Historic data is meaningless because it does not take into consideration new incentives appearing from the new rules. I fully admit here that if the downvotes don't occur, it will be a disaster. But I have a good feeling about that just based on the economics of it as well as various levels of commitment, and the fact that some already do it even though it costs them right now (obviously, talk is cheap).
Posted using Partiko Android
I had high hopes for the down votes, but after runningthis poll, now I no longer do.
Based on this poll and actual data of downvotes and retaliations from the analysis's I used to run from @pibarabot, I'm pretty much convinced there is verry little economic insentive to make use of free down votes.
There is pretty much a culture of fear and retaliation surrounding the use of down votes on this platform.
There's not much stake represented in that poll. What's the analysis from @pibarabot? There's definitely an economic incentive to use all of the free downvotes, and it doesn't even matter who you are (although obviously the more stake, the more impact).
Posted using Partiko Android
Actually there is a tangible economic incentive not to use the free down votes:
retaliation!
I used to run a daily script that processed all flags done on one week old posts, until Steemit Inc killed my code by implementing API rate limiting a few months back (here is a sample graph of one day. One thing that stood out was that the same accounts kept showing up in different roles, either as flagger or flagee. On closer inspection, a substantial amount of down vote weight could be attributed to some form of retribution for earlier flags against accounts with likely relations (voting proxy or non-standard recovery account) with the flagging account.
I don't know what to make of it except that people flagged despite being retaliated against, and despite the current opportunity cost from doing so. I get that retaliation is a deterrent, but I'm pretty sure free downvotes will be flying in different ways than what you are analyzing. (Whether it holds or does something productive or is ultimately detrimental overall all remain to be seen)
Posted using Partiko Android
Retaliation isn't all bad. Alice and Bob get into it with each other, and that means that Charlie walks away with the money. Maybe Charlie was actually doing something right by not being seen as downvote-worthy by anyone.
Point is, if Alice and Charlie both behave exactly the same, both notice Bob does something bad, like calling in bid bots on a post tagged for one of the top curation initiatives (something I've used flags for myself a couple of times, and have been retaliated against a couple of times as well), and Alice was the only one acting on it by attenuating the bot votes with a down vote, then behaving like Charlie is incentified above Behaving like Alice by the retaliation by Bob.
For the flagging to truly work without negative incentives, I think we would need some kind of flagging SP that can be delegated separately from regular SP to semi-automated services. Without something that can shield those who for example want to use their free downvotes for attenuating false curation bot upvotes against direct retaliation. As the EIP comes with nothing that shields us from retribution down votes, I'm afraid the incentive to act is still negative.
In this case A, B, and C did not behave exactly the same.
If you think that literally everyone will downvote and/or get downvoted and/or retaliate and/or get retaliated against all in lockstep or that literally no one will and we will all drown in a sea of sameness, we will have to wait and see but I'm very much doubting it.
Somewhat. Certainly there is still minimal if any positive incentive to downvote, and indeed many still may just not bother, or not want to enter the fray. That is a clear potential point of failure. However, all those issues already exist. Compared with now, the overall cost of downvoting is surely reduced dramatically for nearly everyone. How much difference that will make remains to be seen.
That is harmful nonsense. Old data will show what the new insentives are. You can than middel second iteration data accordingly, rinse and repeat, but you need to run the new rules against the old data just once. If you don't, the whole new incentives narratives will be nothing but speculative fiction.
And that's exactly what it is. You can run the new rules on old data, and that's exactly how you get the immediate distribution at hard fork time. But after that, you can't predict anything. I don't see how you can claim that this has any predictive power. After all, you aren't going to be able to estimate how many downvotes are going to be issued, for starters.
Posted using Partiko Android