Definition of a Whale and a whale-like group and the Math Behind

in #steem7 years ago

Definition

Whales and whale-like groups are those who have more than 800M VESTS in total, or around 384K SP in total.

Whales and whale-like groups have (unfair) advantages on the influence on posts.

The Math

To be simple, let assume everyone always have 100% voting power and vote with 100% weight.

Payout is decided by influence.
Payout_of_a_post = influence_on_the_post * total_reward_funds_in_the_pool / total_influence_on_all_posts

Total influence on a post is total_sp * total_sp + total_sp * 384K_sp, where total_sp is sum of SP that people who voted on this post.

When total_sp is below 384K, the influence is near linear. A minnow adds 1000 SP, total influence increase by around 1000 * 384K = 384M.

When total_sp is greater than 384K, the influence starts to accelerate. A minnow adds 1000 SP, total influence increase by 1000 * total_sp * 3.

A whale or whale-like group has the ability to accelerate the influence from zero.

The Influence

When whales are voting, they have most of the influences. Whales voting on one post means other posts / voters will get less reward. Whales voting on many posts means others will get even less.

When whales stopped voting, dolphins/minnows have more influence.

See the numbers (and remember that total curation reward is around 7K SP a day):

The limitations

The number of whales is limited. With this limit, the system won't scale, we will be unable to grow into a bigger user base.

Image someday we have one million new posts a day, 10 whales with 40 votes each can only vote for 400 posts in total, and probably many votes will go to same posts, what will the other 999,600 authors think?

As a member, you want we have more users so each SP your own will worth more in the future, or you earn more SP today but each SP will worth less in the future?

Conclusion

With greater power comes greater responsibility.

Sort:  

So why not simply put forward a proposal to limit the amount of MV that can be used for voting?

And find another way of compensating those with MV above that limit?

That way people can decide on whether to invest in STEEM based on the code and consensus. Rather than the arbitrary actions of a tiny minority.

Without discussion on steemit.com, a proposal submitted to github will be rejected. It's good practice to have consensus before coding.

I agree. Which is why I'm surprised about an "experiment" being undertaken by a handful of large SP holders that affects the whole userbase without discussion, let alone consensus.

How is it that the actions of a tiny minority are okay when it comes to upvoting and a few people effectively controlling nearly all the influence and allocating nearly all of the reward pool, but not when it comes to downvoting? If you look at the overall net effect on the system, the effect of countervoting is far, far less than the effect of outright whale voting given the n^2 rule (i.e. 0^2 is still 0 but 1 trillion squared is one septillion, a number so large that I would guess many people are not familiar with that word).

Yes the current system has flaws, but spam downvoting is not the answer.

The "net overall effect on the system" as you see it is neglecting the most important part, the human element. This is far more nuanced than math.

The 'human element' is being stirred up by people such as yourself who probably know better and actually understand how the system works, but enjoy fanning the flames to manipulate those who may not. It is sad to see. Perhaps try working to reduce the misunderstandings rather than encourage them?

The fact that you genuinely think 'dissenting voices' on this are "stirring" and "manipulating", just shows you're out of touch with how regular people think.

A process of "counter-voting" to address defects in the system, begs the obvious question... why not seek consensus on how best to fix voting imbalances and fix it?

Why make a platform that is already confusing for the average user, even more convoluted?

A process of "counter-voting" to address defects in the system, begs the obvious question... why not seek consensus on how best to fix voting imbalances and fix it?

As I stated earlier, it is not entirely clear that it can be fixed that way (not saying it can't, just not necessarily). Some problems require a degree of subjectivity; that's why we have voting at all, otherwise we could "fix the code" and make it reward the right content all by itself. It may just be that we need a two-way voting structure. Most equilibrium systems in nature (or created by humans) achieve stability through a balance of opposing forces. It is very difficult to make a system of upvotes without downvotes stable, perhaps impossible (see post by @bitcoindoom).

This does not mean that improvements in algorithms and code can't be made, and I support that, but they aren't necessarily the same issue nor solutions to the same problems.

@smooth

You may be right - fixing the problem may not be code and may require subjectivity. That can be debated and discussed.

Who exercises the subjective judgement and under what mandate – that can also be discussed. It may be there are more elegant solutions than what is being done at the moment. E.g. Moderators voted on by the SP holders (a la Witnesses), and the method may not be a flag but a way of reducing the voting weight of specific votes. I don’t know. However having that conversation and getting people to debate a proposal, tends to lead to more buy-in, in my opinion. At the very least people understand what is happening, when and why.

Anyway, 'the experiment' is here, you're well within your right to conduct it, I'm well within my right to disagree, we’ll see how it plays out. Hopefully I'm proved wrong and the platform benefits from it.

Oh I like spam upvoting. I hate spam downvoting.

It is possible to see the negative connotations of both and seek some kind of consensus on how best to address it. However I can see that point is lost.

I guess we'll see how the experiment plays out.. cross our fingers and hope for the best.

Thanks for doing this!

Interesting idea. Maybe it could be combined with my idea. 'Grey out' the SP/MV above a certain level and compensate in others ways?? https://steemit.com/steemit/@mindhunter/a-simple-question-on-whale-flagging-anti-abuse

'a proposal to limit the amount of MV that can be used for voting' That is completely vulnerable to stake being split up into multiple accounts and voted together. There are some things, that being one of them, which can't practically be enforced in code and effective stakeholder consensus is what really matters. That is equally achieved by the exisitng voting mechansm (or to say it another way, what is happening now is exacty enforced by the current code). What could change in code is adjustment of the voting rules so that it doesn't support whale monopolizing influence far beyond their share of stake, and I do support that.

I agree with most of what you've written here.

Pretty much any 'solution' can be gamed. I explain the rationale behind the voting power limit in my last post, however ultimately I'm not wedded to any single idea. Because as you say

effective stakeholder consensus is what really matters.

Which is why,

I'm surprised about an "experiment" being undertaken by a handful of large SP holders that affects the whole userbase without discussion, let alone consensus.

What is being missed by some, is that everyone has a vote here. Some whales are choosing not to vote and others, including myself, are choosing to vote in disagreement on allocating rewards on the basis of those who are voting. The 'handful of large SP holders' are part of the user base and we are voting our stake as we choose. In fact we are doing so in a manner that minimizes the in-built power imbalance. There is no getting away from that handful of large SP holders having a huge influence, the only question is the nature of that influence, which is slightly different now than it was a few days ago, and indeed has far less effect on what you call the 'whole userbase' now than then. What we are doing, really, is nothing more than getting out of the way.

You are absolutely entitled to upvote and downvote as you see fit.

Just like I'm entitled to agree or disagree with how votes are being exercised and voice my opinions.

I agree there are is an obvious "in-built power imbalance." However my point is simple. If the intention is to 'get large SP holders out-the-way', let's get some kind of consensus on how best to achieve this, if this is what the majority of stakeholders want.

whale-like groups

Hang on! You are treating a bunch of dolphins who are srteaming votes as whales?

Is this really an attack on services like Streemian and autosteem?

Lots of dolphins use those services because they haven't got time to vote manually but want to use their curating power to help new authors. I don't understand why you are against that!

Good point.
We need real readers.
Most bots are voting without reading the content, but predicting others' voting.

They were predicting whale votes. It will be interesting to see how successful they will be without the predictable whale vote to front run. I think we can't stop experiment until the robots have been reprogrammed. But robots won't work well with a bottoms up curation method.

Bots can predict "the mass" as well.

now when there are clear rules whos a whale and whos not and good % to counter them, this is a lot (not a bit!) better @abit

Defining terms... Great Idea. Now look up the word 'experiment'. Whilst you're at it look up the words 'hypothesis', 'methodology', 'control group' and 'scientific method'.

Conclusion
Power tends to corrupt. Absolute power corrupts absolutely.

"Whales and whale-like groups have (unfair) advantages on the influence on posts."

Kinda like what you're doing huh...

The system can scale as new content creators join, and older, more seasoned creators continue to gain SP and vest. Let the new ones work their way to that position. No need to penalize good content to bring about good content and growth. That will kill content creation. Is the experiment finished now?

How is people with more vests having more power "unfair?"

I thought this was the entire point of a stake-weighted platform. You earn/buy more stake, you get more influence.

Is it imbalanced right now? Sure. Is the n^2 algorithm skewing rewards more than they otherwise would be? Of course. Is the solution to turn Commie, complain about unfairness, and effectively and arbitrarily punish certain content creators and investors? No.

The solution is to fix the damn code.

We all know this and pretty much everyone agreed that the n^2 algorithm needs a change. Let's start there and see what happens. That is how you address issues with a blockchain, so that everyone is actually playing by the same rules and the results of "experimentation" are actually measurable/legitimate. All of this arbitrary/subjective "testing" is a farce.

Yes, fix it. Just do it. Before it's done, what can we do?
The experiment is playing by the rule as well.

Couldn't have said it better. Thank you.

Thanks for the explanations! I´ve been a fan of this experiment right from the start and I am still. Here is a suggestion (an idea) regarding the power and influence of whales:

With greater power comes greater responsibility.

Exactly. That´s why I believe that whales should have influence on the evaluation of content as well.

What if we didn´t give them voting power, but let them mark content they consider worth voting for - and pay them a special type of curation rewards back according on how many users followed their recommendation? Means: the more popular a whale, the more followers, the more curation rewards for the whale.
Nice side effect: if a whale needed to become popular to improve his own curation rewards, he surely wouldn´t ever abuse on his power (and flag without any need, for instance).

Then the rewards pool would be distributed by the dolphins and minnows (more responsibility for everyone), and at the same time powering up would be rewarded as well (investment would stay attractive).

I am not a developer and don´t know if such an idea would make any sense from a technical point of view or if it could even be implemented.

However, I believe that whales need to have a certain influence on the evaluation of content. Evaluation of content doesn´t necessarily mean distribution of rewards. A voice can be worth something in many ways.

I like experiments too!

Your suggestion reminds me somewhat of what @complexring was doing. I guess it was a kind of experiment by him. He would resteem a post that he planned on voting for in a few hours. He evaluated the post as up vote worthy but provided that information for others to act on, before he acted. I personally made some nice curation rewards from this. I didn't automatically up vote everything he resteemed but it got my attention because I felt he was a good evaluator and I did find most of what he resteemed to be of value. The effect of his experiment was to get more steem in the hands of those with less SP. I would like to see one of the top 10 whales try this experiment for a while and see what happens, while we're experimenting! :-)

Of course if someone did this, someone else would make a bot to up vote everything that person resteemed. But then the whale could make a bot to implement this at random; maybe for random time periods and not all the time. Then the bot maker would improve his bot to detect when this is active and when it is not. Then someone will also make a post that the implementation is currently on, and everyone else will know to look at the resteems. This becomes a form of variable reinforcement schedule (https://en.wikipedia.org/wiki/Reinforcement) and everyone becomes addicted to steemit, checking it often to see when to check the resteems, and forgets about their facebook addiction and moves a billion people from facebook to steemit. Maybe! :-)

EDIT: Here is the post where @complexring talks about his reblogging / resteeming experiment:
https://steemit.com/witness-category/@complexring/witness-update-how-i-plan-on-using-the-re-blogging-feature-and-other-miscellaneous-items

Thanks for the up vote of this comment, complexring! :-)

Thank you for the adding, that's really interesting!! I hadn't seen this post yet.
I guess whatever features developers may plan to implement, they have to be kept simple, easy to understand and to handle. We always have to bear in mind that the great majority of users won't have much time neither will to study an algorithm.
Instead of resteeming the content (since resteeming is also a feature used by minnows), whales may select their favorite pieces by giving them "points", like we use when evaluating services and products. A highly recommended article may reach a maximum of 5 points (stars) which could be indicated close to the payout and vote counter. That would simply indicate high quality to minnows and dolphins. The decision to upvote would be still on them.
Furthermore, besides a trending page we would have a recommendation page (or however we may call it), where whale picks would be listed.
Like that, rewards distribution and curation would be perfectly separated and both - the most powerful and those who want to become powerful - would have a relevant influence on the overall performance and distribution of wealth.

Thanks for your thoughts in reply!

I get your point on keeping things simple, but for me, I love the technical and complex aspect of steem and steemit! Would it be so bad if steemit was a place with a high concentration of techy folks who also enjoy up seeing and up voting non-tech content? Do the non-tech people really need to understand how things work here, to post and enjoy recognition and feedback? How many of us have a clue as to how our computer of phone works? :-) Yet we use them all the time.

Also, regarding simplicity, some people would like a simple reward system - you get a lot of votes, you get a lot of reward. Personally, I like the randomness of the rewards! I love poker too, and a great hand does not always get a great reward. The variable nature of it can be addicting, see any casino for example. Yet, there IS a formula! So us techy people can and have figured it out to some degree. We like solving puzzles like this and find steemit to be intellectually challenging just from the perspective of figuring out how it works.

On YouTube, they initially had a 5 star rating system. Most people gave one or 5 stars, so they went to thumbs up and down. I think if a whale, or anyone, resteems a post here, it is the ultimate way to select it as a favorite. If they further make it known that they will be up voting it in the near future, then they are in affect giving an easy opportunity to get a curation reward which redistributes the SP in a way that gives everyone a chance to haave more power here in the future.

It says "beta" under "steemit" in the upper lefthand corner of the page, so let the experiments and the games continue!

I like this idea. Would think about it more and see what I can come up with. Not anything technical too though. Lol.

Looking forward! ;)

How to define "more followers"? A whale can create thousands of accounts to follow himself.

True! Then what about defining the quality of a whale's vote recommendation like that:
Let's assume that you - being a whale - recommend to vote on a specific article. Then thousands of minnows and Dolphins plus your own bots "follow" this recommendation and upvote the mentioned article you pointed your finger at. Now your payback / reward (for having recommended the article) would be defined according to the articles result. To avoid that your own bots are taken in consideration, your reward could be given according to the article's final position on the trending page. The "trendiness" is not depending on the no. of votes (including your bots), but on the rewards given on the content itself and on comments, for instance. That last point is an indicator of the quality of social interaction and which cannot be manipulated by (your) bots.
Does that make sense?

I didn't see the difference.. A whale can still upvote and follow with sock puppets. So the upvote will be confirmed, so the reward will be high, so the post will be trending.

You are the one who started with this idea. Why do you pass the ball back to me? :) Be creative. The experiment has shown that there are ways to distribute power more equally (= more fairly), thus it´s worth to keep thinking in the same direction.
Flags are surely not the solution, they were an effective short term instrument to wake us up. But I am sure you agree on the fact that they have caused a lot of trouble, too.
Try to bundle your smart forces in a healthier way and this is going to succeed.
Peace!

So that makes me borderline...?

and it means you should invest more - your marginal investment now worth 2 times more than we minnows. :)

@abit okay thank you for clarifying this
I'm guessing we got flagged cause there are those accounts who don't know they are already whales or maybe they do but are on auto votes mode which is sweet too I think - that's loyal support !

Whales and whale-like groups are those who have more than 800M VESTS in total, or around 384K SP in total.

so this is your purpose - to grow but how will this lead to growth?

As a member, you want we have more users so each SP your own will worth more in the future, or you earn more SP today but each SP will worth less in the future?

this part just killed my dream of being a whale okay so that would never happen unless a few whales keep powering down and give up whaleness ...

The number of whales is limited. With this limit, the system won't scale, we will be unable to grow into a bigger user base.

I don't have to imagine this is somehow happening now isn't it ? Till you did this experiment. Somehow I felt like a whale - I still only have cents for effects but hey better cents that no effect at all. and yes .. I get how you whales feel now with all that voting power in your hands - it ain't easy casting a vote specially that my whaleness feeling has an expiry date and VI %age limit. Anyway, am loving the whaleness feeling but I also noticed that I still don't have much effect so does that mean that - the worth of my SP is already worth less?

Image someday we have one million new posts a day, 10 whales with 40 votes each can only vote for 400 posts in total, and probably many votes will go to same posts, what will the other 999,600 authors think?

may I ask what we are expecting as a result of this experiment? Higher worth of SP in the future? no one invests in something that devaluates

When STEEM's price is rising, our SP will worth more.

One expectation of the experiment: fairer rules.

@abit so I take it this would pump the price of Steem up?
The more we should write eh?
Okay, thank you for explaining that.

Is the experiment now over?

Coin Marketplace

STEEM 0.36
TRX 0.12
JST 0.040
BTC 70839.13
ETH 3563.41
USDT 1.00
SBD 4.77