Voting System Overhaul

After consulting with PJB I decided is worth posting this:

If the idea is to create the best content rating why, not use similar methods as Facebook and Google use?

In the current system, you can’t accurately determine collusion and the incentive for the best curation is not achieved.

If YUP were to make a system to produce the best possible curation, it would not base the voting system on the assumption that free voting will generate the best results especially when incentives aren’t properly aligned, inner communities exist, and the statistical markup of all curators could be biased.

As such, I would suggest a system where there are 2 phases, in the first phase anyone votes anything they want and will not get rewarded.

The second phase is where the system will randomly pick a number of batches from those votes to be curated, and then send those batches to the curators, curators then will vote and create quality consensus, participating in consensus should be rewarded and the order of participation should not matter much.

This way you achieve 2 things:
1: the distribution will even out a bit and between active quality members and will reduce the pressure on the token.
2: It will create clearly better votes for the network since the incentive will be to do quality voting and be an active member.

This is inspired by how FB and Google do and pay their content rating/validation/curation.

In the current system, you pitch users against each other since half of the user base is in countries known to have a very poor internet connection and maybe users have crappy devices and will FUD that everyone has a bot when the reality is that they received notifications with high delays.

Here’s a poor-man diagram of such a voting system:

I understand this could be difficult to implement, but as a long-time vision at least I think is worth pursuing.

It would be great if 90% of the talk is not about rewards every time, having a simple and robust voting system will reduce this kind of criticism.

As a short time solution maybe just stop rewards that stem from popular accounts, so then even if someone will rate first will get 0 rewards it would be a curious experiment to see how many votes the “popular” accounts will receive then.

But my personal belief is that isn’t a solution long-term, longterm the system must find a better way to incentivize quality votes, which I think is almost impossible if the domain of voting is the whole possible combination of URLs which will always be bigger than what’s possible to curate.

This is my 2 Gwei on the matter, and I would like some feedback.

Thanks.

6 Likes

It is a very interesting proposal. It’s clear that the economy of the platform is not sustainable in the future, and the voting system has a lot to do with that. It’s great that everyone contributes their ideas. We really are a proactive community!

To understand your idea correctly, now there would be three roles: creator, curator and user, right? I understand that what we currently call a curator, would be a regular user. And you introduce a new figure, which would be like a team of super-curators who supervise the quality of the votes. Is that correct?

I find it positive that the factor of speed or the order of the votes does not affect the reward.

The solution of treating certain accounts differently from the rest does not seem very practical, since who would decide that and under what criteria? I think that would only transfer the problem to other accounts and in addition we would need to have someone from the team alert to update that “black list”.

But the truth is, there are many things about your proposal that sound very good.

:clap::clap::clap:

2 Likes

Hey, thanks for the feedback.

It would not introduce a new role, as you say that would be a bit discriminatory. In such a system the curator is any simple active user, ( by active users means someone who submitted posts in the last x days where x could be a chosen arbitrary number like 3,4,5)

The only difference would be that rewards would be granted only for the votes to the posts that are sent from the protocol to the user/curator.

So every active user daily would have a random interval[x min, y max] of paid votes, which means one day I can have 18 paid votes, and the next one 22 paid votes, then after a day 19, etc, is all random to prevent any type of smart collusion.

The system would send those posts to build consensus, and the posts are a randomly selected subset from the unpaid submitted post.

For example, if I like a URL titled: “Qt framework documentation”, there is a chance that another user will receive that post to build consensus on it(rate it). There is even an extremely low chance that the user itself that sent the post(unpaid), will receive the same post later to rate it(paid).

The idea is to align reward incentives with quality voting and disrupt in a provable mode any kind of rewarded collusion.

Because for now, you can’t reasonably prove that a score is not built out of collusion or out of the desire to get greater rewards and ignore the quality of your rating.

6 Likes