I've learned a lot following you, and am interested to see how your bot improves over time. Having thought about it myself, I'd also like to know the steps you'll take to make the bot resistant to exploitation.
You are viewing a single comment's thread from:
I'd like to know those steps also. :) As a first pass, I think randomness and adaptability are two extremely important pieces of it. Currently, my bot adapts. For example, right now it thinks the tag #curation is a strong predictor of large payouts. If many authors start spamming that tag for my votes, the value of that tag in my model will automatically decrease and my bot will stop voting for it. There will necessarily be some learning delay, but it should work.