You are viewing a single comment's thread from:

RE: A Radically Updated Steem Whitepaper

in #steem4 years ago (edited)

Just curious about something...

It was never explained why the previous n^2 algorithm and its ability to mitigate abuse/exploitation was eliminated in favor of a full linear algorithm, which has led to a lot more widespread/rampant abuse of the rewards system. The new algorithm requires not a single shred of "consensus" when it comes to the allocation of rewards.

Does this not present a problem for dealing with serial spammers and collusion, like we have been seeing since the last hard fork (19) was implemented? Are there any discussions or plans to address this, or will we just simply have to continue playing catch-up with all of the new abuse/exploitation - and continue to waste tons of voting power on something that was adequately mitigated by several different protocols which have been removed/changed in consecutive hard forks?


which has led to a lot more widespread/rampant abuse of the rewards system

This is an easy claim to make and when it comes to widespread I would agree it is obviously much more widespread. This makes perfect sense because under the previous algorithm the bulk of the user base (like 90%+) had virtually no ability to grant rewards and therefore no ability to abuse. That's like starving to death as a weight loss diet.

However, when it comes to the 'amount' or 'severity' of 'abuse' it is very difficult to be confident about that conclusion because excessive concentration of rewards and a few whales essentially siphoning off the entire pool (or nearly so) to a few preferred posters could also be considered abuse.

The role of posting, comment, voting, and rewarding in this system is to encourage widespread participation and growth of the platform. The precise allocation of rewards is less important, especially when that allocation is flatter and broader.

The new algorithm requires not a single shred of "consensus" when it comes to the allocation of rewards.

It does require consensus. If A upvotes and B downvotes (with equal weight) then no reward is paid. That is precisely a consensus-finding algorithm. If the community is unwilling to take a stand and downvote, it can't be helped.


If the community is unwilling to take a stand and downvote, it can't be helped.

Agreed, but please help us by making down-voting anonymous. This way we don't get burned by stepping on the wrong foot. I see worthless posts that really deserves a flag, but then I check their SP, and decide not to pick a fight.

Yes this is a problem for anyone with low SP but at the same time if it wan anonymous then people could just go and down vote anything they didn't like all day long and no one would then be able to hold them accountable.
Transparency is the savior and we must protect that at ALL costs.

I agree that transparency is valuable. Unfortunately, it is making the fight against misuse harder. Maybe we need a more creative way to solve this problem. Otherwise, this problem is crippling the whole system. Low-quality content is being rewarded more than high-quality content. I am sure Steem was not designed to do that.

BTW, I have just read a post by @heavey about a new ICO that is similar to Steem. It is called Red Pulse. They have devised some solutions to guarantee high-quality content. Maybe, we should check them out.

I don't think that transparency is the problem. The problem is that almost everyone is here to make money and as usual willing to compromise honesty, integrity, fairness etc to get a piece of the pie.
I am sure there is already a solution that exists to deal with this and I feel the solution is just that this community curates content they feel is valuable and flags stuff that isn't. Guilds/groups are the way to go.
If a lowly wimpy minnow or plankton finds flag worthy content they can just report it to the group and the group with all the power can do the down voting if the warning meats criteria.
Its actually very simple, then the entire community becomes the security camera that sees EVERYTHING and the whales become the enforcement. Its perfect.

I think accounts should not be able to upvote their own posts. This way abusers will be forced to create other accounts to upvote. Those other accounts won't have a lot of SP and as such won't be intimidating to users who want to downvote their posts.

Maybe in the current situation a lot of people who would have downvoted the posts were scared off by the SP.

Its just as easy to transfer or delegate sp to a sock puppet.

The account posting spam and the account with the SP will be different. Other naive users will be more confident to downvote a spamming account with little to no SP. The SP account might still retaliate but might also have to deal with many naive users who downvoted his Spam account.

Let me try and give an illustration:

Mob boss (SP account)
Puppet (Spam account)
Government contract (Rewards from self voting)

Currently the mob boss can go for government contracts and do a bad job. People will keep quiet about it because they do not want to suffer the wrath of the mob boss.

If the rules didn't allow the mob boss to get government contracts, the mob boss would have to get a puppet to get the government contracts. If the puppet does a bad job, people who are not aware of the puppet's links to the mob boss will act to punish the puppet. If enough naive people act against the puppet, the well-informed might join in to punish the puppet. This is how revolutions happen. It requires enough naive (or brave) people to act.

Moral of this: it is easier to scare off individuals before they act than to scare off masses after they act.

I hope I made sense.

It makes sense but not sure I actually agree it would apply/work here on this platform.
Anyone with little sp/rep will not do flagging if they get retaliated on, regardless of it its from the boss or henchman. People are just as scared of the henchman as the boss. Because everyone knows the boss calls the shots and the henchman dish it out.
Steemit does have potential to be a place of honor and ideals but is currently very far from that. This community would probably actually be a lot better without the money yet none of us want to give it up either hahaha

Unless a spam post has two votes and a big reward, it requires some work to identify henchmen and people are generally lazy.

Laziness kicks in before fear.

Somebody will come across the spam post, check the posting account REP:

  • if low, they won't even bother to check the SP
  • if high, as when the boss transfers SP to new account and posts with the old account, the SP remaining won't scare off downvoters

Independent of the questions raised about whether this is a good idea, there isn't obviously any mechanism for doing this. To limit spam, the voter needs to be 'charged' for the bandwidth and vote power, which means the voter must be identified. To alter that would require some radical changes in the system design.

Agreed. The benefits of HF-19 far outweigh the negative impact of the spammers.

when it comes to widespread I would agree it is obviously much more widespread.

I agree with this statement in the most literal sense.

that is to say, under n^2, the abuse was more concentrated. It was a very few people making six figures annually (steemguild, curie, etc).

The total percentage of the reward pool going to spam/abuse is, i suspect, probably about the same (and if its higher, ite because many whales got fed up with fighting abuse and now just sell their votes), there are just more people getting some of it. .

Obviously, neither situation is optimal, but if we're going to pay out a significant share of the reward pool to abusers, the abuser community should at least be inclusive.

The Witnesses voted for linear over n^2 and that is why there is linear. Overwhelmingly, it appears to have positive results in terms of user engagement and growth. Consider other alternatives than n^2 to mitigate "abuse."

I'm sure there are many other ways to address the spam caused by self voting, none of them perfect, I just wish people working at Steemit were more engaged in the discussion of possible solutions, There has been at least 15 serious posts specifically about the issue and possible solutions and nobody from Steemit commented on them if only to acknowledge having read it.

Hopefully, they're actually already working on it behind the scenes.

Even I just wrote about the issues here.

I haven't exactly been here all that long, but the problems are obvious.

We'll see how long they'll let them drag on for fear of upsetting the "witnesses."

If nothing valid is done to combat the problem, we'll all be witnesses to the downfall of steem. :P

As much as I love Steemit, Steemit is very well known for horrible communication as well as PR.
Could be so easily fixed. Maybe one day, one day it will have to be in order to reach the masses thats for sure.

well, that's all detailed in the whitepaper.

Consider other alternatives to full linear in order to mollify new users and users who are not well-versed on how to build their social media presence/following.

And also consider that there were other protocols that have been eliminated, as I stated in my original comment. A series of changes has led to the amount of abuse/exploitation that we see today. If you'd like to know more about what's actually happening around here, feel free to visit:

Also - talk with some of the whales around here who have been doing their best to counter the abuses. People like @transisto and @berniesanders, and other users like @pfunk and myself. If you want lists with figures, I'm sure we can compile some...and a few users are already doing it.


Consider other alternatives to full linear in order to mollify new users and users who are not well-versed on how to build their social media presence/following.

Why don't you be more specific? Do you mean increasing the slope of the reward and curation curves?

It could be interesting trying to improve behavior by allowing these curves to pass an array of float couples that represents a polynomial function. For example if you pass [['1', '1']] it's like the current flat reward curve. [['1', '-1']] for square root curation curve. [['0.5', '2'], ['-1', '1']] for f(x) = x²/2 - x and so on. Maybe in that case just doing a % of quadratic + a % of linear.

Would love Steem to be able to offer those choices, but as I have literally told you before, IME it hasn't been found to be safely implementable. However, you're of course free to code it up and propose it to the Witnesses

Consider other alternatives than n^2 to mitigate "abuse."

Maybe you can be more specific as well?

It's one thing to just ignore what I typed in my original comment. It's another thing to go off about "float couples" and "polynomial functions" instead of actually answering my initial questions honestly. You know...the ones about abuse mitigation that had previously existed but now does not.

If you willingly choose to be ignorant about what's been happening since the last hard fork(s), then just refrain from engaging in discussions about it. Send me one of your other team members to honestly talk about it. Drowning the discussion with posturing, deflections, and distractions gets us nowhere.

He is not ignorant. They know what they’re doing.

He is not ignorant. They know what they are doing.

Why not use an extended error function curve (A Half-Gaussian) for rewards, this would allow newer users to be restricted in their voting strength, with moderate users getting more of a vote (those who have published good content and have made enough in order to have a vote), while still keeping the higher-steem users having more of the vote, this would seem like a much better 'In the middle' approach between quadratic and linear reward system.

Hi @ats-david, as a minnow, thanks for sharing your experience. Do you think more blogs that help to point out the worst offenders would be a welcome addition to the community, or frowned upon? e.g. Perhaps weekly lists of users worth flagging, or similar. Thanks

I think this is exactly what we are trying to get away from needing, and exactly what points out we still have many problems that need to be dealt with.

Can someone help explain this to me? I am a new user (joined yesterday) and all i see is a ton of bots and services to pay to promote your posts...not at all what I consider curated content. Before I decide to spend any time or money with SteemIt i'd really like to understand the plan for preventing that and surfacing actual content.

Content curation/filtering is the responsibility of the interface. Steemit or other interfaces can continue to rely on community curation or scale up using machine learning and other techniques to reduce spam and bot interactions.

@ned I'm sure the communities will raise the level of user interaction in both content creation and curation. They will make content so much more discoverable. I'm not sure if the feed will have changes to it too, to maybe include recommended content? Can't wait to get use the new UI/UX! I'm sure interaction levels will significantly increase.

Ned, you sound allot smarter about blockchain tech than you did before :)

Hum... that's all u got?

What about using a captcha? And why wouldn't Steemit want to filter that? Am I the only one who doesn't want to read a bunch of automated bot comments?

Captcha is ineffective for the reason that the underlying infrastructure is publicly accessible. Steemit can be circumvented by any willing body.

That's a good point - but Ned while I seem to have your attention for a minute here - maybe you could help me understand why someone who isn't into the crypto-craze would want to use SteemIt. Right now to me it seems like a big game of programming bots and/or gaming the system to promote posts rather than to surface good content.

One example of a community i read frequently is Hacker News - if something like that could exist on the Steem platform (meaning the quality of the content) and the contributors could get paid for their contributions, then I think Steem could become huge. Right now I guess i'm not sure if i see how that can happen.

Sorry I'm not Ned, and don't know him, and I hope you don't mind my thoughts on your question. You see, I'm not interested in the crypto-craze.

maybe you could help me understand why someone who isn't into the crypto-craze would want to use SteemIt.

On steemit, I do not need to have hoards of angry bee's following me to take care of any troll issues I might have in other social media. I think the reason for so few trolls on Steemit is that they know what ever they type it will be around forever, and that they can't type, slam someone, then edit and make the other guy look like an idiot. There is a record of the edits. Lack of Trolls is a big plus on Steemit. Another reason, is that even though I am not into the crypto-craze, it is nice to see that my comments on another person's post could be rewarded. Bonus free money even if it is only 0.001 cent or less. It's not the money.

Sometimes, not a lot though, I have my own thoughts to share with people, and it is pretty easy to do, yeah it is hit and miss on how many of the 200,000+ real people, (I saw one blog that there may be upto 10,000+ bots), would see my post. But that does not matter to me, why am I here, for the content, there are so many great writers on here, so many artist, and I do not have to be distracted by the content being obscured by advertisement I am totally not into. No ADS, Great Content. That is why I am here.

Thank you for writing this. I am here for exactly the same reasons. To write about life to share your thoughts or your little adventures, to vote on posts that have real content, to connect with people on the other side of the world, to learn how they live and the difficulties they encounter, etc. etc.
It's great to be part of this community and although it's very nice if people actually read your posts, and write a comment, its primarily the joy of writing itself, a kind of diary for the future. That is why I am here!!

Great response! Thanks for being so detailed and clear.

I actually don't see that much spam or bots and it sure beats the depressing negativity that is a constant in Youtube comments.

Would you consider making the Steemit post/comment views data publicly available via a JSON RPC for example? This shouldn't take much, and would really help collaborative filtering and other machine learning to improve user experiences.

put .json on the end of any url on, like the one you are on right now

Oh, wait, that doesn’t do comments.

Those are all already available via JSON-RPC from steemd at

That's useful info (which I didn't know) thanks, but I meant the views data as stored in Steemit's private database.

I'm doing quite a lot of work with the blockchain data, but it doesn't give the user page views which would be really helpful. Anyway, I don't expect it's possible, but thought it worth asking.

Yes, I'm aware of how the process works.

I'm not following. Witness choices is WHY there is a change.

That's a bit disingenuous. You personally reshuffled the witness cabinet to insure that outcome.

I retract this claim.

That didn't happen. First of all nearly all of the top 30 witnesses (meaning the top 20 plus the next 10 who might reasonable by voted in their place) were in favor of it and second of all there has been relatively little reshuffling of the witnesses at all for months (for better or worse). Unlike some other proposals (including ones which Steemit very strongly supported and were not activated), there wasn't even much controversy over this one.

I remember there being some shuffling close to the hard fork time, but maybe I'm thinking of HF 17? Do you remember if there was more reshuffling then?

Anyway I have apologized for the claim, should have thought more before saying that.

maybe I'm thinking of HF 17? Do you remember if there was more reshuffling then?

I don't specifically remember but that is very possible. HF17 was more controversial (and ultimately rejected/withdrawn).

Sounds about right. Thanks

I don't claim to remember every detail of this past year and a half but I respectfully disagree. I had removed all votes and ordered personnel under contracts at steemit to do the same.

Oh, my mistake. Sorry for the false accusation.

@andrarchy did mention this to me about steemit's personnel under contract, not allowed to vote on witnesses!

Yes the hf19 was very bad for a lot of people, I still earn cents from it, even after having more that 1300 followers and posting everyday, I hope the hf20 will be fair enough to change the situation

Coin Marketplace

STEEM 0.68
TRX 0.10
JST 0.076
BTC 58134.78
ETH 4688.92
BNB 639.71
SBD 7.37