Effective Altruism Forum
EA Forum

by damc4
8 min read 0

3

Existential riskPhilosophyPolicyArtificial intelligenceDecision theoryDraft Amnesty Week (2025)Economic inequalityEthics of artificial intelligenceGlobal catastrophic risk
Frontpage
  1. In this post
    1. How to stop growing inequality.
  2. Draft Amnesty
    1. This post is submitted with Draft Amnesty. If the idea doesn't make sense to you, it's likely that I haven't explained something clearly or I haven't justified something deeply enough, so please stay open to the idea and consider it again, if I share it again later.
  3. Problem
    1. People get increasingly unequal in terms of how much resources they have (money, computational power, intelligence and other useful things...).
    2. That is bad for collective happiness because a person's happiness grows logarithmically with the amount of resources that the person has. So, the collective happiness is the highest when the resources are split evenly.
    3. The problem is amplified by the following facts:
      1. It might be true that on average, it is easier to acquire resources at a cost to others. Because a lot of resources are physical objects and if one person uses them, then the other person can't.
        1. Example: if one group of people uses part of land in certain way, then the other person can't use the same part of land. So, it's easier to take land from other people than to colonize Mars to have more land.
      2. resources can be used to acquire new resources. For example, computers can be used to create artificial intelligence which can be used to create better computers which can be used to create better artificial intelligence. Therefore, people increase their resources exponentially (with some random factor) which increases the inequality over time.
      3. All else equal, the world with the above properties will end with a winner takes it all scenario - one person accumulating close to 100% of resources. That can be proven with a simple program simulating that world. That scenario is bad given the logarithmic relation between resources and happiness.
    4. How can a person maximize its happiness, in this situation?
  4. Game
    1. The below game is a simplified model of the problem. If we solve this game, we will have a simplified solution.
    2. There are players: .
    3. Each player has some number of resources.
    4. At the beginning, each player has 1 resource.
    5. In each turn, the number of resources of each player is multiplied by , where is a random number from a certain range. The range is constant for the entire game, but is different per each turn and player.
      1. That rule aims to reflect reality that resources can be used to acquire new resources, therefore people increase their resources exponentially.
    6. Whenever one player increases its number of resources by , then one other random player looses resources. Except their number of resources can't go below 0.
      1. That rule aims to reflect reality that on average, it is easier to acquire resources at a cost to others. So, if one player gains resources, the other ones loses some resources.
    7. In each turn, a player can give any number of their resources to another player or players.
      1. The number of resources that they give doesn't have to be an integer (it can be 0.5 for example).
      2. There is no limit of how much they can give, but they can give only what they have.
      3. The previous rule (about losing resources) don't apply when a person increases its number of resources by receiving it from someone else.
    8. In each turn, each player receives a payoff equal to , where and - the number of resources that the has. Payoff is what the players aim to maximize in the game.
      1. That rule reflects the reality that happiness grows logarithmically with the amount of resources that a person has.
    9. The game never ends - it has infinite numbers of turns.
  5. Strategies
    1. Greedy strategy (bad strategy)
      1. In this strategy, the player never gives any resources to any other player.
      2. Here's what's going to happen if all players follow that strategy:
        1. As a result of luck, some players will become wealthier than others.
        2. Later, as a result of luck, some wealthier players will become even wealthier.
        3. Later, as a result of luck, some wealthier players among the wealthier players will become even wealthier.
        4. ...
        5. One player owns 99% of all resources.
      3. The key here is the fact that the payoff is a logarithm from the number of resources. Because of that, concentration of resources in the hands of one player is undesirable.
      4. Due to concentration of resources, this strategy results in suboptimal collective payoff. The collective number of resources is the same regardless if anyone gives the resources or not. But the collective payoff will be low comparing to what it could be if the resources were not concentrated.
    2. Equality law strategy (good strategy)
      1. In this strategy, the players agree to establish the following enforced norm:
        1. If a player receives a new resource, they distribute that resource evenly among all players, excluding only the players who violated this law in the past (proportionally to how much they violated it).
        2. How much should they pay exactly
      2. If all players comply with the law, the collective payoff will be optimal. Because there's going to be equality for the entire time.
      3. Once the precedent of complying with the equality law has been created, it's in the best interest of each player to comply with the law. Because the penalty will exceed the benefit of keeping the resource.
      4. It is in the best interest of all players to build a precedent of complying with the equality law, because then it's in the best interest of all players to comply with the law. And when all players comply with the law, the expected payoff of each player will be higher than when players follow the greedy strategy.
      5. Therefore, if the players are rational, then they will comply with the law for the entire time.
      6. This strategy results in optimal collective payoff, assuming that all other players are rational.
      7. This strategy is the best strategy, as far as I know.
    3. Equality law strategy with ignoring previous equality law (bad strategy)
      1. In this strategy, the players agree to comply with the equality law, as in the previous strategy.
      2. However, once some people become wealthy (once they are lucky to receive the resource), they choose to ignore the existing equality law so that they don't have to share their resources with the poor. And at some point they choose to establish a new equality law that applies only to the wealthy (the wealthy players share their resources only with other wealthy players) so that the equality is kept among the wealthy and they don't have to face the downsides of inequality among them.
      3. Here's what's going to happen:
        1. Equality law is established.
        2. Some players are lucky and receive resources. But they ignore the equality law.
        3. The players that were lucky and are now wealthy choose to establish a 2nd equality law that applies only to the wealthy ones.
        4. But some people among the wealthy ones are lucky and become the wealthy among the wealthy. They face the choice of ignoring the 2nd law.
        5. Now, the wealthy among wealthy have the following options about what they can do:
          1. They can ignore the 1st and 2nd equality law and establish the 3rd equality law.
          2. They can ignore the 1st law and respect the 2nd law.
          3. They can respect the 1st law.
        6. If they ignore the 1st law, then:
          1. They build a precedent that a group of players can ignore a previous law and not face any penalty.
          2. If a precedent like that is built, then the next groups of wealthy players will expect that the precedent will continue.
          3. The next groups of wealthy players won't have a reason to respect the 2nd and 3rd equality law because they won't expect that they will be penalized for that.
          4. Therefore, the only good choice is to respect the 1st equality law.
        7. Therefore, the strategy of ignoring a previous law is a bad one. Because it's in the interest of the next groups of wealthy players to respect the 1st equality law. And that law will penalize any group of wealthy players who choose to ignore it.
  6. Applying to real world
    1. The above game is a simplification of the problem, but in real world there are some differences. I have stripped these additional complications from the above game because it would be too complex to communicate all of that at the same time, putting everything into one model (game).
    2. Verification - it's difficult to know how wealthy other people are and if someone has given resources to someone else. I have an idea how to solve that problem, I might share it in one of my next post.
    3. The impact of effort:
      1. In real world, people don't just get new resources because they are lucky. They acquire new resources as a combination of:
        1. their effort to contribute to collective happiness,
        2. luck and other factors.
      2. The effort to contribute to collective happiness should be rewarded because it incentivizes that effort. So, in real world, not all resources should be evenly distributed as described in the equality law strategy. But to some extent, they still should be distributed.
      3. The higher is the role of the effort, the less resources should be distributed. The higher is the role of luck and other factors, the more resources should be distributed.
    4. In real world, it would be difficult and impractical for a person to distribute an resource evenly towards all people, every time after they acquire a new resource. So, in practice, it would mean for example paying a tax annually to some organization (in money and/or other resources). That organization would distribute the money/resources to the people. I think that system should be decentralized to avoid concentration of power.
    5. In the game, we assumed that everyone starts with equal number of resources. In real world, people already have different number of resources.
      1. The default situation (without establishing any norm) is bad for everyone. However, there are many possible norms that would be better than default for everyone, so there is a room for negotiation.
      2. The people who have more resources are more likely to be the only winner. Therefore, they have a better BATNA (best alternative to the negotiated agreement). Therefore, they should be able to negotiate better conditions.
      3. How much better depends on some variables. If we assume that how long they live (or how long the game lasts) is significantly higher than any other variable, then the split will end up to be quite equal - the longer the people live, the more equal conditions of the equality norm they will be able to negotiate. Why?
      4. Given the fact that humanity might be close to achieving LEV (longevity escape velocity - a point in time from which scientific advancement increases a person's remaining life expectancy faster than the person ages which leads to people living super long), people should be able to negotiate an equal split when it comes to the equality norm.
    6. In the game, we assumed that there is infinite number of turns. In real world, people die and their game ends.
      1. If a person expects to die soon, it might be in their best interest to violate the equality law, because they won't live long enough to receive the consequences.
      2. However, given the fact that humanity might be close to achieving LEV (longevity escape velocity - a point in time from which scientific advancement increases a person's remaining life expectancy faster than the person ages which leads to people living super long), that doesn't necessarily matter a lot.
    7. In the game, we assumed that acquiring resources is a zero-sum game - one entity wins a resource at a cost to other entity. That is not necessarily true with regards to all resources. Because of that, the threat of concentration of power is lower than we assume, but it still exists.
    8. In the game, we assumed that each player can easily communicate with other players. In the real world, creating a network of people who respect the law is more difficult but still doable.
    9. The conclusions are based on some model. That model can be missing some statements that are true and relevant, so the conclusions might be false.

3

0
0

Reactions

0
0

More posts like this

Comments
No comments on this post yet.
Be the first to respond.
More from damc4
View more
Curated and popular this week
Relevant opportunities
View more