You are viewing a single comment's thread from:

RE: Downvote Pool Deep Dive

in #steem5 years ago

The collective body that's deciding which version of code to run has a fiduciary responsibility. To some extent, that includes everyone, but in reality, it's a relatively small group of people.

You can't know with certainty what will happen without trying it, but you can gain an increased level of confidence by doing formal analysis of the change, and developing research-backed theories that are more reliable than our intuitions. You can also increase your level of confidence by running simulations.

Flag wars, much like self voting cannot be countered through code changes

Where is the evidence for this? If self-voting can't be countered through code changes, there's no point in implementing the change. As suggested in item (iii), however, I suspect that they actually can be mitigated by realigning the voting incentives.

Sort:  

They don't have any such responsibility, what they are responsible for is running the nodes/servers and updating the price feed (ugh), all else is a matter of desire not of necessity.

I don't understand what you want me to provide since you seem to want me to demonstrate a negative proof. There have been a lot of suggestions and, much like your 2nd price action, they weren't considered fully in how a bad actor would overcome them. In the end, they all have one common thread: tax/burden the community and obfuscate undesirable behavior, unwittingly. You seem to think that by splitting stake up to jump over the hurdle is making abuse more obvious, but frankly I don't know how you can reason that, how it's easier to link multiple accounts together in a scheme than to simply recognize one abusive account.

Again, the way to realign incentives is not through Law but the enforcement of law, likewise, the flag enforces the law of "thou shall not voturbate" because implementing any such law is only at the expense of everyone who doesn't abuse self voting and it barely can be considered a hurdle to those determined to do the least for the most.

Posted using Partiko Android

I don't want you to provide anything, and I'm not the one that needs to be convinced. I'm content with either decision. I'm saying that the people who decide which version of code to run should demand more than an intuitive demonstration that the change will make things better.

What does better mean? In curation, "better" means that it is more likely to rank a set of posts in the correct order, according to user preferences. So, it seems to me that the witnesses who will run the code should ask whoever is proposing the change to provide some level of evidence that the post ranking after the change is likely to be more correct (closer to matching user preferences) than post ranking before the change.

Again, the way to realign incentives is not through Law but the enforcement of law

Your argument seems self-contradictory. On one hand, you say that the rules don't matter - and we need to just depend on curators to downvote, but you're making that argument in support of a rule change. If we can't solve the problem of incorrect ranking of posts by changing the rules of the game, then why are we having this conversation at all?

They aren't asking for any demonstration because there is no testing or scoring for such a demonstration, the only way to get data, the. Only. Way. Is to try it.

Curation isn't what you make it out to be by far, it's another word for saying review / rate. That's it. It's not about correct or incorrect ordering, it's about "I rate this as x dollars". After all let's not kid ourselves that we are curators at a natural history museum or the like...

What this proposal is about is not "fixing curation" but incentivizing policing the network, making it not profitable to self vote, bid bot, or circle jerk, make it cost little to nothing to counter plagiarism / fraud etc..

You're asking for some level of evidence that this proposal is going to make users more likely to rank a set of posts in a correct order but the preposterous notion isn't only that this proposal is not about that or that there aren't any metrics to measure "a correct order to user preference" (kinda oxymoron, as preference and correct are contradictory ideas, one is objective t/f and binary correct or incorrect, the other is subjective and variable) but that there isn't any place or system that could be used to determine that besides steem.

My argument is not that rules don't matter, I'm saying that having rules without enforcement is redundant, and putting it into code only burdens everyone, much like your suggestion, while those that seek the least amount of work for the most profit will exploit it either way, if they have to go around the obstacles you place they will. By allowing people to enforce rules that needent be even expressed or explained but are simply "unwritten" (ie George Carlin's unwritten rules of the road) then you don't have to deal with figuring out exactly who is who as they are going around your obstacles. I hope it makes sense, I'm sure there are others who could better explain the conundrum.

Posted using Partiko Android

The point of a content curation system is to produce a ranked list of content. Yes, from the voter's perspective, it's just "I rank this as x dollars", but a good content curation system will aggregate all of those individual decisions into an ordered set that approximates the actual combined preferences of the users so that readers can quickly find things of interest.

In that context, it is possible to analyze the strengths and weaknesses of a particular voting scheme before injecting it into the blockchain.

You should read A Puff of Steem: Security Analysis of Decentralized Content Curation. There is much to learn, and it suggests several techniques by which the strengths and weakness of any proposal might be quantified before slapping it into the running block chain.

It's very simple to calculate. What is the Median power of an average voter? What is the power of a community vote? Of a bid-bot? Of a whale?

Case closed :D

There's no system or place to analyze that. It's preposterous to think that this could be tested anywhere but in the real world.

Posted using Partiko Android