You are viewing a single comment's thread from:

RE: SBI in an HF21 World - Part 2: Convergent Linear & Downvote Pools

in #hf215 years ago (edited)

Unfortunately down voting is too easily abused by high stake users where they can effectively ban by auto downvote users/accounts they disagree with. Yes, there are many spam accounts that exist, but it is way to easy right now for a high stakes user to downvote everything an account posts, destroying their reputation score, making their account pretty much useless. The alternative score that is based on follows, which isn't based on votes on posts, if it were to become the main score to replace the existing reputation score, I could see as being one way to reduce the abusability of downvotes hitting reputation. Personally, all I'm seeing with the new "free" downvote pool is the potential for a rapid "abuse of commons" for massive downvote campaigns. Think of someone setting up a downvote bot "swarm" by registering lots of accounts...

Sort:  

When I used 'reputation', I meant actual social reputation, not broken blockchain reputation. As you've stated, many users have a high blockchain reputation but low social reputation, and the opposite can easily occur when people with high blockchain reputation choose to trash those with high social reputation (but not as deep pockets).

The alternative score that is based on follows, which isn't based on votes on posts, if it were to become the main score to replace the existing reputation score, I could see as being one way to reduce the abusability of downvotes hitting reputation.

This I believe was the original intention of @steem-ua, though their implementation proved that a score based on followers can also be gamed, though not as easily as the blockchain reputation score.

I would expect any type of reputation system to be gameable to some degree. While the @steem-ua approach is not perfect, I do consider it way more useful as a metric than the existing and broken reputation score system.

That I agree with. Steem-ua's system is imperfect, but way better than the existing system. Unfortunately it did not gain enough support even to be added at condenser level.

One could make a condenser fork that does this, but only the users that trust the administrator would ever use that website.

There is not actually any good reason that I'm aware of for 'reputation' to be calculated at protocol level. If a condenser created a better reputation methodology that was found by users to be more meaningful, it would become a feature that attracts people to that condenser.
Better reputation calculations are not something any condenser marketing as their 'competitive advantage', as far as I know.

Right now it is a chain level calculation (to the best of my knowledge) based strictly on votes. IMHO a very abusable system. I only see a free downvote pool as making it more abusable.

There is the @steem-ua project which demonstrates an alternative reputation score calculation that isn't based on votes.