You are viewing a single comment's thread from:
RE: Operation Clean Trending
I would see a centralised solution as a last resort if we can't build the tools to make a decentralised system work. This does mean a certain responsiveness.
There are ways @steemit could intervene without compromising the decentralisation of the blockchain itself. Firstly, the main problem is they way their 'trending' algorithm works. If the use of bidbots excluded a post from appearing in 'Trending' pages, this would go some way to solving the problem. Steemit Inc could do this for their own site. @busy could do it for theirs, if they wanted to. I am beginning to think they should.
Buying votes and self-voting are promotion, and posts with either belong in Promoted, rather than Trending.
Neither are curation, and no curation rewards should inure to either.
Only people have any capacity to judge value of a post, and curation rewards should be limited to human curation.
Agreed about 'Promotion', I have said as much elsewhere. Also your point about neither being curation.
The difficulty is in how would the platform add a feature to distinguish between bots and humans? Who has the authority to determine the difference, and how do we ensure that authority is not abused?
AI has not attained competence to pass the void-comp test yet, and fairly simple automated means of verifying humanity are extant. Captchas, 2FA, etc. These require no central authority that isn't already necessary to have keys, nor any additional layers of complexity on the blockchain.
Just add one to the module that verifies keys. Should be easy. If Yahoo! can do it, I am sure @ned can do it better.
Yahoo is centralised. The various sites that access the steem blockchain do so via a stateless API. Same API used by bots. Doing a captcha would mean doing so for every comment, every like, every share...
How can the system circumvent that without a massive rethink about how the API works?
The system requires no rethink. Each of the transactions that occurs requires the system verify the authorization of the transactor, the key used. This simply needs to have an additional authorization metric for votes added to it, a captcha, 2FA, whatever is chosen to mitigate non-human actors on the blockchain.
It would dramatically impact Steemit users, who would have to solve the captcha, or reply to a txt, every time they vote. It would also dramatically reduce the number of bots, and make bots unable to vote unless they were able to solve captchas, or reply to txts appropriately. Those are very expensive features to add to bots, and few bots would be able to undertake them.
Further, those mechanisms bots used to surmount the secondary authorization validation may also be attackable, allowing them to be eliminated as well, with additional features.
It requires no further centralization or great changes in how the blockchain operates, simply adding an additional verification module to the one that operates now. It requires human intervention to authorize votes by the voters.
Unless this is done, Steemit, perhaps Steem itself, may lose all value to people, outside of the ability to mine rewards. Investors should carefully consider how that will impact their holdings. I have no doubt they are, and presently the bots are increasing their holdings.
Once the tipping point is reached, there may be no way to recover Steemit or Steem utility to people, and investors will lose what holdings they cannot extract in time to capture nominal value through exchanging Steem for other media. Stinc will be far more vulnerable to this than individual investors, as Myspace shows.
Presently Stinc is practically servile to whales, as a matter of necessity. In this matter, the interests of Stinc and the whales are not aligned, and Stinc may need to unilaterally act to preserve it's market, even over the objections of that market. Each whale also has common interests with the others, but also points of divergence from the group, concerning their own increase relative to the group, for example.
If I have misunderstood anything regarding the implementation of botproofing mechansims, please do make every effort to enable me to understand. I am not intent on any particular means of reducing bot competition with people. I just am certain it needs to stop.
Thanks!
If I had to solve a captcha or enact a 2FA just to vote, I would not just stop voting, I would leave the platform, as it will have turned into a horrendous nightmare of a user interface. I really don't think what you are proposing sounds like a workable solution.
If anything, it would create a situation where the cleverest bot writers would gain a significant monopoly (perhaps attaching the 2fa to an automated sms responder, or the captcha to a web faucet), whilst simultaneously drastically reducing the number of votes from actual humans.
It would also create significant centralisation of user interfaces, by introducing significant barriers to interface designers, be they web interfaces like busy, dtube or utopian, or mobile apps like esteem. Chainbb has already called it quits, how many more projects would collapse if such a significant barrier to entry was introduced?
You are quite correct in your criticisms.
I also point out that I would sooner expect the pope to convert to Islam than I would expect Stinc to seek to limit votebots. It's their bread and butter.
Yet, the alternative is the eventual destruction of any socially redeeming value Steemit has. Bots suck the value of upvotes out of the users and increasingly concentrate it in the accounts of whales.
This is ongoing, but there is a limit to how much longer it can go on. Presently Steemit has a ~10% YOY retention rate for accounts. 90% of users give up before a year is out.
The bots are getting worse, and making the problems that are already insoluble for 90% of folks worse.
Bots or people. Steemit can live without bots, but it cannot without people.
While I mentioned captchas and 2fa, I am no specialist in botproofing, and others may have better ideas on how to manage it.
As to impediments to additional interfaces to the blockchain, again, you are correct. These impediments aren't necessarily insoluble. Impediments aren't roadblocks or unbridgeable chasms. Dunno why Chainbb gave up, so can't comment. 90% of new accounts gave up too, though, and that might be why.
Even without you, as long as the bots were gone, Steemit would be a better place.
I'd miss you though!
A bad user interface would drive everyone away, not just me. No-one wants to face a capture every time they click 'like', leave a comment or make a post. I understand your concern about the platform, indeed I share it. I just define it in different terms. To me the problem is not bots verses humans, but good actors verses bad actors. Both sides have bots and humans. I think we need more bot good actors to use downvoting to counter the bot (and human) bad actors. It is a bot arms race and downvoting is the weapon good actors have to discourage bad faith actors.
One thing that bothers me as well is the pandering to whales (a lot like in the real world). If a whale posts something, anything, every plankton in the pond rushes to upvote because they know the post will get huge rewards regardless of its quality. It becomes self-fulfilling as most whales are friends, having been around for long while and to a certain extent upvote each other.
I'm less bothered by the fact that "trending" shows posts that have been upvoted by bots, it's a bit like advertising. If you believe your post is really good, what means do you have to show it to people ? You have "promoted" but there you waste your money. And then you have the bots and trending and you mostly recoup your money.
Now if you bring to trend a shitty post maybe there should be a parallel mechanism to gather opinion, that would be less directly linked to money.
For instance an additional "like" or something. You pay for advertising and I am grateful to @yallapapi and @suesa they did so because otherwise I woldn't have discovered their very nice posts. But people could also indicate that they didn't appreciate your post, without linking that to the rewards, at least in the beginning. Regardless of how much @suesa post made (I don't think $800 is fair but it's not outrageous either. It's not $15 000), we could have a parallel signalling: I enjoyed / didn't enjoy that post.
With this information, we could afterwards implement a more fine grained mechanism to discourage shitty posts somehow.
The key message is: the system needs to be made more sensitive, fine-grained, accomodate more varied inputs, not just monetary rewards. Communities is probably going in the right direction
It is precisely the advertising aspect that bothers me. If all the posts in Trending paid to be there, what makes Trending different from Promoted? I saw another comment that suggested all bidbot posts should be listed in Promoted rather than trending for that very reason. One could argue that self voted posts be moved there for similar reasons. Other than 'New', all the tabs are really just showing posts that paid to be there.
Indeed ... could be the right thing to do - move bid-bot upvoted posts to Promoted ...
There is also a difference between bidbots (as they currently work) and advertising. When you pay for an advertisement, you have no guaranteed return, no-one is obligated to buy your product. There is an element of risk. Bidbots should also have no guaranteed return. However, if we refuse to downvote them, then we are basically allowing them to force us, as a community, to purchase their content (via the reward pool).
It is precisely the lack of risk inherent in them that means they are scamming everyone else. This is why I do not see a problem in them being downvoted. Introducing that element of risk will make people more cautious in using them, especially for low quality content.