You are viewing a single comment's thread from:
RE: Follower Count Report for Thursday, August 25th
Personally, i think all available data is valuable. What would be really valuable imo is if this data could be filtered to eliminate spam accounts. This would take some creative data manipulation. For example you could eliminate all inactive accounts and set some thresholds for eliminating likely spam accounts as well. Not sure if others will find such information valuable, but it seems like it would potentially be more useful than the raw data itself, especially follower counts seem to be highly spammed these days.
Great point, @trogdor. That's what I've started to do with the "trying too hard" list. For the most part, those accounts just bubbled to my awareness in an "Wait, that can't be right" sort of way. True human inactivity is a difficult thing to measure. Those accounts can post regularly and even vote up their own stuff (and comment too) via other bots to make it look like they are real, active accounts. I'm thinking the payout amounts are currently the only way to know what's real and what's fake (or, at the least, what someone spent a lot of money purchasing influence to produce). If someone has hundreds of followers but no content paying out more than a few cents, that makes me suspicious. At the same time, it is possible for someone real to go unnoticed by the whales. It's a difficult challenge to provide meaningful data as opposed to just raw data.
Good point. I wonder if it would be possible to make use of the reputation scores too. For example, if someone has rep 25 and 100 followers, that's a pretty clear red flag, unless it's some whale account that has very low activity but followers nonetheless for celebrity status. Then again you don't want to discount new members either. One thing that could be interesting to look at would be total steem power of your followers. That is an easy way to solve the sybil issue, i think.
Ah, I like that! It would also emphasize the very real incentive to purchase influence here which can drive up the price of STEEM by creating more of a demand for Steem Power. I may have to play around with that idea. Instead of going off of numbers like "has 10 followers" maybe I could look into something like "has more than the initial starting Steem Power balance."
Great ideas, thank you! Now I'm thinking a programatic look at this data may still be useful.