Assume a can opener -- A critique of the snowflake plan
A physicist, a chemist and an economist find themselves stranded on a desert island with nothing to eat. One day, they find a cache of canned goods washed ashore on the beach. However, they have no way to open the cans.
After pondering the problem for a while, the chemist says "I think i have an idea. It will take us a few days, but if we take some of the seawater, and concentrate it, we can eventually get a solution corrosive enough to eat through the top of the can."
He's barely finished before the physicist interrupts him "Wait! I've got an even better idea. It shouldn't take us more than a day. We collect some sand from the beach, then we build a fire and melt the sand into glass to make a lens. We use the lens to focus the rays of the sun until they melt through the top of the can"
But the economist interrupts him "wait! wait! I've got an even better idea. I know how we can open the cans right now... OK... First, assume a can opener"
I've linked snowflake and timcliffs posts about this at the bottom of this one.
To offer a quick summary, snowflake has an idea to change the structure of steemit voting by
- Eliminating curation rewards entirely
- Adopting a two tier voting system where users with steem power above a certain threshold only downvote, and smaller SP users upvote and downvote.
On its face, this plan seems like a good idea to give regular users more influence on the steem platform. However, there are multiple (probably) insurmountable problems with the plan. I am posting this not just to blindly criticize, but in hopes that snow or tim will come up with a viable way to address some of these problems.
Problem #1 -- the benefits of Snowflake's plan depend upon unproven assumptions about the reasons for current voting patterns
In many cases, these assumptions are not in line with what is actually observable on steemit everyday. In one of the posts @smooth pointed out that the majority of whale upvotes do get some sort of curation reward. And he may be right. But the current curation reward system severely penalizes bandwagoning, and voters, whale and otherwise, bandwagon nonetheless. The current curation reward system severely penalizes whales for voting for the same authors all the time (because of frontrunning). But they do so nonetheless.
The current structure of curation rewards might encourage whales to vote as often as possible while maintaining their full voting power in order to maximize rewards (and many of them don't even do that in spite of the fact that not doing so is tantamount to throwing away free money). However, aside from that, the current problems with curation that snowflake aims to fix are problems in spite of the structure of the curation rewards system, not because of it.
Its easy to scapegoat curation rewards and assume that whales and others who are voting in undesirable ways are doing so because of them. But there is a paucity of compelling evidence that this is the case. If it were, we would expect the same desire for increased curation rewards to prevent whales from doing things like bandwagoning and voting for the same people frequently.
Removing curation rewards will will almost certainly have the impact of making it less profitable for people like @biophil and @ats-david to do what good they are able in finding and upvoting original content with their trail (both have said that they do it for profit). That is to say, there are certainly people who are voting in ways that many would consider desirable for whom curation rewards are the primary incentive.
In all, starting from a null hypothesis, we have significant evidence that many continue to vote in ways considered undesirable (like via bots) despite a penalty in curation rewards. We also have evidence that users who vote in a postive way (ie attempt to find new, undervalued content) are motivated to do so, at least in part, by curation rewards.
Problem #2 -- Snowflake's plan depends on assumptions about whale curation behavior that history has already proven to be false.
One can assume that "good whales" will accept the moderator position, and "bad whales" will split their account into many self-upvoting sibyls. Because we will have removed a great deal of the upvoting power from the steemit ecosystem, these bad whales will have a level of influence in assigning rewards far beyond even the disproportionate level of influence whales enjoy now. In snowflakes mind, this is where the good whales who are moderators step in. When they see upvoted content that doesn't meet some standard of quality, and they will "moderate" it and reduce rewards. Per snowflake's reply in his post:
Moderators would downvote content that's overpaid. It doesn't really matter if it was upvoted by many sockpuppets, if the content is shitty payout will go down. We would have a system were post would go through a lot more scrutiny, which will increase the quality of content selection.
However, for this system to work, the moderator whales in question would have to be willing to spend a considerable amount of time and effort to subject all posts and comments on steemit to this level of scrutiny. And they would have to be willing to do so for free. There is no incentive in snowflakes plan that they do not have under the current system, and therefore no reason to believe that these moderators would be any more inclined to spend time finding and downvoting bad content than they are currently. If they were willing to take the time and effort to provide the level of scrutiny necessary to ensure that "if the content is shitty payout will go down" the problem that @snowflake is trying to fix wouldn't be a problem.
After all, the moderators that snowflake is depending on to provide this level of scrutiny are already bandwagon upvoting low quality, cookie-cutter posts (with little or no financial incentive to do so) by SG curators in exchange for having to scrutinize steemit content less. That is to say, they are willing to pay money (in the form of votes) to provide less scrutiny. Why would one expect them to be willing to provide more scrutiny for free.
You might argue that, under the current system, these whales just want to save their voting power for curation-reward-eligible upvotes. However, a look at voting history will show that that is not the case. Dan didn't dust off his big account @dan until a few weeks ago. @ned has been somewhat more active in recent months, but i remember when i wrote my post about the 5 votes a day change he was casting something way less than the 40 votes a day limit for maximum voting power. The same is true for the val accounts , jamesc, and many others, all the way down the line. RIght now, given the choice between letting their voting power go to waste, and scrutinizing posts to DV bad, overpaid content, most of these prospective moderators choose the former far more often than the latter.
Nearly all of the biggest whales have unused voting power with which they could downvote bad content at no cost to themselves, if they wanted to do so. They just don't want to do so (or they don't feel much content is bad enough to warrant a dv). There is nothing in this plan that would make them want to do so any more than they do currently. They get the same amount of inflation money if they do nothing as if they downvote a bunch of bad posts every day. So why deal with the hassle?
"because they have a large investment in steem, and responsible curation will improve the price of steem" has already proven to be insufficient. Maybe its social loafing, maybe its just that they feel their time on steemit is better spent in other pursuits, but there is no reason to believe that the same long term concerns that fail to motivate them now will somehow become more compelling under snowflakes plan.
problem #3: the level of inflation necessary to adequately incentivize being a moderator vs a group of self-upvoting sibyls would be too high to be sustainable
Right now, there are a bunch of bots with a lot of steem power. Will they really go extinct if there are no curation rewards, or will they just turn to self upvoting (again with a disproportionate level of influence)? Are the moderators, with no compelling financial reason to do so, really going to be willing to play John Henry to the bots' steem powered hammer (<-- thats clever right).
Of course, the bots could choose to be moderators and get free inflation money. But how much inflation money would you have to give them to make that a +ev decision over self upvoting (or sockpuppet upvoting) even at a very modest level of success? Especially in a spot where we just had to reorganize the universe to cut inflation. You would have some extra free money to play with as, you could convert the money currently going to curation rewards for moderator SP incentives, but thats pretty trivial (spitballing but something like 50K steem a week) Thats 50K non compounded you have on top of current SP incentives. Just taking into account the top 44SP users(those with with 1GV or more and not including the steemit account, though im not really sure how you can exclude it) thats 50k SP devided up porportionally for 150GV... so just over 2.6 kSp per week on a 10GV account. SO as an example, dan would be making around 2.6KSP/week on an approximately 5 million SP balance.. around .05 percent weekly or around 2.6% per annum. thats not terrible for doing nothing, but he could still do better powering down and holding SBD. ANd out of the 280 full power equivalent votes he could cast every week as a group of sybils at that power, he would only need one or two full power ones to sneak through the moderators and be successful (or even just to sell a few for a percentage of the proceeds).
And even that underestimates the advantages offered by being a group of self-upvoting sybils. Because his vote would actually be worth way more as a bunch of sybils, because there would be fewer votes being cast (because some whales wouldnt be voting to take the incentives)... also, ive probably slightly overestimated his returns on balance, since theres a non trivial amount of vests that would still be above the threshold but that aren't above 1GV.
problem #4: the SF plan would give many moderators unchecked power when downvoting at 100%
Reducing the amount of SP with the power to upvote would give whales who split their account for the purposes of self-upvoting far more influence than they currently have (as outlined in problem 2).
However, there is another side effect to making most SP upvote ineligible. The mirror image of this problem is that it would be exceptionally difficult, perhaps even impossible, to counter a whale downvote. Even if a modestly sized whale like @enki downvoted someone, it would take a non-trivial amount of consensus among the regular voters just to take the post out of the negative rshare range. I'm not even certain it would be mathematically possible to counter a downvote by someone like @dan or @ned or @blocktrades even with 100% consensus of all upvote-eligible accounts. Currently the check on irresponsible whale downvoting is that other whales will (at least sometimes) respond to it by upvoting the material that was unfairly (in their perception) downvoted. The SF plan would remove that check which, afaik, is the only one there is.
In all, the snowflake plan seems, to me, to assume a lot of can openers that only exist in an ideal world.
https://steemit.com/steem/@timcliff/whales-can-the-community-buy-out-a-portion-of-your-influence
https://steemit.com/steem/@timcliff/snowflake-s-really-interesting-proposal-turn-whales-into-moderators-and-give-dolphins-all-the-voting-power-tldr-version
https://steemit.com/steem/@snowflake/guardian-of-the-steem-universe-a-different-perspective-on-the-role-of-whales-within-steem-ecosystem-part-2
https://steemit.com/steem/@snowflake/guardian-of-the-steem-universe-a-different-take-on-the-role-of-whales-within-steem-ecosystem
Yes, this is what I've been trying to point out to snowflake and timcliff. The proposal is based on unproven assumptions about curating habits. They're looking at the effects of a tiny user base and massively disproportionate distribution and extrapolating that curation rewards are harming voting and engagement. It's a non sequitur.
I think your critique is quite accurate. "Assume a can opener" is almost spot-on.
Its kind of something ive noticed about improvement proposals for voting/curation. Its pretty easy to come up with a well-intentioned plan, make a few perfect world, can opener type assumptions, and come to the conclusion that it will fix everything.
Because any system for voting or whatever that happens in a perfect world will end up being way better then the one we have here in reality, because there are so many things that, at the moment, are non-optimal.
Exactly. I don't know how many times I have to say this, but i'll give it another shot.
The biggest problem we have right now is a lack of active users. Nothing is going to appear to work right with such a tiny user base and a huge disparity in stake.
Now, the one thing that has not really been explained in the proposal in question is why any particular user would buy into a system so that they can read and vote on content? Is the content on Steemit going to be exclusive for registered members and that much more superior to most other blogging sites? The proposal not only assumes that voting habits are/will be different from what they are, but it also assumes that people will spend their money to do things that they can do for free right now - and not even earn any ROI.
It all looks like magical thinking to me.
Because they would get more influence, they will be able to give more $ to their friend's post if they do.
Of course voting habits will be different, why do you think nobody is voting for comments ? Why do you think people will still cast all their vote 30 min after a post was published? Why do you think people would still subscribe to bot if curation rewards are removed? It's clear that voting behavior will change a lot.
They won't really spend their money as in " not free". The steem power that they bought can be sold back for fiat and will likely be worth more than what they initially paid for ( so their you have your ROI) . This is the dynamic that is needed to create steady demand
So in this perfect world where we're implementing your system, steem price is constantly going up. And the fact that its constantly going up is what causes the demand (not the other way around)
While in the real world, we're flirting with ATLs on what amounts to a 7 month slide... you see how these assumptions are more than a little biased? ANy system is going to sound like it works better than what we have if you take it as a given that the price of steem will constantly go up.
You are misunderstanding, the price will also go down and some users might lose some influence, but it won't matter because they bought it already and that increased the price. Some new users will want more influence which will increase the price again, then the price might go down,etc..but the idea is that more people will buy steem power to get influence than people will sell the steem power that they already bought.
Steem price will have to increase to handle new users, so yeah that should definetely be the goal. I often hear people say that an increase in number of users will lead to an increase in price, this is completely false in the current form, users have no incentives to buy steem power right now and curation rewards is only an incentives for a minority of users, most people who uses steemit have no idea about them.
Who cares about giving more money to their friends' posts if there's nothing in it for them? Do you go out and buy a bunch of stuff for your friends so that they can have more things for themselves? Who actually does this?
Umm...they don't. Have you been watching vote times?
They wouldn't...because they simply wouldn't bother curating at all.
Yes, it will change because there won't be any users.
This is an assumption not rooted in any real-world observation. How can your investment appreciate when nobody will be using the platform because there's no incentive except for bloggers? Where will the investment come from for bloggers to earn? Who will buy into the platform just to vote for other people to earn money? And who will buy any meaningful amount in order to make such blogging worth the time or at least more than any other site that pays for content?
I will say this again - the market for bloggers is vastly smaller than the market that consumes blogging content. The proposal is based on bad assumptions about behavior and incentives. The problems we see on Steemit are not a result of the existence of curation rewards.
this is a great point. Ive never come up with a precise number, but i suspect more than half of all curation rewards are forfeited in the reverse auction.
This is one of the things like bandwagoning and list voting. If what SF thought was casuing current behavior was really causing it, most people wouldn't vote for most things before the 30 minute mark.
Part of this is a game theory decision.... sometimes voting early and forfeiting some percent of curation rewards is a better outocme than voting later (and therefore with more people in front of you). But often, its just a decision made without any rational basis whatsoever.
That's not a good comparison. Like I said in my post steem power is similar to upgrades and power ups in games. People buy it because they like the game and want to gain influence/power in it.
Are you trying argue that bots will still vote at 30 min if curation rewards are removed?
The main reason user retention rate is bad is because people don't have fun on this site. What do you propose to make it fun?
You are not seeing the bigger picture. Steemit is just one site among many other sites that will use the steem blockchain. Many of these other sites will have very different models. Some site will be similar to facebook, where people will just document their lives to friends and family, these people should be able to send a few cents to each other. Do you think these people will want to go through the posts of people they don't know to curate content? The curation reward model is a flawed one for mainstream adoption and isn't going to create significant demand.
The only system that will attract a large audience is one where people buy steem power in order to have more influence in the system.
Curation rewards encourages bot voting and discourage people from commenting/participating in the platform because no one votes for comments.
I don't know exactly what problems you are refering to, but curation rewards are indeed not the main issue, the biggest problem is that 99.8% don't have fun because they have no influence on the platform.
"Who cares about giving more money to their friends' posts if there's nothing in it for them? Do you go out and buy a bunch of stuff for your friends so that they can have more things for themselves? Who actually does this?"
I sent a complete stranger, granted we were part of the same facebook group but I was never "friendly" or a "friend" to the person $1000 cash, through the mail (fuck legality), for nothing in return to do as he wished, simply because I knew that the person he was infatuated with/in love with was traveling across the country to meet with him (not a cupid move or a cupid fb group). People do all kinds of things when it comes with sharing and giving and there is no "real world" who does this in that sense.
Foundation. You have a theory about what causes this voting behavior that you consider undesirable. But you haven't proven it. Just saying "of course im right. Why do you think im wrong? its clear that im right" Isn't proof. its entrenchment.
So lets say you want to build up your power. You decide to buy some steem. In order for you to buy that steem, someone has to sell it to you.
That is your transaction. 1 buyer. 1 seller. Not more buyers than sellers. The same number of buyers and sellers.
Less sellers than buyers = price going up
The number of buyers and sellers is never the same, that's what creates the market price !
According to your theory the price would be flat.
No, i completely understand what youre saying is going to happen (Even the impossible part about there being more buyers than sellers). In your perfect world, the price will fluctuate, but generally trend upwards. In our real world, the price fluctuates, but generally trends downwards.
The problem is, you have no rational foundation for your claim that these changes you want to make will actually change create a higher demand for steem. There were people saying in july and august, about the current rewards system, the exact same thing you are saying now.
Your logic is basically circular. Getting rid of cuation rewards is a good idea because it will increase the price of steem. How do you know it will increase the price of steem? Because its a good idea. How do you know its a good idea. Duh, because it will increase the price of steem.
How is that impossible? Most users will want to build up their power especially if they see significant gain in their influence.
The price is falling right now because there are a bunch of whales that just want to get out. If you look at active users on this forum most are not powering down.
I suggest you look at the cryptocurrency chart since 2012
This is ridiculous, the july price spike was a bubble.
So people buying steem power to gain more influence is irational to you?
What value do you find in a system run by bots ?
Why would giving influence to 99.8% of steemians not be a good idea?
1- I don't think I ever mentionned whales voting for same authors or bandwagoning being a problem. Essentially what I said is that curation rewards encourages bot voting.
And I agree with you that curation rewards are not inherently bad, as long as they encourage natural voting these rewards poses no problem at all. I've been thinking about a system for curation rewards that would be different but it's not easy, I thought about using an average of all post instead of rewards for each post and also have some sort of limitation on rewards but even this has a lot of abuse issue.
2-
A lot of the moderation would be done by bots, moderators could set up all kind of differents filters to spot sybil behavior. Like I mentionned in my post there would also be a report button.
Another thing is that a downvote is basically an upvote to the system as a whole instead of being a targeted upvote. So if moderators sees something that deserve more reward they could simply downvote something that they think do not deserve it, that will effectively increase reward on that post.
3- I disagree. If all rewards are completely removed from users, moderators would have very little incentives to split their account. Also the more moderators do actually split the more inflation will be given to the ones who don't , so essentially moderators who don't split will build up more and more power to counter a sybil attempt.
4- Is a valid concern, @timcliff suggested that moderators would be able to upvote to counter abusive downvoting from other moderators, they would be able to upvote to bring the post back to where it was before the downvote but not beyond.
I realize this. My point was that these voting patterns exist in spite of a significant curation reward penalty. That is to say, not every voting pattern can be attributed to a desire for curation rewards. You assume people bot vote because of curation rewards. And you might even be right. But you have not proven it.
SO these bots exist right now, or theyre just electric can openers? Cheetah is moderately good at detecting the most obvious sort of abuse, but for the most part right now, we find abuse with human eyeballs.
Its not a question of power (they don't have to build it up they already have more than sufficient power in the system you describe), its a question of the will to use that power. And the acuity of discernment to be able to see most of the places where it needs to be used.
fair enough that seems like an OK way to address that.
Here is an update:
https://steemit.com/steem/@timcliff/update-on-the-whale-buyout-proposal
If the curation rewards were removed, how does someone who does not produce content climb the ladder to become a moderator, without spending money? Might as well just start charging thousands of dollars to purchase an account.
So I bought my account today, I'm a moderator. Now to spend my day looking for things I don't like.
Who shops for the things they won't like?
The majority of members now, including new members today, tomorrow and next week will not be able to earn anything unless they produce content...while those with enough SP sit back and earn without contributing. The rich get richer.
How are these new problems supposed to attract new users?
Haven't you heard? Everyone on the internet is a blogger. And those who are bloggers that don't actually blog will simply spend a bunch of money on an account so that they can give other people rewards.
Duh. I thought everyone knew that.
Oh, this was a plan to attract new users? I thought it was a plan to get rid of the ones who decided to stick around. I guess I need to read it again.
My sarcasm detector almost shattered.
So I was on Youtube today enjoying the work of a particular vlogger I enjoy. He rides around on a dirtbike while saying words I find entertaining and/or interesting. One might say to themselves, "How does that bring value to the platform?"
This guy is almost at the 200000 subscriber mark. The video I watched had 69389 views. Yes, I realize youtube has millions upon millions of users who frequent the site. I noticed something peculiar though. Out of all those views and subscribers, the thumbs up button was hit 2822 times. I'm sure far more than 2822 people liked that video. Only 66 pressed downvote, and those were probably jealous trolls.
What could possibly be the reason why the upvote button on Youtube is neglected? No incentive, perhaps?
At the start of the video, the video blogger starts talking about how much he loves his supporters and begins thanking them. He said, "If only there was something I could do for you guys."
If he was on Steemit, he could have said, "Don't forget to upvote! Enjoy your piece of the pie! Thanks for the support, as per usual!"
I'm certain there's a lesson to be learned somewhere in what I just said.
Yeah...I'd say that there's an extremely valuable lesson. That is one of the brilliant ways that this platform can be marketed.
Excellent observation!
[EDIT] - I upvoted you with my trail. I hope the Steem Guild staff takes notice of it and whines about how abusive it is. I would upvote you with just my account, but that's not how it works and there's nothing I can do to change that. I do hope that you understand and that you'll forgive my transgression.
Those content consumers spent 14 minutes on the video. Many stuck around to fill the comment section, so it's very hard to say people don't read these days. A lot of upvotes seem to occur in those comment sections. In this case, a quick tally tells me more people upvoted in the comments than they did for the video. There's another lesson.
This is a blogging platform, and so is Youtube. We should be comparing this model to their model. Marketing Steemit like it's supposed to be some sort Facebook replacement is pointless. We need bloggers to produce and many consumers to eat it up, plus a bonus incentive to hit that like button.
[EDIT]How dare you do what they do within a comment section of every successful blogging platform!
Im pretty sure most of them keep me on ignore
Fuggin' brilliant.
I like that a lot.
This is well put
Then there are those who upvote something simply because they think it's good without thought to whether or not it will benefit them. And comment to let the author know they think it's good, also without expectation of reward. Is that really as rare as this post and its subsequent conversations make it sound?
I haven't time to read your whole post right now, but here's a cute little idea that popped into my head as I was reading timcliff's post: https://steemit.com/curation/@biophil/a-cute-idea-for-curation-rewards
its interesting. If im understand right, you want to people to either be eligible for SP incentives, or curation rewards.
Pre HF-16 i think this actually might have worked pretty well. AORN, however, im not sure that there is actually enough SP incentive money being produced to make it a real alternative.
Yes, that's the basic notion.
And I think I agree with your second paragraph; but it would be worth looking into how big typical whale curation earnings are. In my days as top curator, I was earning around 50% APR effectively. My clients were getting considerably less; say 20%. They were all top-25 best curators. So I need to look into it closer, but I wouldn't be surprised if all but the top 100 curators are earning less than 10% on their curation.
Interesting read.
From a broader perspective, what @snowflake wants to tackle is not undesirable voting pattern itself. As he mentioned, the problem of curation reward is it "has turned the platform into a completely fake system where everything is run by bots and driven by money". In other words, the motivation is an issue rather than the results (voting patterns).
However, I agree that removing curation reward can harm human curators with the current profit mechanism. We maybe able to solve it with taking alternative ways. For instance, submitting articles to curators with reward sharing (Github issue 773), posting editor's pick, etc.
I question the causality. I do agree that the system tends to be inauthentic, and i do agree that bot voting is part of the reason for that inauthenticity. Just showing that two things co-exist (the current curation rewards system and inauthenticity) doesn't show that one thing caused the other. The two things might have a common cause. Or, they might be two completely independent phenomena.
IMO, the current curation system is rigged toward whales and bots, which let many users disappointed and consequently make the system inauthentic. First, for free users, who are potential investors, with around 30 STEEM, they don't earn meaningful amount of curation rewards. And many of them only vote several times per day while bots do tens or hundreds times, which decrease human users curation rewards. Unless you are a whale or a bot, curation reward may mean nothing but attracting bots.
Curation rewards are only a part (and a small part at that) of the curation system. The larger part is author rewardshhich is which are way more rigged, and bigger than curation rewards by a factor of probably close to 7:1.
The guy with 30 steem power, unless hes incredibly lucky, earns no curation rewards at all. Remember if its under 2 cents they round it down to zero. Id be curious to know what the least amount of SP anyone has ever earned a >2c curation reward with was (in current times since steem has been valued at 10-20 cents). If i had to take a stab, id say in the vicinity of like 300 steem.
Then again, its debatable whether brand new users with 30 free steem power should be earning any curation rewards at all. I don't think curation rewards (as they exist now) are a viable steem entry point -- im just not sure they were ever intended to be. I think new users should have opportunity here, but every single feature or reward does not need to be geared to convert new users to investors.
One version of the proposal (maybe only discussed in chat) would allow whale upvoting to the extent of previous whale downvoting on the same content. So whale moderators (upvoting) could still cancel out other whale moderators (downvoting)
@smooth how about if the whales also just regulate their votes so no other whale will have to down vote the post that looks like overly rewarded - it's a waste of voting power doing that or I could be wrong. Anyway, please check your chat I sent you a reminder, thank you very much!
Yeah... SF mentioned that. You could also simply alter conversion of SP to Rshares for mod downvotes to make them more overcomable. Its really the most easy to address objection. though it bears pointing out that youre taking an already relatively complex system (steemit voting), making it more complex by differentiating moderators and regular voters both of which have seperate sets of voting rights and rewards, then making it even more complex by having the mod - on - mod vote canceling privelages (or the seperate rshare conversion or however you address it)
https://steemit.com/curation/@sigmajin/an-opponent-of-the-exponent-making-the-case-for-vshare-linearity
Here are some thoughts of mine of what can happen to improve curation. No can openers that i know of, but if anyone finds any id love to know.