You are viewing a single comment's thread from:

RE: Reward Curve Deep Dive

in #steem5 years ago

First, I'm excited to see how this will go because I've wanted to see how something other than pure linear would work in real-time (wasn't around back in the early days of superlinear.)

However, regarding the curve "converging" on linear at the top, won't this just encourage people to split their stakes into whatever quantity is optimal return (before it starts to flatten) or am I conceptualizing this incorrectly?

Sort:  

There is always some loss from splitting here (since it only converges to linear but never quite gets there) but the idea is that qualitatively the switch over happens at a point that the rewards are reasonably obvious and can't be hidden in thousands or millions of micro-rewarded comments. Since they are visible, if abusive they can be downvoted (part of the EIP is also to reduce the cost of downvotes).

Therefore splitting: a) still has at least some cost, and b) can't be done to the point where the rewards are hidden in many very small payouts, or the cost is large.

Great explanation, thank you @smooth.

In reply to the OP I propose mechanisms I think preferable to manipulations of rewards curves. Given your competence, I'd appreciate you proving my proposal would not work, and is stupid. If you can't, well, you know what to do.

Thanks in advance.

thanks, I was wondering the same thing.

Actually, if one looks at the rshares generated during voting, they are very large numbers, so I suspect the convergence happens very rapidly.

I also wonder if the code will use some approximating function for such large numbers, as happens with the square-root function in curation rewards calculations.

(part of the EIP is also to reduce the cost of downvotes).

So we can have more comments that are relevant to the discussion greyed out by individuals like @ Iflagtrash? Rewarding bad behavior is not the way to go. I have seen many of @valued-customer's comments in this thread greyed out by iflagtrash, so those downvotes of his comments really add to the value of steemit and the steem blockchain? A downvote pool is a very very bad idea and will only aid the bad actors.

What is grayed out or not is a function of the UI and not the blockchain. Steemit.com chooses to gray out some content under certain conditions as is their prerogative; other UIs may do it differently or not at all.

That's a completely different matter from granting rewards, which needs a balancing mechanism against self-voting (where people can just reward themselves regardless of what anyone else thinks of it, which undermines the entire premise of the system).

If and when there are cheaper downvotes (and if there are not, social rewarding will continue to work poorly if at all), UIs might make different decisions about hiding. For example, they might require even more downvotes before hiding something, on the theory that downvotes being cheap and something getting only a small degree of downvoting means it isn't actually that bad.

Getting these systems to work well is always an exercise in balance and compromises, not absolutes. To understand how such a system operates, one must consider not only immediate effects but also resulting adjustments and counteradjustments. If this were all easy we'd have a splendidly working system by now. If it is impossible we should give up and turn off rewards (and stop paying the price of repelling investors with high inflation, resulting in a low and declining Steem price). It seems exploring the middle ground is more reasonable and closer to consensus at this point.

That's all well and good, and I have in previous comments mentioned that it is steemit, and I only use steemit. This is a post from steemit, not busy, not esteem, not partiko, but from steemit, and it is not a post about correcting a UI issue, but a steem blockchain issue.

>Hello Steemians, I’m @vandeberg, Senior Blockchain Developer at Steemit

Consensus, when it comes to anything from the company and relating to changes is 35 people being the mass majority on a poll that was only valid for a few hours before it was decided that there was overwhelming support for the previous change. There is no such thing as consensus on the steem blockchain, so I really do not understand why people pretend there is.

I know, and the vast majority of the people responding against the downvote pool all know it is a done deal so just get on with it, and let the serial flaggers have there way and have an ability to flag and still collect rewards and not only virtually rape users and rake them over the coals but also allow them to rape the reward pool they claim to be protecting.

There is consensus on the blockchain as demonstrated when a hard fork is enabled or (as has occasionally happened in the past), not enabled, the latter indicating lack of consensus.

Other than hard forks we haven't up until this point had an objective way to measure consensus, but the SPS will do so, since it will be voted in a verifiable public way by Steem stakeholders. Previous efforts at "polls" and such things have never been objective or particularly meaningful.

I don't actually know what point you are trying to make by bringing up Steemit vs busy or esteem, but there is only one Steem blockchain, and when you are looking at things like the reward pool being paid out, it is only the blockchain that matters, not the UI. When looking at things like what is hidden or grayed out, that is UI. The issues affecting one or the other are very different.

A downvote pool is not a UI issue. It is a blockchain issue, currently to downvote a post it cost the downvoter mana, it does not matter what UI they are using. If it was a UI issue then people using busy or esteem or another UI could downvote at will if their UI did not charge a mana fee so to speak for downvotes.

(part of the EIP is also to reduce the cost of downvotes).

There is two sides to the downvote issue if not more. At what point does needless downvoting stop? Enabling a cheaper downvote is enabling bad actors. By lowering the cost of downvoting that does nothing to curb downvote abuse as is currently happening. The vast majority of downvotes appear to be going out to comments not plagiarism, not excessive reward, but to normal everyday comments.

People talk about wanting to stop the self vote and self vote from alt accounts, start with the first issue, the self vote. Do away with it. Then see if people continue to vote themselves with alt accounts. People will see a trend pretty quickly, then the downvote for reward abuse will work.

By eventually allowing people to get paid or to pay themselves because of their downvote actions is only going to exacerbate the issue. Because getting a downvote curation reward is where this is all heading, that is pretty obvious with the desire to have a separated downvote pool.

For the last 22 months on steemit, every time the self vote is bought up, it has never been simply stopped, removed from the system. Either it is good for the blockchain, or it is bad for the blockchain. Before creating new problems fix the problem everyone keeps saying is the reason for the downvote pool and that is to stop the abusive self voters. Just do away with the self vote and be done with it.

Self-vote is neither good nor bad. It depends on the content and context.

For example, if I post something really value-adding and want to jump start the voting with a self-vote, there is nothing wrong with that, and indeed could be argued as a very legitimate reason to buy Steem for (non-abusive) self-promotion. (And indeed further argued that if I'm not willing to spend my valuable vote power on my own content, maybe I shouldn't even post it.)

On the other hand, if I post a bunch of empty comments or other worthless 'content' and self-vote them, then I am abusing the system

The only mechanism that allows the blockchain to distinguish between those two cases is having other people either agree with me and add their vote (or at least do nothing and let my vote stand), or downvote it and both reduce/prevent the payout and discourage my behavior since my vote has now been wasted.

Self-voting isn't the problem, but even if it were, it is only a small part of the problem. Paid votes are just as easy to abuse, already exist in very large numbers, and would not be stopped or even slowed down by stopping literal self-votes. So, no, that is not a solution.

As you correctly stated, downvotes have a high cost and as a result there are only a tiny number of downvotes that happen (and yes, as you say, many if not most of that tiny number of downvotes are from annoying jerks), and so naturally we have people abusing the system though self votes and paid votes all day long.

I disagree with your suggestion that the most effective way to address the core problem is by focusing on literal self-votes. Not only is it easy to evade, but even without any obstacle in place to evade it is already only a narrow subset of the problem, effectively a distraction.

At what point does needless downvoting stop?

Probably never. Any system which allows people to interact will always have jerks who want to annoy others. But it also should not be the case that what one or two jerks do matters much at all (or, indeed, the system is much too fragile).

So, again, like I said earlier, probably UIs should demand a lot more downvoting before paying attention to it and performing hiding, negative reputations, etc. (effectively ignoring the few jerks who can still downvote, but will not be able to meet this threshold very often if at all). And for that to happen, we need a lot more downvoting from ordinary non-jerks, where well-deserved, which in turn requires that good non-jerks not have to burn their valuable vote power in order to perform a public service.

Should the steem account be used to fight abuse?
Wouldn't this reduce the burden on the rest of us, and make possible stopping the largest abusers?

No, and if you do that you are making Steem into an illegal security, even more than it already is (essential functions and therefore investment returns depend on the efforts of a central party).

The right way to address this is to improve the underlying economics so that it works as a decentralized system, or if we can't, recognize that it doesn't work and pivot.

How does a blockchain defending itself from attacks on it's value become prosecutable by the sec?

Steem is open source, not owned by any one entity.
What does it care about the rules of violent criminal gangs that force others to pay and obey?

Steem is a meme that cannot be put back in the bottle.
It is viral technology that cannot be stopped, except by the witnesses.
Not all of which are in the united snakes.
(I hear there may be a satellite that might let us rent space on it's servers.)

How do we open source the access to the keys?
Hard code a 85% witness, and 90% community, consensus mechanism that opens the keys to the code to allow changes?

By hard coding a mv influence cap, one that slowly rises over time, it takes the choice from any central parties.
There is no central party, only code.

They can try to punish coin holders, but c'mon, that isn't working for btc.

Maybe I misunderstood but when you said 'the steem account' I assumed you meant @steem which is owned by Steemit. If the critical functions of the blockchain are depending on one company (which also happens to be the company that developed it and launched it) going on four years in then it isn't a decentralized system.

Loading...

I think probably anything is better than pure linear at this point because pure linear is proven not to work at all. But I also think that even if we make these changes, too little thought has gone into the question of "whale creation".

We shouldn't demonize whales. We should want people on the platform to want to become whales. We should think of "whale status" as the product which Steem sells to investors. And so far most of these economic changes attack the honest whales while rewarding the dishonest whales.

I declare the honest whales to be the whales who earned their status by the rules of the platform or who invested in the platform. I totally understand certain whales are grandfathered in and this has created most of the controversy but then what about the whales who earned or and invested their way to their status?

The problem? Being a whale doesn't actually improve the UX for those who are whales. As a result there isn't as much demand to want to become a whale. As a result the investment in Steem is lower. This is bad.

Yes some whales might have had too much power but as I see it if they invested into the community they purchased their power fair and square according to the original rules. If we are concerned with content discovery how does this fix the main issue most of us have which is the vote buying issue?

Yeah, I completely agree that I'd like to try ANYTHING other than pure linear. If I had to sum up what it feels like in one word it would be: STAGNANT. It doesn't feel like anything dynamic is going on. More like currency is just being mined irrespective of any "proof-of-brain" concept.

I also agree about whales. "Blame the rich" mentality is quick and easy, but rarely leads to productive discussions about optimal wealth distribution. You're spot on that there are few incentives to become a whale outside of altruism and perhaps a company combining many individuals' stakes to gain visibility on the platform.

Altruism, when it impoverishes oneself, tends to be draining and imbalanced. Of course some will always complain about "ninja-mined" stakes and who deserves what, but that is subjective and again, rarely leads to productive discussions.

I'm also very curious how these changes will affect vote-buying and bid-bots. At the very least, it should shake things up and force some adaptations which will hopefully make the platform feel less stagnant.

The thing about being rich, we want rich people to join our community to make our community much richer. I am definitely against the "blame the rich" ideology because we want the rich and the poor to both get richer. Everyone should feel like they have opportunities to get richer while having fun.

The issue I think with Steem is merely there was never enough opportunities. SMTs were promised, Oracles, communities, etc, which may have given people more ways to earn Steem and become whales (rich) in the correct way.

But to go pure linear and then also not have anything going on? To be honest playing DrugWars is more fun than posting on Steem at this time which is why I don't post as much. DrugWars might not be earning me anything at all but at least it's fun and feels like I'm earning something (gamification works!).

I'm also very curious how these changes will affect vote-buying and bid-bots. At the very least, it should shake things up and force some adaptations which will hopefully make the platform feel less stagnant.

Vote buying made the whole thing pointless. Imagine if society were set up where you essentially have to buy your job. And the jobs are offered by a computer which takes digital currency and then hands out the jobs to all the people. In this we'd have the worst of communism and crony capitalism combined. In my opinion that is what we currently have on Steem.

The votes are essentially rigged by bots which sell rewards. How can it possibly get more broken than that?