RE: A first thought about improving curation
I that might constitute the verbal component for summoning me.
From a game theoretic and social network design point of view, your suggestion has three significant issues:
The suggested scaling doesn't actually improve the implied ranking more effectively than just nakedly looking at SP.
Which you probably shouldn't feel terribly bad about, because 99% of the other systems I see people suggest have the same problem.
Consider, that the more established that any author is on this platform, more likely they are to have a higher SP than someone else who has not been here as long. Simply by dint of the fact that they've had more time to accumulate SP. This doesn't actually have anything to do with the quality of their posting, just their tenacity, but will get to that here in a moment. For now, think about the fact that systemically, an established author who has been on Steemit/the Steem blockchain has an extreme likelihood to have a higher SP, natively, that a new author. Which means that your linear scaling is just taking the SP ranking and repeating it.
SP is not distributed in a linear fashion, it's absolutely true. It is almost aggressively logarithmic/exponential. But if you take that log and linearize the curve, what falls out is still ranked in exactly the same order. The numbers are just closer together.
It assumes that the best curators are also the best content creators.
That's just patently false. On the face of it. It's a false premise. In fact, a few minutes' thought will lead you to that conclusion because good curation takes time and could content creation takes time, and those are not tasks that you can pursue simultaneously.
A good content creator spends time creating content. A good curator spends time finding content with the crappy tools on the Steem blockchain, considers what the best volume of bet for any given piece of content is (and that's really what an up vote is for the majority of curators on the platform who want to use that as a means of making money), and does so as quickly as possible.
Believe me, I know that creating good content takes many hours to get right, and those are not hours that you can spend being a curator.
And then there's the next point, which is neither mechanical nor mathematical but philosophical.
What you are rewarding is not what you actually want.
What this system rewards is properly gassing and betting appropriately on what other people will think is good. It has nothing to do with what you will think is good, and nothing to do with what I will think is good, but everything to do with what people who neither one of us care about and have no contact with think is good. Those people will be rewarded by your system, just as they are by the current system, and you will get exactly what you reward – more things which have nothing to do with what you think is good or what I think is good.
You get what you reward.
As @por500bolos says somewhat more clumsily below, by implication, good curation, useful curation, is rooted in people reading content – but moreover people who share my tastes reading content and being able to signal that such content is good. As long as, and entirely as long as, the means of assessment of curation rewards is collectivist, as long as it remains effectively of bet on what other people are going to vote up and has nothing to do with real content that I or you might care about, that's both going to affect what those bets are laid on and what content is created – because creators really want to be rewarded for that act of creation.
That leads directly to the situation that we're in now, where bots really do have the efficiency edge on human interaction in terms of curation, because nobody cares about the content. The content is unimportant to the mechanism. Your suggested system would just continue that, because it doesn't care about what individuals want or like, it assumes that everyone does and should like the same thing.
When you base your means of presentation on the assumption that the individual doesn't matter, you get results that state that the individual doesn't matter, because the individual doesn't matter.
You get exactly what you have declared that you don't want, and it's no kind of surprise.
Now, I don't agree with the other suggestions that Por makes along the way – or at least most of them. Requiring 100% of an up vote all the time just means that individuals would be limited to a very, very small number of votes, and that's really insufficient for signaling to a system where what you desire and expect is to have more content than someone can truly consume in a day to have any kind of discovery. A better mechanism would be to say that you can have as many votes as you want a day, but you only have X SP which will be distributed over those votes every day – and then you don't have to make the decision of how much to drop on any given post. Much simpler. Likewise, isolating activity on the blockchain based on what interface someone uses to post versus what someone uses to curate is both stupid and self-defeating. It means that no tools will ever be developed, and bots won't be stopped at all because they will be able to mimic any platform they like. It only impedes legitimate users. So that's kind of crap.
Requiring that someone actually read content is, at least, putting the heart in the right place, but it is mechanistically and logically impossible. There is no way to prove that someone read a thing, and honestly you wouldn't want the kind of intrusive surveillance operation that it would take to validate that they had. Nobody wants that. So really – I guess all three of his points are kind of garbage.
But he means well. And he's right about the idea that this is a social network and that content should be curated with an eye to the value to the individual as content, not as a complicated minigame. That I am absolutely in agreement with.
What we need is a system which is rooted in taking the signals that individuals give to the architecture by way of votes and provides an individuated lens onto the blockchain which, given what the individual has indicated about what they like and who they trust, then presents them with a view which is unique to them or at least customized to "people like them" and which is likely to be content that they would like to vote for to signal that they like it going forward.
Without that central mechanism, without that hinge, no plan will be useful nor will it give anyone who cares about content, who cares about seeing things that they like, those things. It's impossible. Logically, mathematically, and philosophically impossible.
That won't keep people who believe in the authoritarian, collectivist idea that everyone knows better than I do what I should like from trying to make that vision a reality – but it does mean that I can take comfort in the fact that they will fail at every turn, so I've got that going for me.
I feel like I've given this lecture once a week for months on end, and that maybe because I pretty much have. And I'm saddened by that. There are a lot of people who have a lot of energy and a lot of desire to get something better than what we have, but there obsessed with the idea that more people liking something means it's better with the assumption that their idea of "better" is somehow more functional than everyone else's.
Patently not true. Observably not true. Empirically not true.
And yet it comes up all the time.
One day things will be different. Just not this day.
In fact, not so small number of votes. 10 full 100% upvotes daily are enough to cast before my eyes. I've been doing this since the beginning and it seems more than adequate to reward the good content of the ones I follow and also go hunting to discover those I do not follow.
After all, how many posts can you read daily from top to bottom to really appreciate its quality and decide to upvote it and reward it if you are also a content creator yourself?
Yep, I could agree with that. That could be another novel approach to achieve similar results.
As an authentic human test of decentralized 'Proof of Brain', having to login to different interfaces in order to have access and truly be able to interact with the content created within each dApp/platform independently, this would reinforce the impression that we are being curated by living beings who are not lazy and truly read, appreciate and value our content.
Maybe this way things would be a little more difficult for those external platforms of automated self-voting Trails, Guilds, VoteXVote Bots, etc. competing against humans to obtain more significant incomes through the curating activity.
I disagree. All those new tools, platforms and dApps can continue to be developed. As they have been doing it. As different communities with different specialized niches of interest. The blogger platforms for bloggers/writters. And dSound, dTube, dLive, etc. for Musicians, Podcasters, Vlogers, Visualists, etc. The same with all the others, each specialized in its own niche.
Yeah! admittedly humans are lazy. But then, they wouldn't deserve call themselves curators if they are not willing to go the extra mile as to let themselves be seized the gains in curation incomes by a band of brainless bots and rapist automatisms which they will always be on the lookout to take the lead.
Nah! there is no need to prove that someone has read a thing. With just avoid that the posts can be voted from outside the user's blog, without having clicked on them first and allow them to be fully deployed & displayed in the browser, we would make life more fucked for any internal/external automatism, dumb AIs and brainless bots to play the Curators game contributing to liquidate the true social sense of the platform in relation to deserved rewards. So, there is no need of any authoritarian intrusive surveillance operation to assure that either.
Now we are talking Hahahaha. };)
Emphasis mine. That works for you, but for me, partial vote strength has been a huge boon. I do a biweekly curation post, featuring anything I found personally interesting. There's usually about 40 to 70 articles per post and I try to upvote all of them. Combine that with the number of comments I get, which I like to give at least a minimal vote, and I'd never have a voting power above 1% ever again.
The other use case which does not reflect your reality is that of mega-curators like @steemstem and @utopian-io. They are arguable some of the most effective crap-sorters and their entire model of work would not be possible without fractional votes.
From my perspective, those are two of the biggest crap sources most of the time. Utopian's rules basically mean that it is far more effective to write a cryptocurrency-adjacent crap application, hype it up as much as possible, and then do little bits of update which don't actually represent anything meaningful, then it is to write content which is intended to be evergreen and educationally useful. Steemstem is an almost incestual curation group, promoting writers within their small circle of acceptable voices for the most part and occasionally finding posts which are on STEM subjects elsewhere, but nowhere like consistently or well rewarded.
But that said, I am perfectly happy with them doing what they do on the blockchain. Thrilled, even. I want there to be a multiplicity of communities who want different things and have come together however they like. It thrills me to the bone that there are groups with whom I have serious disagreements in terms of the content that they promote doing what they do. In theory, that means that there could/should be communities which promote, curate, and reward content that I'm interested in.
Except that's not really the case, because the tools to do so don't really exist or are very difficult to deploy. In fact, in the case of Utopian, the best part of that communities toolset was the web-side UI that would present only Utopian-content, meaning that if you wanted Utopian content, split up by what type it was, it all you needed to do was go to their website, click on a header, and there it was – easy and simple to examine and stream and enjoy.
It would have been very nice if Utopian had devoted themselves to, as part of their mandate, making that web app mobile and easy to set up for any community that wanted to curate its own type of content. That would've been great.
That never happened.
We are in absolute agreement that the ability to scale voting percentage is a great tool for all users and allows a multitude of different ways of interacting with content on the blockchain that would not otherwise exist. We are 100% together on that idea.
Well @effofex, then you are clearly doing a lousy job. Because if you do that curation post 'biweekly' you should read, digest, curate and upvote no less than 140 posts at least having 10 votes daily at your disposal. ¡Grab your abacus! };)
Upvoting comments is a total different beast to contemplate here. But I can tell you that you don't need either become a willy-nilly slut spreading kisses and hugs everywhere to the point to deplete your VP until levels of 1%.
If they are such effective crap-sorters in their entire model of work, then I suppose they wouldn't have issues granting only 10 upvotes daily on prominent and really outstanding posts only. Yes, due their massive high SP I concede that they can't upvote good content at 100% of their VP. But surely they can cleverly upvote at least 20 posts daily at 50% of their VP being truly selective in what really deserves such sort of high rewards. And the next day 20 posts more and so on indefinitely.
You know, I really can't decide if you're being a deliberate troll or just an idiot. It pains me, not to be able to distinguish the two, even though Poe's Law has been true since the beginning of the Internet.
@effofex was very clear when he said that the biweekly curation post contained things that he found personally interesting, and that they usually contain somewhere between 40 and 70 articles per post, and he tries to get up votes on all of them.
Any sane, intelligent, experienced person would recognize that firstly that's a lot of posts, and secondly – they don't all show up on the same day. And they don't all show up on different days. If my experience is anything to go by, they clump, and some days you have a whole pile of them and some days you have nothing and all the time you spend reading doesn't turn up anything that you think is worth voting for. The narrower the niche, the more likely that is to happen.
And that's part of the reason that "10 votes a day" is a stupid idea. Just a ridiculously dumb idea. Some days you get 20 good articles, and the next day you may have five, and the next day you may have 30, and the next day you may have zero – and either you build a custom tool to keep track of all of the things that you've come across in the past week and automatically drop your votes at the appropriate time to maximize the impact, because that's what the system demands to be efficient, or you drop the votes when they happen and try to scale your vote percentages so that you are able to reward them when they happen.
Which, ironically, is a solid justification for a single-user vote bot. The system incentivizes automation, and so it gets automation.
They would if they get 20. And that's the problem with your idea – it assumes, and we all know what happens when you assume, that content is and will forevermore be limited, and you should only have a fixed amount of things that you can reward a day.
But that very architecture requires that you be prescient. You need to be able to see the future and know how many good articles you're going to run into in the next 12 hours. How will you ever know that? It's impossible. If I find an excellent piece of content right now, and I've already dropped nine votes on excellent pieces of content that I like, how much do I want to gamble that I won't find another excellent piece of content five minutes after this one? Because that's the gamble, that's the forced decision that your system would require.
If you want more vote selling, that's how you get more vote selling. And more bots running accounts purely for the fact that they would have more votes. And we've already discussed why you really can't stop bots interacting on the blockchain, starting with its unverifiable.
It doesn't take experience with game theory and game design to understand why this is a bad idea. All you have to do was walk through how it would actually play out in practice and you can see, immediately, where it fails.
Admittedly, this may be a talent that few people have.
Mm, sort of. It also accounts for cashouts, and using SP instead would encourage cashouts. On the other hand using SP would be computaitonally trivial, so that option remains.
I think you're misreading something, because there's literally nothing in here about quality of curators. We're not basing anything on the curators' rewards, but on the authors'.
Yes, and this is annoying. I totally agree with you that the whole white-paper idea that the most-desirable content will rise to the top through the wisdom of crowds is essentially useless.
That doesn't change the fact that we still feel the need to incentivize people to vote with some form of returning partial value through a gambling mechanism, though. This could be entirely separate from the content-delivery system, which seems like it's what you want to improve. Maybe we should try to get Pocket integration somehow; they do a really good job recommending things to me that I want to read.
But I think in general the problem you've identified indicates that voting shouldn't be used to deliver content, not that we need to improve the voting system so that it delivers content that you want to you.
Personally I enjoy finding things I like myself and would not be pleased by a Facebook-ish algorithmic determination of what I want to read, but presumably that could go on a separate tab that I could ignore the way I do Trending now.