RE: What's a Minnow to do? The Game Theory of Steem, Part 4
This is a really, really excellent series. You write very clearly and explain things well. Well done!
Having said that, I don't think it describes my voting strategy. I usually consciously avoid voting up highly popular posts because (1) I have limited voting power and (2) I realize that voting after whales doesn't earn me much curation rewards, whereas voting before them does. So, I'm constantly scouring the "new" page for content that I think is great AND that I think whales will think is great, and then I try to vote it up early before the whales have spotted it.
How does my approach fit with your theory? Or, is that covered in your next installment?
Also, wouldn't the fact that voting awards scale exponentially rather than linearly change your analysis? By assuming linear, I think you did "rig the game" a little.
I'm looking forward to the next installment!
Thanks for the kind words! Your support and promotion is always greatly appreciated. The more of my posts I can push to the front page, the more I'll write.
You're right on all counts, except that assuming linear/proportional rewards actually isn't rigging the game in my simplified model. We'd get the same results using the proper quadratic rewards.
Your suggestion is to add in
Adding in any of these things would definitely change the answer, and I suspect that we could very easily argue that your voting strategy is game-theoretically justified and individually optimal. We could model you as a Type-1 player who has probabilistic beliefs about the preferences of whales, and votes for posts specifically in an effort to push them up high enough to get whale attention. Then the Type-0 whale comes along and votes for what he likes. I probably won't do this analysis myself quite yet - I'm itching to get on to analyzing whale behavior.
What makes you think that?
My understanding is that earlier votes are weighted more than later votes so as to incentivize people to scour news posts for good content. No?
I don’t remember reading that in the white paper.
Edit: I just re-read the applicable sections of the white paper and I do not see any mention of your expectation. I note the curation reward algorithm was finalized after the release of the white paper.
Here: https://steemit.com/steem/@abit/new-curation-reward-algorithm-huge-penalty-to-early-voters
The current curation reward system is not detailed in the whitepaper, and the section on the website where it explains the math is nothing like the current implementation in the code.
Part 5 is up!