gain accolades or other rewards for the massive upside.

1: Nuclear Power Currently we (the developed world except France) rely mostly on fossil fuels for energy with some usage of renewables. This has been the case since WW2. This is bad. It's bad because global warming and other kinds of pollution. It's bad because we rely either on coal, which is extra dirty, or oil and natural gas which largely comes from brutal autocracies and causes us to be needlessly involved in wars and influence battles. Renewables are coming but they're too inconsistent to be useful. until we have better energy storage tech.

None of this was necessary. We've had nuclear energy since the 1960's. It's cheap, safe and essentially limitless. France generates 70% of its energy from nuclear power and has done so for decades. As a consequence it has energy independence, low emissions, low energy costs and is generally better off. The only reason it's not mainstream in most nations is due to environmentalist movements campaigning against it and high profile nucler accidents leading to a public misconception of the risks of nuclear as compared to other power sources much the same way as people are more scared of terrorism than car crashes despite the latter being overwhelmingly more dangerous.

2: Immigration Canada and Australia do immigration well. They have points system. High skilled, probably beneficial to society immigrants can easily migrate with minimum paperwork.

In the Uk and USA, immigration is hell. Low skilled illegal migrants number in the millions. High skilled legal migrants have to fight through regulatory hurdle after hurdle to get into the country. This is bad.

It's not clear why immigration is so bad. Massive majorities of the public support points based migration systems. Nevertheless we have terrible systems which tolerate massive illegal migration but make legal migration unnecessarily difficult.

3: Corruption Private individuals and businesses can give money to politicians and parties. They usually do this because it's in their interest. This is colloquially known as corruption. There is massive support for ending this kind of practice. Almost everyone from elites to the public see it as bad. Still year after year it persists. In the US this may be because of the law imposing a barrier to reducing campaign contributions. In the UK or other european countries there's no such problem yet still nothing is done.

4: Education A lot of education is signalling. This is even more true of lower tier education. One memory that sticks with me is on my first day of a graduate schema another person telling my they studied airport management. I asked them why. They said just to get a job and that they'd learn nothing practical over those years.

The problem with education is a collective action/externality one. Education is good for individuals but imposes costs on others by inflating the level of signalling needed. Many people would be better off if they collectively decided to not waste years on useless degrees so they can apply to unrelated graduate schemes. Still, it's not possible to cordinate.

This is a thing I've heard unofficially from a guy in the industry, but I'm not sure how big of a problem it basically is. I mean, I think it's a big deal, but I would not care to quantify it.

In Ukraine, we have a separate branch of research institutions dedicated to agriculture, forestry, fishery etc. I mean, we've got universities where there's some science being done; the Academy, where most of it happens; and several "applied" branches like the above-mentioned Academy of Agrarian Sciences. The "applied" people are generally not well-integrated into the whole community, which is a problem in itself. One day I took part in an agrarian conference (I was looking for possible collaborations), and this guy showed me a sample of wheat they were working on. He said they have to keep producing new varieties, ever better varieties. Which means in practice that they don't actually go all out to breed superwheat, they just keep making it incrementally super.

And I just hate this - local culture? Official requirements? I mean, if only they were actually acknowledged as researchers by the overall community (heck, if they were acknowledged as researchers by their own authorities). People would be working, ridiculous expectations would be laughed at, things would be moving along!

Because when they say Lysenko had left a legacy, this is what it looks like.

dating on dating apps (I think) men sends messages or tries to match with as many people as possible to increase their chances of talking to someone. women receives a lot of messages/potential matches so they get more selective. which in turns creates less matches per message sent for men, which reacts by sending more messages

this is from what I read and from my experience and experience of people I talked to. also, it looks like there is still a big heterosexual dynamics in this case (this might not apply to all dating apps but I guess it must apply to the big ones). I stayed a big vague in the description since I have seen different reasons for every of those points (that are not necessarily contradicting each other though: they could be additive)

An EY category #3 example from emergency medical services (EMS) in the USA

>How things are currently, and why that's bad.
When someone tells a public safety access point (PSAP) that there might be a problem, the dispatcher generally sends an ambulance at a minimum. Then, the city or ambulance provider takes the nearest obvious patient(s) to the emergency room. The resulting EMS bill is often extremely high as well as the ER bill. Sometimes, the providers are not in-network and the patient must pay in full even if there was not a true emergency.

>How they could be instead, and why that's better.
The PSAP could pre-commit to only sending certain types of units when a set of criteria are met. And, if none of the criteria are met for a public safety response, then the patient could be given self-care instructions ranging from watchful waiting to calling the patient's doctor. If a crew is sent to the scene, then the crew could send the patient to the most appropriate place (ranging from "stay home for a week" to "Uh oh, you really do need an emergency room."). This would result in greatly reduced spending by the PSAP, public safety agencies, ambulance providers, hospital, patient, and other healthcare-related entities. The patient would also have better outcomes due to decreased financial stress, decreased risk of iatrogenic infections, etc.

>What's blocking the transition from 1 to 2.
A public safety agency/ambulance provider is only paid when they transport a patient to the emergency room because that's what all the insurance companies have agreed to do.

If a hospital dares to tell an ambulance provider that they can send patients to another type of location, the hospital will lose money.

If an ambulance crew does not transport a patient to an emergency room, the crew takes on liability for practicing unconventional medicine.

Dispatchers are generally not permitted to send anything less than an ambulance and a fire engine. Sometimes, ambulances are only provided at the most costly level (ALS). The crew may even be required to go lights and sirens as a matter of policy despite the increased risk of death and injury. But if a city does not do a "full response", then it is on the hook if anyone complains about reduced responses (aka political fallout) and there may be legal liability for the reduced response. Dispatchers are also often not RN or MDs. So they may not be permitted to definitively diagnosis or recommend treatment outside of a very narrow set of pre-defined situations.

For what it's worth, Coronavirus disease 2019 has forced many cities to change their practices because there are not enough units. CMS has also had on-again-off-again efforts to send patients to alternate destinations (e.g. https://www.cms.gov/newsroom/fact-sheets/emergency-triage-treat-and-transport-et3-model) but many such programs eventually end and the local city must find alternative grant funding in the long-term. As usual, the necessary know-how has existed for decades (e.g. https://en.wikipedia.org/wiki/Medical_Priority_Dispatch_System#Response_Determinant).


April 25, 2021 Update: Today, there is a New York Times article talking about the closure of ambulance services in rural America. As the article says, the situation was entirely foreseeable (article reasonably claims that the direct proximate cause was lowered patient transport volumes from Coronavirus disease 2019).

12 comments, sorted by Click to highlight new comments since: Today at 7:29 PM

I think some of these examples are not real examples of "inadequate equilibria".  They are instead situations with real switching costs, or where there are conflicting beliefs or interests.

To illustrate, the authors example of moving from proprietary to open-source journals seems like a real example to me.  But their example of using Bayesian rather than frequentist statistical methods does not seem like a real example.

Note that I'm a Bayesian (though not a rabid one), and that I've taught introductory statistics.  There isn't some easy way to just switch to Bayesian methods.  First, students need to understand the scientific literature, and that includes the past scientific literature.  So for a considerable period of time, students will need to understand frequentist statistics. This is a legacy compatibility problem, not a coordination problem. Second, there are many scientists who don't know Bayesian statistics. This is a retraining problem (which is not cost-free), not a coordination problem. Third, not everyone agrees that Bayesian methods are better.  This is a persuasion problem, not a coordination problem. Fourth, Bayesian methods aren't always better - there really are problems where the correct Bayesian approach is much, much more difficult to carry out than a simple frequentist approach that is usually gives mostly-correct results. So it really is necessary for at least some people to understand frequentist statistics, though it would be good if the emphasis eventually changes in a Bayesian direction. There may be some coordination aspects to the current mess, but to a considerable extent the mess reflects real issues with real costs and benefits.

I didn't put a deadline for the bounty, but I'm now setting the deadline for June 1, 2021. I'm going to count up all the answers and reach out to winners some time that week, only counting answers that came in on or before that date.

Feel free to try and slip in last minute. 

There's the large (huge?) category of market failure in politics, both local and national. 

Let's take the simplest example: long term projects. When the private market wants to invest in a very long term project, there are plenty of financial instruments that allow them. For example you can plant a slow growth forest, which takes 20+ years to mature. Then in several years you can sell bonds, or just sell the forest itself at a value proportional to the time left to maturation - to somebody that either wants to wait, or more likely is just looking for a medium term investment and also intends to sell it in a few years for a moderate profit. 

When a local administration wants to invest in a long term project - and by "long term" I mean a project where the benefits will be seen and collected in a few years - there is just no incentive structure to do that, because the elected officials are there for something like 4 years, maybe 8. After that most of them intend to use the quick results shown locally to move to a bigger scene. There is just no way they can start something now and collect (political) benefits partway. Except by the expenditure itself, instead of the results, which creates even more perverse incentives. 

That's compensated for, perhaps over compensated for, by career civil servants who regard elected official as as an inconvenience to their long term goals

The problem also exists for career civil servants when they are moved around before the project completes. 

With Plan S and Project Deal there seems to be good movement on the Open Access front. It wouldn't surprise me if in ten years a majority of the research is published under open access.

Links? Or should I just google?

https://www.coalition-s.org/ and https://www.projekt-deal.de/about-deal/ . Reading the Wikipedia pages might also be useful. 

Just to clarify, what is allowed here.  I can think of tons of scenarios where there are probably better equilibria  (e.g. nuclear disarmament, no starving people, etc) in which clearly the current state is not optimal, and there's some other theoretical equilibria that's more optimal. 

At a minimum, there should be a good argument that the new equilibrium is stable in that no one can benefit by unilaterally defecting.

Yeah. The new equilibrium has to be realistic, and, as Measure says, stable.

Situations where there's almost certainly a better equilibria, but we don't know what it is, don't count.

Every entry should include:

  1. How things are currently, and why that's bad.
  2. How they could be instead, and why that's better.
  3. What's blocking the transition from 1 to 2.

I'm a bit confused about the edges of the inadequate equilbrium concept you're interested in.

In particular, do simple cases of negative externalities count? E.g. the econ 101 example of "factory pollutes river" - seems like an instance of (1) and (2) in Eliezer's taxonomy - depending on whether you're thinking of the "decision-maker" as (1) the factory owner (who would lose out personally) or (2) the government (who can't learn the information they need because the pollution is intentionally hidden). But this isn't what I'd typically think of as a bad Nash equilibrium, because (let's suppose) the factory owners wouldn't actually be better off by "cooperating"