How would you like rewards to be allocated?

in #steem6 years ago (edited)

Most of us would agree that the current approach to allocating rewards is a bit shit, where the bid bots and whale whisperers take home almost all of the reward pool, while most actual content creators fight for scraps.

So, what would you suggest is a better way to calculate rewards? You can be vague, or suggest precise formulae.

Snax is going to start rewarding Steem users from 19th May. Their team has been very receptive to feedback thus far, and made several changes as per popular requests. If there are good ideas here, I'm sure they'll tweak their reward algorithm accordingly. But moreover, it's a thought experiment that'll also let Steem developers know what users really want.

Assume that you have access to the Steem blockchain, and you can use any data within it to calculate rewards. It can be something simple like some combination of # of comments on the post, follower count, # of votes, pending rewards; or something radical.

My idea is to look at relationships between accounts, and reward those that generate engagement (comments) from a diverse set of users. Each user gets an engagement score. A user that's known to be engaging with a diverse set of other users will be able to influence the rewards more for whichever post they leave their comments on. If they start abusing the system by collusion or Sybil attacks, their engagement score will drop automatically. I don't have any details about how this can be implemented, but hey, I did say we can be vague!

Sort:  

Thanks for this important discussion. I believe there is no mechanism that cannot be abused. But, we can do much better than we are doing now.
I would in part keep the current formula, but include a soft cap of max rewards. If your post hits let's say 100 SBD just as an example, adding another whale vote worth 30 SBD should be worth significantly less. Also, although voting strength should still be based on SP, it should also include different parameters like user interaction, the diversity of voting (a certain % of votes for always the same accounts or yourself should trigger a reduction), reputation score (which should be calculated differently to correlate better with the actual reputation within a community), and expertise in a field (e.g. if you earned a lot of SP withing the "art" tag, your vote on posts using that tag should get a bonus).

Anyhow, this is all super-heavy to implement and would need some serious balancing work.

I have proposed using an algorithm that dispersed rewards based on Huey Long's proposal for income: no one should receive less than 3% of the median, nor more than 300%. This leaves two orders of magnitude to create incentive for quality content, but doesn't leave any content void of incentive. It also eliminates the ability of whales and bidbots to mine rewards, as the median is so small that massive stake has no interest is such pittance.

This would be unpalatable to everyone (we all want massive payouts, and like lottery players, fail to note that we dream impossible dreams) unless a means of gaining higher profits exists. I therefore proposed nominal dividends from delegating or directly funding development via the SPS mechanism @blocktrades has recently completed, and devised for the use of the @steemalliance foundation that has recently determined it's corporate structure, and is intended to fund development of the Steem ecosystem.

There is a fundamental difference in affect on the investment vehicle between capital gains derived from increasing the value of the investment vehicle, such as development is intended to create, and extracting profits from the investment vehicle, which is what bidbots presently do. What I propose is to promote the former as a means of enabling ROI, and discourage the latter.

Any way that is achieved will improve the current situation regarding Steem's market performance. Water flows downhill because that's what the code (physics) makes profitable in energy cost. Investment simply follows code, just like water.

Thanks!

I like this idea - a flattening of the reward pool. This is similar to a publication I used to work for as a video game journalist. Everyone got paid the same for their reviews, whether it was a romping, open world, 80+ hour RPG, or a whimsical puzzle game.

If something was longer form, or was to bring in huge amounts of traffic, or warranted a follow up article or response to an editorial, that meant more money (by virtue of more ads being served) - but could sometimes be a short extension of the original.

We can always reward comments more handsomely than posts in some instances, especially when the author goes out of their way to add real depth in their comments section.

Thanks @mayb for the tag! To be honest there will never be a truly ideal system that everyone feels is fair! Whales and bid botters will do what they will with their stake be it emotional or ROI motivation based! Getting that out of the way I would like to see a 2 fouls reward approach that could help distribute income better and encourage things we want like engagement

  1. Front end IE busy rewards a user based on regularity of post, size of follower base and follower base vests! This encourages regular posting and growing your audience actively

Then like steem UA based on your comments, engagement and upvotes form human accounts you get rewards based on that

System Still has room for abuse but what I really just want to encourage is regular contribution and engagement!

Posted using Partiko iOS

There's one fault with Steem-Ua - I don't think it filters out its own posts when ranking your engagement, or those from well known curation communities such as curie, c-cubed, or ocd.

It's a good problem to have when you get a jumble of those interspersed in your own comment section! I used to use busy for a little while, but haven't touched it in ages, in favor of steempeak.

That voting technique that you have described that they use sounds like an enticing model for genuine content creators.

It's a competitive place for front-ends right now!

Something that folks haven't seemed to realize about reputation is that it is ultimately subjective. Each of us assigns reputation to others based on our personal values and experiences. Any blanket reputation metric cannot do this.

Accordingly, I reckon that the metrics on which reputation is based on should be ranked by individuals, or the reputation number simply plugged in by individuals based on their personal subjective assessment. This means that no universal reputation score would actually exist, but that each account would assign the reputation score relevant to them personally.

When your buddy recommends a contractor to you, your assessment of your buddies' reputation is the basis for evaluating his assessment of the contractor, and that's how reputation works in the real world. That's how it can work on Steem, rather than simply being the result of gaming algorithms.

If people you hold in high reputation hold another in high reputation, that is a useful gauge of folks unknown to you. The opinion of people you do not hold in high reputation may well be contrary to your own assessment. For example, folks that Trump or Clinton hold in high regard are all people I will not. This makes reputation necessarily only as useful as it is subjective, and this particulary is a poignant aspect of a real social media platform. We are not algorithms or bots. The values of algorithms or bots are not those we hold, and it is our actual humanity and society that is the point of our engagement.

Reputation is only useful to people insofar as it is actually the result of our subjective opinion, and this makes it impossible to assign objective measures to. Bots and algorithms cannot assign reputation, except per our individual use of them per our preferences. A mechanism to reveal reputation of someone unknown to us but known to folks we follow based on the assessment of those we follow is useful as a means of recommending them to us, or discouraging us.

Money is not reputation. UA is an attempt to reflect that fact, but it is not able to reflect the fact that the esteem of folks I do not esteem is actually contrary to my esteem, and I don't see how an algorithm might reflect that fact. The truth is that the reputation of any person is different to each other person considering them, and no universal reputation score has real value.

Thanks!

I definitely agree with your statements. Reputation depends so much on your personal level of trust with an individual on the chain. For instance, some people in the Steem Monster community have previously asked for me to act as escrow, as they held the view that I could be trusted to not take the money and run.

However, if that same trust needed to transfer into the steem silver gold community, I probably wouldn't have that same level of engagement and "trust-worthiness" to them,

I'm a big fan of the 1 person = 1 vote, and for you to be able to "review" people's profiles based on your feeling for their reputation. Reputation is a funny thing in that regard, however; and could be gamed. If people pay for votes, there's surely someone out there that will pay for reputation.

That's what a lot of political activists end up doing. There's no perfect system. I like to use my gut. The reputation number is nothing than a number.

Actions speak louder than numbers. :)

Thanks for your well considered response!

Can you elaborate or link something about this '2 fouls reward'? I don't know what that means.
UA is nice but I never really understand the way it works and those standard comments could provide a bit more help on how to follow their suggestions, like: "The followers like your work Try reaching more follower!"

Yeah, but how so, ua? :)

I think what he means by 'two fouls' is like strikes in baseball, but I am not reading his mind. Hopefully he'll expand a bit, because I also would like to know, as he seems to grasp that a universal reputation metric is void of actual meaning to real people, who hold people in regard based on their subjective values.

I'd love to see a page-impressions model. Which pages, which posts (regardless of age) generate the most traffic to the platform? There would be a good market for Steemit Inc to then use their stake to provide votes to that content that captures an external audience, especially now that they've got ads on their front end.

Engagement, however, is a huge thing - but there has to be a way to determine real engagement versus the typical "nice post, deer" that we all remember from the early days. With the change to RCs, that isn't as common, but there's a lot of really hollow comments that don't provoke discussion on a lot of posts.

The sad thing is that they aren't comments like "This post made me so speechless, all I can give you is a vote." They're the sorts of engagement that don't provide any evidence that the post has even been read.

An interesting way, like you said, would be the relationships between accounts. I had an incredible interaction with @mayb on a recent post of mine, and ever since then, by following them, I've uncovered a few other Steemians who engage in a deep, meaningful, and intellectual manner.

I guess that's the sort of people that I'm chasing, and the sort of content that I enjoy.

The fact that this place is somewhere genuine engagement and intellectual satisfaction can sometimes be found makes it feel like a much more mature environment, and I try to reward that wherever I see that happening.

I was so happy to read your comment that I upvoted before even checking the context! :D So, mayb read it first …

Now I've read it all and I love the idea. One of my latest posts was about the commenting behavior which is definitely improvable. I agree with both of you, @holoz0r, @liberosist. I want to find those people who truly engage and amaze me with their different opinions. A lot of the steem experience, for me personally, is the part of leaving my comfort zone, learning new things and I like all those challenges. We all can profit from communicating more. Monetary and in our personal growth.
Below my above-mentioned post @frankbacon left this hilarious comment:

… still looking for that one interface that has it all! Show me interaction, show me where the party is right now, notify about what my steem friends are discussing when I come online.

Perhaps a trending page that detects genuine discource in the comments section of a post might do.

Do you use Steempeak? It's an excellent front-end. You can save drafts, schedule posts, and save your favourite topics. For responses, I use a combination of Ginabot via discord, partiko (when I'm not at the computer) and most importantly, the stuff that lives between my ears - the original plan for Steem.

Proof of brain. :)

I am using … look for yourself … Bildschirmfoto 20190515 um 12.58.05.png

And I love it. Btw, @chekohler was going to write a post this week to make some suggestions for steempeak. So, he should also be part of our communication, I feel.

But yet you're not using the night mode! :) @jarvie and @asgarth are really approachable as some of the driving forces behind steempeak. They're done a phenomenal job that has gone very unrewarded to date.

I am also a Steempeak user, and use the beneficiary mechanism to pass 5% of my author rewards for each post I make to Steempeak. It's easy, and hope lot's of folks do the same.

SEE?!? That was what I forget to mention in my last comment! I was going to say "but there's still a lot I haven't fully explored in the dapp. It's simply not that neat&clean as assbook or twatter. Nightmode??? FTW! CHECK!

Unfortunately, I've recently reactivated by assbook, asstagram, and I never left twatter. It's all part of a ploy to try and get some more people across to this platform where they may actually be rewarded in a minuscule way for their content, words, and opinions.

It'll be a slow burn, and I've tried before. I'll try some more. It's much easier to turn a creative, thinking, breathing thing into a crypto person than it is to turn a normie into a crypto person.

most definitely. It takes the will to learn and the creativity to overcome frustrations that might appear in a phase like that where everyone is still figuring out how we want to 'manage' this whole community. I believe in us. First, improve the blockchain, then the world! 🚀

That would be a far more useful trending page IMHO. Still, given the pandering rampant on Steem, the comments of folks pandering would still skew such a metric, albeit to a lesser degree than payout.

I reckon that we should each be able to create filters that recommends content to us, and that a real world reputation metric that is truly subjective would be a great utility for that purpose. There's probably not a better gauge than you suggest for an outward facing trending page, as folks not on Steem can't be availed of such a reputation metric. Engagement is probably the very best metric for such an outward facing trending page.

Thanks!

Edit: also, your parrot seems broken, you pirate. =p

I'm yet to see a comment earn more than the post it is linked to for providing additional context or opinion; that isn't charged by a self-vote army, or some unscrupulous vote trading. That would be neat to see. Perhaps the "value" of all comments on a post relative to the actual post itself would be an interesting metric to show true engagement.

i.e post is $1.00, value of votes for comments is 0.20c, votes are from other parties that are not the post author or comment-leaver. That could have an engagement score of 20%, or something. Just thinking out aloud. :)

Regarding the "parrot": It's a chicken! The pet of my soon to be mother in law. It is named "Special". It once tried to wander into a pre-heating oven on its own accord. She's a silly chicken.

I have seen comments that organically were upvoted higher than the OP, but it's not common at all.

Chickens are the most vicious predators I have ever met the gaze of. I am damn glad to be larger than them. Prolly make good pets.

"The sad thing is that they aren't comments like "This post made me so speechless, all I can give you is a vote." They're the sorts of engagement that don't provide any evidence that the post has even been read."

I find it a bit funny that the comment you refer to is exactly an example of comment that doesn't provide any evidence of actual engagement.

Thanks!

Irony is a great mistress. :)

I don't have any details about how this can be implemented, but hey, I did say we can be vague!

That's the whole problem, pretty much any algorithmic system that humans can / have devised can be abused on some level through bidbot-esque behavior. Some would argue that algorithmic curation with token rewards is a broken model in itself, and an even more radical alternative (not based on votes and steem power) would be the only way -- but at that point, it would be a different blockchain entirely.

At this point I feel like the Steem experiment is already in orbit, and we're just going to see how this concept plays out.

Well, we're still in beta, and changes will come. Major changes are constantly incoming, and I don't see any real limits on the potential of changes to radically alter the platform, in both beneficial and detrimental ways.

We're in orbit, but development is ongoing and is having enormous impact on the evolution of the platform and community.

Seems engagement is largely a function of VP and willingness to upvote comments, as many opportunistic commenters seem to flock to these accounts. Any type of algoritmic rewarding/voting is bound to fail unless true manual curration is factored in as the primary variable. A secondary variable that I think is usefull is the non-usage of bidbots. Aggressive bid bot users are usually easy to identify. They have high reputations that truly don't match their VP status. I would suggest some sort of reward correction based on the VP to raw-reputation ratio.

You're spot on that there's no substitute for manual curation. That said, our goal here is to do the best we can with an algorithm, and I'm sure we can do a lot better than what currently exists.

Excellent post and ideas. I also think that page view metrics should be used. The simple fact is, people should read and reward if they enjoy or if it's considered thought and effort has gone into the post. Engagement creates more users and is the only thing that will drive growth and set us apart from all the other 'content providers' on the net.
System will never change though because there are too many greedy people with too much influence and too much to lose if they can't shit post their way to as much cash as they can.

Posted using Partiko Android

Unfortunately, there's no way to determine page views metric on the blockchain.

To listen to the audio version of this article click on the play image.

Brought to you by @tts. If you find it useful please consider upvoting this reply.

Honestly, if reward allocation was based on metrics like views/blog (only counting with recorded interaction like page scrolling and time spent), length of post (word count), pictures, structure comments (and length of comments), and an author reputation system that was actually based on real engagement like time spent on others blogs as well as how many posts they put out, then we'd at least have a steem that would more often reward based on quality.

the current approach to allocating rewards is a bit shit, where the bid bots and whale whisperers take home almost all of the reward pool, while most actual content creators fight for scraps.

I absolutely agree that this is the case with steem currently. I've had to reduce my time spent in community activities simply to be able to focus on keeping a blog presence on steem while earning from freelance writing elsewhere. Posts and comments are all I can afford to do here and that's directly proportional to the reward system. I'm still hopeful, although less so than I was, that steem might take off again with another bull run. So I keep doing what I can on here, while earning what I need to live elsewhere.

Posted using Partiko Android

It is useful to note that water flows downhill, and that is because the code that water is governed by makes that the default operation of water. This may seem facile, but it's why rewards flow to substantial stakeholders on Steem. It's the default response of rewards under extant code. Whale panderers are just flowing downhill.

I have proposed changes to code to make funding development more rewarding than rewards pool rape (I go into more detail in reply to @sco above), and am confident that the problem stems from the developers of Steem not being seasoned investors understanding the difference between gaining capital and profit. Funds that derive from increasing the value of the investment vehicle have the opposite impact on the investment vehicle than funds derived from extracting value from the investment vehicle, which is what extant code promotes.

Dunno how Snax can impact this.

I am sure that this is the fundamental reason bidbots exist, and only creating a dividend stream from promoting development while discouraging rewards extraction can change that. If you can envision a way Snax can do that, I reckon that will improve the market position of Steem.

Thanks!

Coin Marketplace

STEEM 0.19
TRX 0.18
JST 0.033
BTC 87757.32
ETH 3103.63
USDT 1.00
SBD 2.75