RE: Play your part in making Steemit better
You are oversimplifying the scope of abuse if you think auto-downvoting a one word post, or a few words for the sake of discussion, would resolve things. There are such things as a misclick of an incomplete post or a failed test, even with rewards declined. What about other languages? What about photographs with little to one word descriptions?
Are you gonna waste voting power downvoting those? Are those the only kind of abuse that exists on this platform?
Don't you think with millions of SP, @steemcleaners or @berniesanders would have done that long ago? What about @therealwolf? He's a developer, why do you think he hasn't done that already?
UA and Truffle Pig are works in progress. While I believe AIs have that sort of potential, but the technology is not refined to that point yet. Take @aicurator for example, it's great at being efficient. It doesn't see quality. How about @cheetah? It's ran into some hiccups with non-English languages.
Have you even seen the current UA algorithm in the works right now? It correlates with whether or not you interact with Witnesses or not. It's still got some ways to go.
Excellent points but using keywords and word count to trigger an autodownvote bot is at least something. None of the points raised in the original post are an ideal solution either but as Steve rightly points out, everyone has a part to play and every little helps.
None of this addresses the question of how autocuration is a good thing. The point still remains that if you arent read by a person you dont have chance to be upvoted. Thats life and we need encouragement to actually read and then decide if what we read is worthy of one of our upvotes.
If we follow the route we are on, we will get re spun articles using one of the many online article spinners being upvoted by ai powered auto upvoters, with people following only what they know will pay them a curation reward.
The languages issue is a substantial one but shouldn't be as this is allegedly a decentralised environment.
Many of the views expressed on the abuse topic come from 'privelidged' for want of better word, context. This also needs addressing and throwing into the equation.
I look forward to reading @slobberchops article on reward flagging as a career move!
At the end of the day, the rewards reclaimed from flagging the post were repaid to the community minded people who flagged it! If there was no reward, how many would have still flagged it?
It is something, but keywords and word count can be gamed; hence why we need to be careful when we consider which part of the process to automate.
Auto-curation is a gray thing: it all depends on the user. For example, @johngreenfield taking a break from Steemit and set authors he knows he enjoys the content on auto-votes. It's like a sponsor vote, the creator is encouraged to continue to work while he is gone.
@fulltimegeek does something similar. He has a whitelist of authors.
Like you said, it can't be entirely a good thing, especially when we hit over 41% on Steem Auto. It's like half the people aren't even here. At some point, auto-curation tends to be abused anyhow.
As for @slobberchops career choice, well good luck to him. Checking out all their wallets, I think only the actual Steem Cleaners and maybe some industrious Reporters in their group have good compensations. Other than that, I don't think even Bernie's Abuse Reports can top those in terms of returns.
Spun articles are already around. Some are more apparent. Some are not. And yes, I do agree with the reading part because that is one thing the AIs won't be able to do for some time and picking up spinning is one of them.
Automatic voting will always have limits. I think we could use AI to alert humans about good and bad posts that need attention. A lot of those we flag could be picked up that way. If be tempted to create a bot that picks up SFR confirmation comments and automatically does a follow on flag of appropriate size.
Hopefully, that will be in the works as we improve our workflow. I think @admiralbot is supposed to fulfill that role at some point when it becomes strong enough.
There are other ways we can shave time such as having a bot that picks up all SFR mentions, so all the approvers have to do is review all links, but like you said, there are limitations. We could be spammed by throwaway accounts, etc.