Splinterlands Economics: Prisoner's Dilemma

in Splinterlands2 years ago (edited)

Hello everyone! We are back again today with another edition of Splinterland Economics - a series in which we introduce a basic economic concept and then apply it to Splinterlands. If this is your first time reading, just to tell you a little bit about myself: my day job is in an unrelated area but I consider myself a little bit of economics nerd - I read a little (or maybe way, way) more news than I should, double majored in econ, and am obsessed with optimization. I love the way Splinterlands is equal parts card game and resource allocation game. My goal with these articles is to share a little bit of what I know with you all.

Our subject today is is one that you may have heard of before, and comes up from time to time in game theory. It refers to a specific situation where suboptimal overall results can occur as the result of individual decisions, and is closely related with the ideas of cooperation and/or collaboration. I would like to give a shout-out to @eijibr who suggested this topic a while back - it took me a while to figure out the best way to present this subject, but I finally cracked it (or at least, I would like to think that I have). Today, we will be talking about the Prisoner's Dilemma!


Splinterlands Economics Thumbnail (29).png


What is the Prisoner's Dilemma?

The Prisoner's Dilemma refers to a hypothetical situation where two parties (or alleged criminals) have been arrested by the police. Each prisoner has been separated from the other with no communications, and is offered the opportunity to either cooperate with the other prisoner and stay silent, or to defect and to testify that the other party did, in fact, commit a crime. If both parties cooperate with each other then the police will have no strong case and both parties will receive a light penalty. If both defect and tells on the other then both will receive a heavy penalty. And if one cooperates and the other defects then one will receive no penalty and the other will receive a very heavy sentence.

This chart here shows the possible amounts of prison time for each combination of outcomes. The main point here is that each person will receive less prison time by defecting, assuming that the other person makes a specific choice. For example, assuming the Prisoner B cooperates, Prisoner A can serve a 1 year sentence by also cooperating, or a 0 year sentence ("goes free") by defecting. If we assume Prisoner B will defect, then Prisoner A will serve a 3 year sentence if the cooperate, or a 2 year sentence if they also defect. In other words, the individual incentives encourage both prisoners to defect and tell on the other.

Notably, if this happens then the overall outcome for the prisoners is much worse - 1+1 = 2 years total jail time if both cooperate, compared with 2+2 = 4 years of jail time if both defect. You can see here what the "dilemma" is - both prisoners are incentivized to betray the other, but if they do that and maximize their own outcome then the collective outcome will turn out much worse than it could be. This problem is not limited to prisoners facing jail time, and can be generalized to many situations in real life where there are two parties with individual incentives, who must make a decision about whether or not to cooperate with each other.


How do we apply it to Splinterlands?

While there are not many examples of true prisoner's dilemmas in Splinterlands since communication is nearly always allowed, Splinterlands does have several instances of situations where individual payouts are not directly aligned with collective results. One example of this is when players work together to play in a guild. Guilds require time and investment in order to do well in brawls and for everyone to receive more merits (and soon, SPS). Each member of a guild would be better off if they slacked off and did not contribute to the guild and instead invested those resources in cards or other items for their personal use. However, if players choose to cooperate and invest in a guild then they will be able to level up buildings, participate in higher level brawls, and get a much greater payout.

Another example of individual and group incentives not being perfectly aligned is when it comes to voting on proposals as a part of the DAO. DAO members are asked to vote on a variety of proposals. Each proposal will have financial consequences for the voters, and it is possible (or even likely) that members will be faced with proposals where their individual incentives encourage them to vote one way, while the long term health of the ecosystem may be better off if they vote the other way. If or when such a proposal comes up to vote, members must choose whether to "cooperate" and vote for the long term health of the game, or to "defect" and vote based purely on their own interests. Again, this is not a pure prisoner's dilemma since communication between voters is allowed and there is a shared outcome, but it does represent an example of incentives pulling decisionmakers in opposing directions.


I mentioned at the start of this section that there are very few true prisoner's dilemmas in Splinterlands. Well, "few" is not the same as "none", and this final example is the result of relatively large amounts of controversy within the community. When you participate in ranked battles, you do not know who you are playing against, and generally try to select your summoner paired with your best possible team of monsters. However, thanks to the details of the Splinterlands ranked rewards, it is possible to get a higher total reward if the battle has specific outcomes. For example, is one player is on a win streak or has a higher ECR than the other, then there will be more total rewards paid out if that particular player is the victor. And it is also in the interest of both players to avoid a draw, as there are no rewards at all paid out if that occurs. I should also mention here for anyone who is relatively new to the game or who may have forgotten - back before this became an issue you could see the names of the players you were matched up against before entering your team, and therefore be able to figure out exactly who you were playing against and their current status.

If there is no player communication then there is no issue with this system. However, if there is the possibility of collusion - say, if players are chatting with one another, or if the same person is operating both accounts in the battle - then there is an incentive to cooperate in order to increase the total rewards payout by having the player with less to gain submit a suboptimal team (or simply surrender). THIS is a large issue, and resulted in significant developer time and resources being spent in order to prevent win-trading and matchmaking between multiple accounts operated by the same person. The result is that today we have (relatively) anonymous matchmaking, where you do not know who you are playing against and simply must play whichever team you feel has the greatest chances of success.


Why should we care?

As we play Splinterlands or navigate our everyday life, we are constantly faced with decisions. And since we are not alone in this world, many of those decisions require us to decide whether to cooperate with others or not. And while not every decision requires cooperation and not every situation will have a Prisoner's Dilemma element to it, being aware of this possible scenario is a reminder to keep in mind the interactions between the individual incentives of involved parties and overall outcomes shared by everyone collectively. I am not going to say that choosing to cooperate is always the best option, but understanding the dynamics and game theory of group decisions helps us to make more informed choices for ourselves.


Thank you so much for reading all the way to the end. Interested in seeing some more of my writing in the future? Be sure to give me a follow! In the meantime, if you'd like to see some of my recent posts:


Thinking about giving Splinterlands a try but haven't signed up yet? Feel free to use my referral link: https://splinterlands.com?ref=bteim, and be sure to reach out to me if you have any questions!

All images used in this article are open source and obtained from Pixabay or Unsplash. Thumbnails borrowed with permission from the Splinterlands team or made in Canva.

Sort:  

Nice work! 😃 Thanks for the mention

!PIZZA

Thank you for the suggestion! 😁

PIZZA!

PIZZA Holders sent $PIZZA tips in this post's comments:
@eijibr(3/5) tipped @bteim (x1)

Join us in Discord!

Thanks for sharing! - @rehan12