In case you missed it, I put some time into producing a version of that graph that removes the pumpkin vote, so we can see what it would look like without it. Many of the top 20 actually stay in the top 20, but a few move out.
I also put a couple of days into coding an algorithm that produces the list of witnesses as it would be if witness decay was introduced - without attempting to guess who would revote (which obviously many voters would).
I'm about to launch a new channel on Steem called Steem Ocean Undercurrents that includes regular blogs on this kind of data. Stay tuned!
Awesome work! I had missed it. Also, we need to take into consideration that if @freedom's vote didn't exist maybe some votes would be cast differently than what they are currently.
Very interesting. I don't your description here is partially correct but I've look at the post and the data is indeed very interesting.
Thanks! Oh yes, I'm sure many things would change in such a dynamic system with so many individuals involved. I could improve the algorithm to take into account how active each of the voters are that were removed due to the decay, to make a guess at how likely they would be to revote - but I have more important projects to put time into currently really. I will post more updates showing the 'movers' in the witness list when I get time, soon.
Faster decay should occur for inactive witnesses.
I was glancing through the posts, but was noticing that the mention of automatic re-voting services does not seem to be addressed. Is there anything we can do about that, because wouldn't that nullify the whole idea?
Exactly. As long as any account is selling its vote to an upvote service or something similar or even being part of a curation trail it will show up as a live account.
we are mainly talking about witness votes here, not post votes.
Think you are misunderstanding @a0x1's point. Seems they are touching on the point of people using a witness auto revoter and there beinh another metric, presumably involving other account activity such as post voting, to determine if the account is active.
Think the premise is that a witness revote for an otherwise inactive account should not count.
Yes, an auto voter could be created for witness votes too. Ultimately, bots can carry out any activity on here, so I'm not sure there is a foolproof way to code these kinds of features, but they will have some effect in important cases, such as where users literally physically die (and don't set an auto voter).
@reggaemuffin had the idea of requiring a witness vote to have been changed during a 6 month period in order to prevent votes from starting to decay or being cancelled. This would mean that autovoting would have to include the changing of accounts being supported too. This could be done by crafty bot coders - but I still think the benefits are worth it.