List of Mechanisms
- Jeremy Nixon
- Aug 13, 2024
- 7 min read
Updated: Nov 21, 2024
Mission: Create a list of implementation-ready mechanisms.
Previously:
Quality Standard:
Refine the mechanism to a point where it can be implemented by a team at the Mechanism Design Hackathons without ambiguity about what the mechanism needs to accomplish.
sha
List
Nonprofit Funding Parties
This mechanism addresses the free-rider problem (in the case of crowdfunding for personal gain), or coordination and amplification (in the case of altruism). Individual contributors request a match (e.g. 99x) or their money back. The program fills these requests in parties. Additional UI information is possible (e.g. “up to 99x match - current match: 37x”).
Alpha Leaks
A lot of value is lost to information asymmetry in the investment markets. This is because there is a strong disincentive to share investment theses. In this project, a writer gets a credit line from readers who are convinced by the thesis enough to offer a high-yield loan specifically to buy the mentioned ticker (this earmark, and a limit price, is enforced). The alignment here is that the author must have skin-in-the-game. In particular, the reader-creditors can only go bust if the author goes bust; this is in contrast to readers following financial authors into pump-and-dumps.
Generalizing Matching Markets (cf https://en.wikipedia.org/wiki/National_Resident_Matching_Program)
This is a subset of market design. Can the ideas behind residency matching, kidney exchange, etc be applied to other domains like grantmaking, hiring, and collaboration?
Coordinated coming out of the closet
Based on Cass Sunstein’s “social cascade” theory of change. People privately commit to a socially risky belief, and a threshold X such that if at least X other people share the belief, it will be revealed. This may help surface valuable contrarian perspectives. There may be a cryptographic solution for this, I’m unsure.
Liquid, scalable Webs of Trust:
Fast ways of curating and selecting lists of experts you trust in specific domains
Useful for better voting (liquid democracy), peer review (expert review of papers and proposals), product suggestions (restaurant recommendations)
Can bootstrap off existing social graphs and lists (e.g. Twitter following / lists, Wikipedia lists, etc) to help resolve cold-start problem
Need to figure out ways of adding new entrants into the Web (similar to the challenges of starting a new Twitter with no following account)
Surprisingly popular information elicitation from crowds
Based on Nature paper, “A solution to the single-question crowd wisdom problem”
“Here we propose the following alternative to a democratic vote: select the answer that is more popular than people predict. We show that this principle yields the best answer under reasonable assumptions about voter behaviour, while the standard ‘most popular’ or ‘most confident’ principles fail under exactly those same assumptions.”
Example:
Is Philadelphia the capital of Pennsylvania, yes or no?
Is Philadelphia the capital of Pennsylvania?
Yes: 65%
No: 35%
What do you think most people will respond to that question?
Yes: 75%
No: 25%
The difference between the answers to the right question and the popular question:
Yes: 65% − 75% = −10%
No: 35% − 25% = 10%
Thus, the No answer is surprisingly popular (10% > −10%) – Harrisburg is the actual capital
{Risks to mitigate given adverse nature of most actors who need voting for zero sum outcomes. When can this break or be manipulated in a critical scenario? Social media ads to influence an election swaying the predictions? Gossip? Vote loading (a bad actor only needs to add a bunch of fake votes and it would sway the outcome with a small margin of extra votes vs a much harder to achieve majority.)}
A coordination mechanism where conditional contracts solve the company team building problem.
Have an institution designer post a company design which can be opted into by team members who would form the company conditional on other team members committing.
All team members must be above a trusted person’s quality bar. This person/people are named in the design.
Funding can also be allocated to the design conditional on the team committing. This allows for design specialization and solves the main coordination problem that typically requires the founder / CEO to fundraise and build the team.
{Risks to mitigate: The pagent problem -since it’s a contract and multiple may trigger at once, participants would not want to risk being punished for contracts they are interested in that overlap. So participants will either opt to participate in fewer or just one contract, or wait until last minute, which could trigger odd or bad behaviors - aka an ambitious but less ethical project leader might lie saying they are close to subscribed to create fomo akin to VC fundraising and scoop talent for a less than ideal project.}
Necessary Features:
Log in / Out
A page for browsing / viewing designs
A page for design posting
Conditional contract for funding
Conditional contract for joining (to be checked by trusted hiring bar)
Code to check whether all conditions for fulfilling the contract are complete
Space to put money on which the contract relies / collect penalties for failure to comply with the contract.
Superforcasting focused research prediction system.
Find research problems that have already resolved that aren’t incredibly well known
Have researchers predict their outcomes
Rank researchers based on prediction ability
Superforcasting focused research prediction system with a mix of reputation and/or money:
Aggregate and organize many peer-to-peer bets to create a prediction-market like vehicle but with more explicit social sharing/reputation involved
Recent example of Holden Karnofsky bet with Zvi Moshowitz on Omicron: https://www.cold-takes.com/bet-with-zvi-about-omicron/
Followed by Applied Divinity Studies bet with Stephen Malina et al: https://applieddivinitystudies.com/zvi-holden-bet/
Build norms and products to let people challenge each other and bet publicly on their predictions, e.g. Anthony Fauci can go on Tucker Carlson and they can bet on COVID deaths in the next 6 months with odds that are defined by other social bets on the platform
Respectability Cascade Surfacing.
Perhaps a simple questionnaire which computes a ratio between how important and tractable people believe a research field is with how much fear / anxiety they feel about publicly posting their support for progress in that field.
Use the ideas with the most intense ratio as the raw material for a respectability cascade organization.
Many worthwhile ideas are bottlenecked by an unwarranted loss in respect that comes from working on them. Coordinate between popular scientific twitter / social media accounts to systematically make worthwhile ideas respectable.
OR, conditional contract to tweet about a controversial idea at a coordinated time if enough in-network others commit to tweet about the idea
Technical roadmap decision-making mechanism.
The Genesis optimal house auction code for envy free resource allocation can be made accessible.
We can also build high quality auction front ends as we aggregate and centralize open source auction implementations.
Implement equity for people.
Create a clean implementation that allows people to sell a percentage of their future earnings for money up front.
Meta-mechanism
Bounty for mechanism implementation which gives the creator of the bounty ownership/equity in the implemented mechanism.
Mechanism Design Library (Modular code for implementing mechanisms)
Payments system
Contracts system
Federated Services
federated web services: profit / recognition sharing across multiple organizations, or a linked service
federated IRL services:
Shared Truth Mechanisms
Two people answer questions about their beliefs and are shown the beliefs that they have in common.
Could run a heretecon like this: conversations between groups of small people that share the same taboo beliefs.
Socialized Truth Mechanisms
Motivation: From the earliest days of philosophy, it has been recognized that the socialization of knowledge is a critical process in refining that knowledge into truth. Classically, rhetorical discussions formed a large part in socialized epistemology. In judicial systems, adversarial discussions have formed a critical basis of adjudicating blame. In the modern era, peer-reviewed journals are often held as a standard of excellence, even though that mechanism may not cogently apply to all fields. More recently, “fact-checking” website have arisen to (ostensibly) centralize the burden of validating simple claims. Few would argue that the recent wave of fact-checkers has performed a stellar job. Yet their rise has certainly indicated this: that modern mechanisms for generating consensus around truth no longer sufficiently work.
Goals:
Theoretical Prompts:
What taxonomy of knowledge is most useful for thinking about socialized truth? What kinds of “truth” should be adjudicated in different ways?
What kinds of truth have the most straightforward adjudication? Are there any “quick wins” in this field?
Which kind of truth are the most complicated to adjudicate?
Is there any universal formulation which would work across all kinds of truth?
Practical Prompt (for the hackathon):
Pick on kind of truth claim about which there is uncertainty or disagreement. Design a mechanism which coordinates multiple people to reduce the uncertainty or disagreement, in a way reasonably expected to perform better than any individual doing their own research.
Modular Startup
e.g. something like:
each engineer makes a microservice (or multiple) and retains that IP
each sales person earns commission in perpetuity
some amount of overhead?
how do we maximize re-use of work across startups
{Risks to mitigate: The kernel of value is often the data, or customer lists, or IP, or preferential placement of products in a marketplace, or brand assets. What happens if a bad actor abuses that segment and extracts value way beyond their share of contribution at the expense of intangible trust towards the entire company? This is a frequent problem with political corruption in small countries.}
Raw Ideas to properly formulate:
Real world escorow automated service
Implement web 3 tech in web 2. Make web 3 companies describe their mechanism and then angel engineers to implement the mechanisms in web 2.
If you can identify or predict gaming - the person who predicted you would not be well calibrated can win money (for using uncertainty in prediction markets to buy research). Similar to reinsurance.
Mechanism for turning a power law distribution of risk into a normal distribution.
Equity for people looks like this
Mechanisms for transitioning from surveillance capitalism to alignment capitalism m{Risks to mitigate: complexity is the problem with most of the above and current financial systems. Value rests in those who understand and can manipulate the complexity ahead of the masses who don’t know or care or can’t comprehend what is happening. Do have a system that works you need trust among participants, which means no one party can gain an information advantage by gaming the system. For that you need a simple enough system that is so transparent to understand that every participant can see if it’s gamed..}
Comments