+20 -11

Impact Evaluators.md

reviewed

··· 70 70 - Information asymmetries - Suppliers control the metrics and optimize for growth rather than user outcomes 71 71 - **Information elicitation without verification**. Getting truthful data from subjective evaluation when you can't verify it requires clever [[Mechanism Design]]: 72 72 - [Peer prediction mechanisms](https://jonathanwarden.com/information-elicitation-mechanisms/) that reward agreement with hidden samples 73 73 - - [Bayesian Truth Serum](https://www.science.org/doi/10.1126/science.1102081) that uses both answers and predictions 74 74 - - Coordination games where truth serves as a Schelling point 73 73 + - [Bayesian Truth Serum](https://www.science.org/doi/10.1126/science.1102081) that uses both answers and predictions. 74 74 + - Coordination games where truth serves as a Schelling point. 75 75 - **Collusion resistance**. Any mechanism helping under-coordinated parties will also help [over-coordinated parties extract value](https://vitalik.eth.limo/general/2019/04/03/collusion.html). Countermeasures include: 76 76 - - Identity-free incentives (like proof-of-work) 77 77 - - Fork-and-exit rights for minorities 78 78 - - Privacy pools that exclude provably malicious actors 79 79 - - Multiple independent "dashboard organizations" preventing capture 76 76 + - Identity-free incentives (like proof-of-work). 77 77 + - Fork-and-exit rights for minorities. 78 78 + - Privacy pools that exclude provably malicious actors. 79 79 + - Multiple independent "dashboard organizations" preventing capture. 80 80 - They should be flexible as it's hard to predict ways the evaluation metrics will be gamed. 81 81 - - [The simpler a mechanism, the less space for hidden privilege](https://vitalik.eth.limo/general/2020/09/11/coordination.html). Fewer parameters mean more resistance to corruption and overfit and more people engaging 81 81 + - [The simpler a mechanism, the less space for hidden privilege](https://vitalik.eth.limo/general/2020/09/11/coordination.html). Fewer parameters mean more resistance to corruption and overfit and more people engaging. 82 82 - Demonstrably fair and impartial to all participants (open source and publicly verifiable execution), with no hidden biases or privileged interests 83 83 - Don't write specific people or outcomes into the mechanism (e.g: using multiple accounts) 84 84 - [An allocation mechanism can be seen as a measurement process, with the goal being the reduction of uncertainty concerning present beliefs about the future. An effective process will gather and leverage as much information as possible while maximizing the signal-to-noise ratio of that information — aims which are often at odds](https://blog.zaratan.world/p/quadratic-v-pairwise). ··· 102 102 - Reinforcement Learning? 103 103 - Genetic algorithms? 104 104 - Is the evaluation/reward better than a centralized/simpler alternative? 105 105 - - E.g: on tabular clinical prediction datasets, standard logistic regression was found to be on par with deep recurrent models 105 105 + - E.g: on tabular clinical prediction datasets, standard logistic regression was found to be on par with deep recurrent models. 106 106 - [IEs need to show how the solution is produced by the interactions of people each of whom possesses only partial knowledge](https://news.ycombinator.com/item?id=44232461). 107 107 - **Exploration vs Exploitation**. IEs are optimization processes with tend to exploit (more impact, more reward). This ends up with a monopoly (100% exploit). You probably want to always have some exploration. 108 108 - Do IEs need some explore/exploit thing? E.g. Use multi-armed bandit algorithms to adaptively choose between evaluation mechanisms based on historical performance and context. 109 109 - - Use maximal lotteries to enforce the exploration 110 109 - Having discrete rounds simplify the process. Like a batch pipeline. 111 110 - The more humans gets involved, the messier (papers, ... academia). You cannot get away from humans in most problems. 112 111 - [Campbell's Law](https://en.wikipedia.org/wiki/Campbell%27s_law). The more any quantitative social indicator is used for social decision-making, the more subject it will be to corruption pressures and the more apt it will be to distort and corrupt the social processes it is intended to monitor. ··· 122 121 - Ostrom's Law. "A resource arrangement that works in practice can work in theory" 123 122 - To create a permissionless way for projects to participate, staking is a solution. 124 123 - You want a reactive and self balancing system. Loops where one parts reacts the other parts. 125 125 - - Feedback loop with the errors of the previous round 124 124 + - Feedback loop with the errors of the previous round. 126 125 - The entire thing needs to be like a game. People want to participate because is fun and they get some rewards. 127 126 - Decide metrics so that gaming/exploiting them means having a better tool, system, process. 128 127 - Which algorithm is the best assigning weights is not the best question. 129 128 - What would you change about the algorithm? 130 129 - What would you change about the process? 131 131 - - Have a democratic way of expressing the values of the community and some representatives 130 130 + - Have a democratic way of expressing the values of the community and some representatives. 132 131 - Economist might be good at analyzing economies but doens't mean they're good at creating them. A phisicist or ecologist might be a better fit. 132 132 + - Making it so people don't have to do somehting is cool. Makeing it so people can't do that thing is bad. E.g: time saving tools like AI is great but humans should be able to jump in if they want! 133 133 + - If people don't want to have their "time saved" have the freedom to express themselves. E.g: offer pairwise comparisons by default but let people expand on feedback or send large project reviews. 134 134 + - Information gathering is messy and noisy. It's hard to get a clear picture of what people think. Let people express themselves as much as they want. 135 135 + - Complex model of people aren't always good (performative reactions, noise, ...) 136 136 + - Prioritize consent and community feedback. 137 137 + - Community should steer the ship. 138 138 + - Design a democratic control that reacts to feedback. 139 139 + - Allow people to express themselves as much as they want. 140 140 + - Super expert with lots of context already have the weights! 141 141 + - Pairwise shines when all the context is in the UX. 133 142 134 143 ## Principles 135 144

+1

Incentives.md

reviewed

··· 19 19 - Clear target metric to improve. 20 20 - Intentional system design. 21 21 - Commitment to study the metric. 22 22 + - Human values are highly dimensional. Nudging people in the right direction is hard, specially because nudges usually are very low dimensional. 22 23 23 24 ## Incentive Framework 24 25