You are viewing a single comment's thread from:
RE: How-to solve SPAM and Democratize Steem: Introducing UserAuthority
Very interesting. The SMT white paper talks about oracles for controlling who can access an SMT rewards pool and I wonder if algorithms like this might be more effective and/or help aid oracles in making decisions about accounts and the value they bring (or take away) from the network.
I'd love to see an open market place for various spam prevention algorithms that users could implement to improve their own experience. Maybe it could involve shared mute lists or something similar along with regularly published reports of abusers taking from the reward pool but adding no value so that certain accounts like @steamcleaners (or something similar) could go through and downvote them and other busy whales could delegate some SP to help.
May I add a simple suggestion for a bot. First of all I'm not a programmer. Just a smart guy who love steemit.
You could use a bot to check for repeated comments of a user. Lot's of spammers copy/paste their spam. There are also words like upvote,plz,follow used repeatedly. If a bot could be made no find users with high amount of copy/paste content and for common words used in upvote begging, It could generate a list for whales to review. Then those who are clearly spamming could be picked and put on a public list.
I'm very happy to see many people fighting the good fight. You guys give me hope regarding the future of steemit. My simple suggestion wouldn't go too far. Spammers will adopt. I actually came across serial spammers with reputation above 50. It's just nuts: https://steemit.com/steemit/@vimukthi/serial-spaming-your-way-into-a-reputation-of-56-how-did-this-happen
Wish you guys best of luck and hope my suggestion helped :-)
@vimukthi
Hi @vimukthi , your suggestion on how to algorithmically detect spam / repeated comments (= content analysis), could be another extension of my UserAuthority (UA) spam identification capabilities (= user authority / popularity analysis).
Your suggestion could also be implemented stand-alone via bi-directional hashing encryption, by calculating "character proximity". Lots of difficult words here ;-), but your solution on its own is not really needed. Compare this principle to copy-pasta webpages: it is impossible to stop authors publishing such webpages, but it's Google's only task to prevent those pages getting to the top of the search rankings (SERPs).
Seen from a Steem ecosystem perspective, it's merely important to stop those comments from receiving high-value author rewards. Hence the need for my UA algo to be implemented Steem-wide (See my HF22 proposal UA * SP at the bottom of my article.)
Mathematically, it looks like your UA algo will work just fine :)
Some folks may not like it, but as with all algos there's always room for improvement!
BTW: here's a link to an article on "bot tells"
https://steemit.com/steemit/@torquewrench1969/tips-i-use-to-identify-bot-accounts
Spot on! ;-)
Another post comming out about this project shortly!
An equivalent of this algorithm was used, in another form, as the only relevant algorithm, when Google launched in 1998. In this form, the same problem Google had in 1998 is addressed: which page should be at the top results, ref. which user is to be regarded as authorative.