Yes, the bunny has been fluffing a few of those already... In a way it's nice for the receivers of these votes when the bunny get's triggered by them, but as they don't express anything at all, I was also thinking those votes should definitely be ignored.
Now I did implement a blacklist for fluff-up-receivers from the get go, but I hadn't initially considered at all to also filter for certain incoming voters until I saw a bunch of these random(?) zilch-votes drawing the bunny's attention.
I'm trying to think of possible reasons that would require a separate list for that or whether I should just apply the same blacklist for both "ends"... Yeah, that's definitely something that's already at the top of my to-do list.
The 0 value upvotes from these curation snipers also are probably the largest drain vs typical dust votes which will vary more in total value averaging a much higher rate. Finding a good way to ignore them should allow more votes to be cast.
Maybe when comments will have to be ignored give preference to those which need the least assistance. Better to save 20 comments at .01 vs 10 with 0 value votes IMO. It save much more dust as the dust saved had value and you can save more of them.
I think the vote value of the usual suspects like penghuren or cheneats is actually roughly comparable with what a new plankton account would be able to produce in terms of rshares... So filtering by a min-impact level or something like that doesn't sound like a good option to me.
So, for now I'm going with the blacklist filtering specific account. I've already added the the names of the two mentioned above. Do you know of any other accounts like those from the top of your head?!
In terms of prefering dust that needs less fluff to maximize the number of items overall... I see what you mean there, that would turn the prioritisation mechanics into a knappsack problem and put more focus on optimizing momentary capacity. But as the "capacity" here also has a time vector given by the recharging VP, I think of it more like a multiplexing issue, so I've chosen to go by a demand-centric weighed max-min fairness distribution.
This is just another reason we need to be able to downvote the voter not the content creator. The content creator has no control over who votes in most cases. I myself would like to kick the freeload voters (those with a 0 value vote) off my content. So many people voting at 1% and zero value. I did not understand that for them they get a curation reward in return, a surprise to me. They fixed self upvote but not here have nothing but give me something in return people. I was wondering why I was getting so many of those. Also it invalidates @steemchiller's CSI attempt I mean really: Voting CSI 11,135.8 ( 0.00 % self, 7872 upvotes, 2004 accounts, last 7d ) there is just no way to look through and see if they have ever given a valuable vote at all
Yeah... I think it's a bit of an ethical grey zone... it's not like they actually do much damage. They basically just collect a tiny bit of the curation rewards created by those voting after them... But it's certainly a "smart" way to maximize curation returns without actually partaking in any "proof-of-brain"...
Much rather have the bunny save that VP to be able to fluff up more genuine dust.