You are viewing a single comment's thread from:
RE: Hardfork Update - Curation being added back to comment rewards, revised curation and reward curves for both posts and comments
Thank you for the update and continuously parsing GitHub for us dev-challenged folks! This is great news. There's been a lot of antagonism towards Steemit, Inc. of late. I hope people give them credit where it's due.
I'm OK with the multiple pools. I can see why going modular can be beneficial for future initiatives. I recall Ned propose short-form reward pools, Disqus-competitor solution etc - they have gotta be considering these. However, I do feel 38% is far too high for the comments pool. We'll see what the stats say, if there's an order of magnitude increase in the quantity and quality of comments, without a decrease in post activity, I'd be happy to be proven wrong.
Modular is fine, that's a coding issue. You can have modular reward pools with only one such module instance existing initially. I feel we are much better off evaluating the change to the reward function (which is likely to be a very significant change, with far-reaching consequences, some as yet unrecognized) before also making additional changes and significantly muddying the water.
I'm going to go out on a limb and guess there would certainly be an order of magnitude increase in the quantity of comments with the proposed split (ignoring the effect of reward curve). As for quality, well...
Comments that lack quality can simply be flagged. I hear some of y'all are pretty good at that.
Vote power, as well as unrewarded effort, is extremely limited. A few posts can easily be flagged. Thousands of poor and bot-generated comments in practice can not, and even if they were this runs up against the problem that doing so push rewards right back into the comment pool, increasing the incentive for more "replacement spam".
There is no way around the fact that arbitrarily choosing a percentage is almost guaranteed to result in a split that is either too high (attracting spam), or too low (starving comment-focused modes of usages). Given the current user base and UI, short term it would likely be the former.
Perhaps. I think we'll see relatively quickly and can move on to our next round of blockchain whack-a-mole^H^H^H^Hiteration
Exactly, that's why we should try one substantial change at a time, so we can actually understand the effects before pulling out the hammer yet again.
With essentially every proposed hard fork so far (at least those implementing rule changes), the feedback from the stakeholders and community has strongly favored fewer changes being pushed out together as one big package. Sometimes that is impossible or impractical, but here we have a case where there are two completely independent major rule changes (flattening the reward curve and splitting the pool) that can easily be rolled out, voted on, and evaluated separately. I see no good reason not do so.
@smooth (nested reply) I would agree except that we don't have unlimited time. We have to compress the timeline as much as we can.
@smooth Fully agree with 1 change at a time wrt important items such as power and funds distribution; absolutely the only way to find out what the results are. @sneak When rolling out multiple changes a time, what are your ways to determine what are the causes/effects?
@edje By talking to our users, same as always. :)
Maybe we can make a new reputation algorithm, make it part of consensus, and make accounts below a certain threshold ineligible for rewards. Thoughts?
Reputation systems re extremely hard. There are very few that ever turn out to be strongly abuse resistant (and there are other hazards as well, some already seen in the existing rep system, such as creating an entrenched system perceived as unfair or unwelcoming where those with high reputations are strongly favored by the rules at the expensive of those who are new or less well connected). Directly attaching rewards ups the ante significantly. I won't say it is impossible, as I rarely do, but why not try easier approaches first.
I'd even say take away their posting rights, or at least limit their bandwidth severely so they can only make one post/vote per day or so. We have seen some discredited accounts continue to spam randomly and incessantly, often with hateful messages. They give up after a while, but it's an unnecessary inconvenience. No new user posting an introduceyourself message wants to be greeted by a message, even if hidden, saying "Steemit is a scam, X whale is evil etc. etc.". (Yes, I have seen new users respond to these.)
As for Reputation, the current system simply doesn't work. Anyone who makes a lot of posts and gets voted by the same people over and over again stands to gain a high Rep. Conversely, any one could single out a new account, get some help from their friends and repeatedly flag it to oblivion; even with relatively low Rep or SP. As smooth points out, I'm not sure if there can be an effective Rep system, but I'm sure we can do better than this.
Off-topic - I'd love to see you comment more! Not just about Steemit stuff, but in general. Your comments about Vaccination yesterday were a tour de force of rationality - much needed on Steemit. :)
@liberosist Thank you for your kind words.
The problem with the system you propose is that it assumes 1 account per user. It is not resistant to a person who registers a thousand accounts.
Also, I am ideologically opposed to anything that limits posting rights. There are better solutions that involve letting people choose what they want to read via UI.