More n^2 #47
If you follow my blog, you might have noticed, that I am not happy about the way rewards are being distributed since hardfork 19 Equality.
Thanks to linear rewards I can blog about nothing else but this topic and pay myself for it. If that makes any sense to you, feel free to ignore the issue.
How to game n^2
For this post, let's look at how easy it would be to collude and 'cheat' with non-linear rewards.
Let's assume:
- Every author has 10 votes/day
- Rewards are based on n^2
- All votes are with 100% Voting Power
- All authors have the same SP
- All votes are worth exactly 1 r_share
If 10 authors colluded and agreed to each make a post every day and upvote each other, they would gain a significant advantage over 10 authors, who each made a post every day and only upvoted themselves.
Even though the first group only made 10 posts and the second group made 100 posts, the first group would get 10x more rewards out of the pool:
Group 1:
10 posts with 10 r_shares -> 10 x (10)² = 1000
Group 2:
100 posts with 1 r_share -> 100 x (1)² = 100
Let's do the numbers for a group of 100 authors, who each only make one post within 10 days, and always vote for each other:
Group 1:
100 posts with 100 r_shares -> 100 x (100)² = 1,000,000
Group 2:
( Let's assume they just post and selfvote 10 times/day for 10 days )
1000 posts with 1 r_share -> 1000 x (1)² = 1000
Even though the first group only made 100 posts and the second group made 1000 posts, the first group gets 100,000x more rewards.
There are a few errors in the numbers above.
One being that Group 2 has an even bigger advantage on any given day.
Those numbers were just to illustrate the idea.
If you were to actually try and game such a system and collude like described above, you would probably soon run into problems with the distribution.
Who's turn is it to post when ?
Obviously your profits would grow by magnitudes if you can grow the number of authors, who collude with you.
If you were to get 1,000,000 authors to join you, sure they would be impossible to coordinate ...
I think you would soon come to the conclusion, that it's the easiest and also the fairest way to pick the posts that have the most relevant content at any given day to circle-jerk over.
If the content is halfway decent, you might get even more people to join in on your scheme !
Congratulations !
By trying to game n^2 I just re- invented social media.
... Oh wait ...
I got it but I make 5post per day and I vote on other people post but no one visit my post because I have less steem.power. people with heigh steem power visit on the post of the people with high steem power.
It's sounds crazy but he's making a lot of sense!
nice post sir and very informative post ....... you calculation is right about post nice work sir.
Does the reward pool under the Scenario from group A accually grow bigger?
I thought that the Pool-Size is fixed and it's only an matter of distribution - some gain, some lose ...
Am i wrong about this?
That's exactly how it is.
I should have maybe mentioned that detail above, too ...
The numbers I end up with are not absolute rewards but only to compare the shares.
Hey, If I am wrong here, you can berate me as you correct me.
If all votes remained the same based on n^2 Haejin and Sweetsssj and friends, would get more or less.
I think a lot more... So, while it wouldn't really hurt large networks with rewards, but eliminate some of the smaller networks, correct?
Can't someone do a mock-up of say.. Yesterday's rewards?
It would also multiply the rewards of those using multiple, large votes on the voting bots such as some of the posts the trending page has been filled with.
Yes, you wouldn't get a fraction of what Sweetsssj is making.
lol, well I don't either way. Nor do I expect to. That wasn't my point is that if spam and crappy posts are being rewarded at high levels, wouldn't that just get worse?
you are absolutely right
important something like this might help a lot
One problem I see would be, that some kind of automation could probably eliminate the coordination difficulties.
Not sure how easy or difficult it would be to automate this though. I'm lacking in the programming skill department. There would probably be the need for some quite complex calculations (including variables like r_shares, Voting Power, order of voting) involved... It's probably a lot more complex than the current way of gaming the system, which I'd see as a good thing.
Probably worth trying out especially if it actually leads to your conclusion. What an amazing idea, this social network thingy you propose! ;)
I think you can handle also 10 accounts by yourself ;)
Is this here something like a rebellion against the system?
interesting post @felixxx
i am your follower and notice that ......
Reading this is making me feel like when i first informed myself about steemit. Its interesting but i still dont understand it fully ^^
What ist the n^2 funktion? Is this based on upvote and reward statistics?
I just jumped into the ocean of steemit there live whales and dolphins like you. I trying to get hag of this thing. Anyhow I am gonna understand it and I will improve myself @bronkong