Extracting Value from Inadequate Equilibria

by [anonymous]3 min read18th May 202012 comments

49

Assurance contractsCommitment MechanismsWorld OptimizationRationality
Frontpage

[Much expanded from my comment here. Pure speculation, but I’m confident that the bones of this make sense, even if it ends up being unrealistic in practice. Cross-posted from Grand, Unified, Crazy.]

A lot of problems are coordination problems. An easy example that comes to mind is scientific publishing: everybody knows that some journal publishers are charging ridiculous prices relative to what they actually provide, but those journals have momentum. It’s too costly for any individual scientist or university to buck the trend; what we need is coordinated action.

Eliezer Yudkowsky talks about these problems in his sequence Inadequate Equilibria, and proposes off-hand the idea of a Kickstarter for Coordinated Action. While Kickstarter is a great metaphor for understanding the basic principle of “timed-collective-action-threshold-conditional-commitment”, I think it’s ultimately led the discussion of this idea down a less fruitful path because Kickstarter is focused on individuals, and most high-value coordination problems happen at the level of institutions.

Consider journal publishing again. Certainly a sufficient mass of individual scientists could coordinate to switch publishers all at once. But no matter what individual scientists agree to, this is not a complete or perfect solution:

  • Almost no individual scientists are paying directly for these subscriptions – their universities are, often via long-term bulk contracts.
  • University hiring decisions involve people in the HR and finance departments of a university who have no interest in a coordinated “stop publishing in predatory journals” action. They only care about the prestige and credentials of the people they hire. Publications in those journals would still be a strong signal for them.
  • Tenure decisions involve more peer scientists than hiring, but would suffer at least partly from the same issue as hiring.

What’s needed for an action like this isn’t a Kickstarter-style website for scientists to sign up on – it’s coordinated action between universities at an institutional level. Many of the other examples discussed in Inadequate Equilibria fit the same pattern: the problems with healthcare in the U.S. aren’t caused by insufficient coordination between individual doctors, they’re caused by institutional coordination problems between hospitals, the FDA, and government.

(Speaking of government, there’s a whole host of other coordination problems [climate change comes to mind] that would be eminently more solvable if we had a good mechanism for coordinating the various institutions of government between countries. The United Nations is better than nothing, but doesn’t have enough trust or verification/enforcement power to be truly effective.)

---

The problem with the Kickstarter model is that institutions qua institutions are never going to sign up for an impersonal website and pledge $25 over a 60-day campaign to switch publishing models. The time scale is wrong, the monetary scale is wrong, the commitment level is wrong, the interface is wrong… that’s just not how institutions do business. Universities and hospitals prefer to do business via contracts, and lawyers, and board meetings. Luckily, there’s still value to be extracted here, which means that it should be possible to make a startup out of this anyway; it just won’t look anything like Kickstarter.

Our hypothetical business would employ a small cadre of lawyers, accountants, and domain experts. It would identify opportunities (e.g. journal publishing) and proactively approach the relevant institutions through the proper channels. These institutions would sign crafted, non-trivial contracts bound to the success of the endeavour. The business would provide fulfillment verification and all of the other necessary components, and would act as a trusted third-party. The existence of proper contracts custom-written by dedicated lawyers would let the existing legal system act as an enforcement mechanism. Since the successful execution of these contracts would provide each institution with significant long-term value, the business can fund itself over the long haul by taking a percentage of these savings off the top, just like Kickstarter.

This idea has a lot of obvious problems as well (the required upfront investment, the business implications of having its income depend on one or two major projects each year, the incentives it would have to manufacture problems, etc) but with a proper long-term-focused investor on board it seems like this could turn into something quite useful to humanity as a whole. Implementing it is well outside of my current skillset, but I would love to see what some well-funded entrepreneur with the right legal chops could make of something like this.

Thoughts?

49

12 comments, sorted by Highlighting new comments since Today at 10:03 PM
New Comment

Thanks for writing this. It feels like it's grappling with the right questions.

I think this still doesn't make for a good primary business model – I think basically each instance of such a contract would need to be very bespoke, and require lots of domain knowledge in a particular domain. (i.e. to solve a medical in-equilibria, you probably need both lawyers who are familiar with the relevant domains and hospital admins and doctors, and this is a lot of specialized personnel who probably wouldn't also be able to solve Academic Journalism, etc)

The open question (given my current knowledge) is to what degree the lawyers and business angle here would be transferable between domains, which could at least allow some economy of scale.

Offhand, this feels more like the domain of philanthropy than capitalism. It doesn't seem like it'd scale well, so it'd require someone who just Really Cares A Lot about fixing these problems to sink a lot of resources into it. (Alternately, you might have a firm where this isn't there primary business model, but they specialize in a lot of institutional deals, more generally, and then they subsidize Solving In-Equilibrias when they find a good opportunity)

Obviously I defer to lawyers who have a clearer idea of what range of expertise would be transferable here.

There's also a quote, which I don't remember the provenance of and can't quickly find, which was something like "the main purpose of think tanks is to generate ideas that are ready to be deployed in times of crisis." 

"Only a crisis—​actual or perceived—​produces real change. When that crisis occurs, the actions that are taken depend on the ideas that are lying around. That, I believe, is our basic function: to develop alternatives to existing policies, to keep them alive and available until the politically impossible becomes politically inevitable." - Milton Friedman, 1982 preface to Capitalism and Freedom

That's it, thanks!

There's a good chance that the last encounter with the quote (Axiomata spells it out) was my post https://www.lesswrong.com/posts/ZFpA3ZbKndRpNbpXg/in-defense-of-politics which quotes Friedman.

Upvoted for an important topic. But I think you miss a fundamental element of the equilibria under discussion. It's not _only_ a question of coordination. It's also a very hard to solve alignment drift. Many participants see the current situation as necessary to their well-being, and will actively fight to preserve it.

That's not a coordination problem, except in the greater sense that people with different timeframes or goals for each other need to be convinced that their approach to life is wrong.

[-][anonymous]2y 2
Many participants see the current situation as necessary to their well-being,

I'd be curious for an example of this. Obviously e.g. Elsevier doesn't want universities to solve their coordination problem, but they're not actually involved in any of the contracts which would be required to solve that one.

Other problems may be less tractable, but within the class of "Nash equilibria that aren’t even the best Nash equilibrium" I don't imagine you'd get much resistance to change?

An example would be that those who've previously had success in this publishing model, and now have some influence over future papers (because they're faculty and advisors, based partly on their prior publishing). They correctly believe their status will decline if a different measure of success takes root. Likewise those who get some of their worth by being referees for the limited "accredited" journals - they would be justified in worrying that their status would decline with a more open process.

Note that the optimal position for these conflicted participants is to publicly decry the system while privately supporting it, often by pointing out that there's no coordination mechanism. This leads to the appearance that coordination is the only barrier, because the real problem (conflicting desires) is not publicly acknowledged.

This situation repeats in MANY examples that at first look like just a coordination failure - there's a lot of overlap with stakeholders who benefit from (or perceive they do) the situation, and those who choose to continue it rather than defecting.

What’s needed for an action like this isn’t a Kickstarter-style website for scientists to sign up on – it’s coordinated action between universities at an institutional level.

It's worth noting that we have that in Germany with Project DEAL.

When I look at the US, I have a hard time imaging that it's easier for someone outside of the research system to get enough univerisites at one table then an organization like the Association of American Universities. If individual universities have a position that they want to cooperate they can raise a proposal through organizations like it.

Other industries have similar Associations that can allow coordination.

If you have an organization like what you propose it's also easy to attack. There are many laws against anticompetetive behavior that can likely be used and can be much easier used when attacking a for profit organization.

I wonder if some of the discussion in the posts related to Moral Mazes might offer any insights.