You are viewing a single comment's thread from:

RE: 8th update of 2021 on BlockTrades work on Hive software

in HiveDevs3 years ago (edited)

I'm concerned that removing the curve is going to invite back the high-volume low-value spam comments we've had in the past but mostly gave up since it became impossible to earn a reward with most low-HP votes.

But with the curve removed, you can, in an automated manner, spin out a very large number of comments all over the place at very low cost and if even 1% of users are fooled into thinking that your bot-posted "Great post!" is sincere or just reflexively want to reward "engagement" then you can make a profit.

Downvoting can work to counter this but it relies on already overworked and under- or un-compensated abuse fighters. Most people don't bother. The balance of effort and payoff very much favors the spammer.

I guess we can see how it works out and then reassess, but IMO removing it and replacing it with nothing is a mistake.

Sort:  

I never really understood this argument. HIVESQL is our friend here and a list could easily be maintained like the current blacklist that can slap down comment abuse of this nature in a similarly automated way.

It just means that our abuse fighters need to be tech savvy and use some automation tools of their own.

They can, but it's adding more targets and workload on to an already thankless task.

Also, downvoting isn't really the ideal solution either. It adds even more spam to the chain, driving up RC costs for legitimate use. It also gets controversial in gray areas (and sometimes not even gray areas but just disagreement with the spammer who feels they should be entitled to do whatever they want with their stake) which further adds workload (and sometimes outright harassment) to what abuse fighters have to deal with.

If downvoted consistently like that, the spammer might give up (though not necessarily right away), but also may just switch to another account or accounts and keep going. It becomes a game of whack-a-mole (which it already would be to the extent that if incentives exist, and I'm pretty sure they do, it won't be one or two spammers doing it, many will).

All else being equal, not having the incentive be there in the first place is much, much better than needing to counter, which is what the curve has done. Now all else is not equal, and there are other factors, but there is definitely a cost to consider.

Adding even a relatively small real cost to post would also kill all of this spam immediately since, as with most spam, the incentive depends on posting being nearly free which justifies the return from a relatively low to tiny success rate. But of course I doubt anyone wants to do that since it would get it the way of normal social use.

Yes, though most abusers fairly quickly stop when they get punished, rather than rewarded for their effort so the RC and chain spam argument is not that strong either.

It does not matter what system is envisioned there will be a way to game it. I question whether we are just shifting goalposts again....even going full circle, but I was wrong about the free downvotes so maybe I'm wrong about this too.

Stakeholders need to provide support to the abuse fighters IMHO as they are the beneficiaries of a more valuable chain. It can be tough for abuse fighters to get the balance just right, but it IS worth fighting abuse and the chain IS improving in quality and integrity as a result of their efforts.....so we should be rewarding them.

It does not matter what system is envisioned there will be a way to game it.

No, that's repeated a lot in many contexts here but it is a wrong argument and I'm pretty sure @blocktrades would agree on this. You set up the system to reward the behavior you want and/or penalize the behavior you don't. Sometimes those can be cleanly separated and the end result is easy to predict, but even when they can't, you can understand the tradeoffs and still set up the rules accordingly.

Throwing up your hands in some fatalistic "no matter what we do it will be gamed" makes no sense when designing and implementing a system like this. The entire premise of all blockchains is based on setting up incentives for certain behavior and (in most cases) getting exactly that behavior.

The free downvotes are a pretty good example. They have worked out more or less exactly as predicted (which included things like retaliation and apathy). They're not perfect but also not getting gamed to any significant degree. The curve too for that matter. It has greatly reduced spam comment and votes. Getting rid of it will likely bring those back. Maybe worth it, but important to think about the consequences and what will be done about it.

Another counterexample would be adding a small direct cost (not just RC which is negligible at this point) to post. I'm quite certain that can't be gamed and wouldn't be gamed. It would simply kill most if not all spam (good) while also discouraging legitimate use (bad). So that's a tradeoff, but it isn't an issue of gaming at all.

Unrelated question :)
Does the HIVE sent to the DHF is being instant converted to HBD, or it is in a slow mode conversion 0.05% per day?

Instant

.

Not even close to accurate. As an account linkage detective, you are really, really bad.

.

..

ROFL, just saw this while replying to your other messages. You are really, really bad at this linkage stuff.

You obviously have no idea how funny this sounds to anyone who's familiar with both smooth and berniesanders to suggest that they are the same individual. That would be a true case of a Sybil attack :-)

..

Is this that big of an issue? I find spam in the hope of some small reward on every social platform. Whether it is Instagram accounts hoping to follow/like people at random in the hope of bringing some % back to their page where they have some affiliate links or products, or Facebook friend requests with links to scams.

Imo, it would be a mistake to trade of any improvement to user experience in the effort to fight abuse. It would be better to lower the barriers of entry as much as possible, to increase user retention as much as possible, and make our "product" easier to understand and use as possible, even if it meant allowing for some more abuse.

We have far bigger problems to solve than any of the worst abuse that our chain is currently experiencing, or has experienced in the past, namely how we can attract, retain and sustain a growing userbase with properly aligned economics to benefit the token value. We can min-max efforts to limit bread crumb farmers after we've solved that.

I don't think we can compare the comment spam on traditional social media to hive. The reward system changes everything. Although I agree that we have bigger issues I feel that removing the economic barriers to comment farming will exacerbate rent seeking behaviour. The real question is if the tradeoff is worth it.

The real question is if the tradeoff is worth it.

That's the question I am asking. And the only point I was making. Namely that the tradeoffs in terms of added barriers to entry of new users, and added difficulties of growth for grassroots communities, is not worth the small benefit of less abuse.

The big picture that we need to look at is that hive is a value extracting platform. That is the reality of our situation as exemplified by the flow to exchanges.

At the end of the day hive is mostly purchased to speculate on it (like 99.9% of all cryptos). More than 42% of the available supply (total supply minus the DAO) is on exchanges. For example bitcoin only has ~13%...that is a big difference.

The potential increase in spam may turn out to be a "nothing burger". I am not convinced that changing the convergent linear rewards curve is necessary but I am not opposed to be proven wrong.

I'd be more interested in comparing with numbers for other smaller blockchain projects rather than bitcoin.

The convergent linear reward curve devalues the tokens of anybody without enough to over come the tax.
It also killed comment rewards.
If it isn't removed it needs to be lowered to 100hp, or less.
160k hp is unrealistic.
Anybody that buys in at less than ~80k usd gets a diluted stake?
Who buys that?

It's not about how much HP you have, it's about how much is payed out per post. I have ~20k and if I vote on content that does not get to the point where the payout becomes linear it is worth about half of what it could be but it is worth more than that on higher rewarded posts.

One the one hand the convergent curve discourages spam but as you pointed out the tradeoff is that accounts that do not get regular votes from high powered accounts have a lower incentive to post or comment.

If we had millions of users with a more even distribution of stake it would not be much of an issue. Content creators with a large following similar to what we see on regular social media could make bank without the support of large stakeholders. But we are not there yet.

16hive x .57 = 9.12htu = ~310k hp to overcome the tax.
Everybody with less than ~310k hp are having their stake reduced in influence on the pool.
If you have 155k hp, your tax is ~25%.
If you have less than 155k hp, your tax is up to 50%.
Would you buy hive, and power up, knowing that unless you powerup over 176k usd's worth your influence is reduced the less you do powerup, then on top of that, if you don't vote 100%, or combine on a post that equals that 100%, you lose money because you are too poor to play with the big people.
I can't believe I've missed this angle for as long as I have, I'm slipping.

I doubt we will ever have a million daily posters, just for the record.
We don't need them, 100k is plenty.

I do think the spam hurts user experience. It's not so much the rewards themselves I'm concerned about it is the secondary effects. 1. Spam itself is distracting, and insufficiently controlled can be an outright denial of service. 2. Comments are very expensive operation on the blockchain (vastly higher than votes, transfers, etc.). Out of control spam comments will drive up RC costs and become an impediment to the value adding user experience we want.

We mostly lack the abilities that centralized platforms have to shut down (some, probably the vast majority of) spammers programmatically, or at all, by identifying bots and such (since it violates their ToS, and we don't have one), and to some extent to delete it or block it (since most UIs are on board with the "decentralized" ethos and don't want to be censors, and again, because it isn't violating a ToS), so we face a fundamentally bigger problem than they do.

The rewards incentive also changes the picture a lot (for the worse in terms of spam), as @onthewayout stated earlier.

Give authors free downvotes on comments on their posts?

Seems plausible.

Sounds like a decent idea and gives the author a little more sense of control over their own blog.

Is it possible to remove Tax Curve from the posts and leave it as it is on comments? Comments under $20 will be automatically taxed and it will be easy to find comments over $20 (abusers will have to vote during the first window, otherwise they will be taxed by 2nd or 3rd window rules)

I think part of the intent of removing it is to encourage small votes on comments, so this wouldn't be effective for the goal.

In the past, I've seen posts on Steem with 5 cents pending payout value and 10 comments under the post with $15 pending payout each (upvoted on 6th day from publish date). Code should protect us from situation like that.
Alternatively, there are many engagement bots like misterengagement, beer, wine... not a perfect solution to promote engagement but if one of them had a strong token price...

I agree, also within Communities if a community mod notices bot comments they can easily mute that account and preserve UX within that community and I think communities will have the most traffic in the end anyway so I think the damage of spam comments in the grand scheme would be minimal.

If I understood correctly only the curation curve is being modified. So this change would only affect how the curation is distributed but the logic behind how posts are rewarded is intact. Payouts below 20 hive will still be none linear.

It's actually something I'm planning to review.

We are taking out the code that is called out as the linear convergent reward curve. This code has a clear impact on curation rewards for small votes (less curation rewards). I reviewed the associated algorithm changes as far as it impacted curation rewards and was involved in the final algorithm selected.

When I asked about impacts on author rewards, I was told there wasn't any (and there's not as far as authors haven't got extra rewards for early votes since HF14), but I plan to double check that it has no impact on author rewards for small votes. Even if does allow for a little more author reward for small votes, I don't think it's likely to cause any problems, but it's good to know all effects of the change.

Yes that is the impact was suggesting. It does impact author rewards (they're increased for small votes).

I do like the idea of getting rid of the extra complexity and buggy sqrt code for something that may not be all that important.

EDIT: I removed the comment I made about the curation curve, since I'm not sure how that will work with the three-piece curation strategy. So I assume someone is working that out.

We are taking out the code that is called out as the linear convergent reward curve

Is the code change setting the reward curve back to linear or something else?

If it's the former then again it will become more profitable to mine/extract value via comments. Although this would not be the same situation as before (when comment farming was rampant) due to the "free" downvotes. I predict it will be an issue.

The immediate impact of the change will be to increase the curation ROI for the bigger accounts and viceversa. Essentially front-running wont be a "thing" anymore.

Rewards will be strictly linear with total rshares.

I'm not particularly worried about comment farming. Not only can downvotes be employed to counter that, such accounts can also be added to popular mute lists that will likely be subscribed to by "newbies" who might potentially vote on such spam comments.

Yes, the larger impact will be that it ends the use of bots to front run for curation rewards, putting manual voters back on an even footing with the bots.

.

The effects were figured out to the best understanding of the coder doing the work. But he's working on code he didn't write, and the code is difficult to understand and has no comments. In such cases, you don't always know if you've understood the code properly until you write tests to check its functionality.

This is doubly true because the code can simply have errors that make it operate in a way that wasn't even intended. We already found a case of this, where a manually written sqrt function contained a bug. So testing and third-party review of both the code and the test results is important. You can't just stare at the code and be sure you understand it entirely. That kind of "ivory tower" programming is a recipe for disaster.

As to witness voting, I vote for witnesses on different criteria. In abit's case, despite the fact that he devotes most of his time to BitShares, I vote for him because he has more knowledge of the core code than most of the other witnesses. This can be very valuable at times. But note that my vote alone doesn't keep him in the top 20. He's fallen out at various times, and I think he's only on the edge of being in now.

/

As far as I know, he does handle his own witness. You're the only one to claim otherwise, and after some of your previous claims, you probably can understand why I'm doubtful as to their accuracy.

But even if you are correct, and he does hire someone to do his IT work, I don't consider that a big deal. He's the one that is responsible for its operation, his reputation is on the line, and I believe he will act appropriately.

...

The convergent linear curve affects total post value so that's both author and curation.

This is separate from the curation change, at least for the most part (there may be some interaction, not sure).

The plan is that post rewards will be a strict function of total rshares voted (including downvotes, ofc) with the new algorithm.

Yes, understood. That's a change from the convergent linear curve.

Because rewarding is always in context, I don't think they should be taxes like the current system.

In general, I see no problem.

spread hive in many hands is better as cycle jerk around. The best thing to do is with comments.

I love to upvote comments. And I do, doesn't matter the extra tax. But I would more like to see no extra barrier for it.

If there will be comments with 50$, they get downvoted and all is fine.

My logic behind this is, if many people have a 1-5 hive it is better as a little group has Millions for the long term.

The 5 hive people are less likely to wait 13 weeks for power down. So it's more like a burn. Sure some do, but most will sit on it forever.

So this should be long-term more beneficial to the network.

Nobody disputes that comment engagement is good. And in fact my concern isn't even about the rewards themselves, but about the negative side effects of having a lot more spam. That's bad for (sincere) engagement in a lot of ways. I'm sure we've all seen comment sections and such on the internet which have been heavily polluted with spam, or tried to use email services which don't filter spam well, etc. None of these are enjoyable or sometimes even functional at all. They also don't have the additional challenges of a blockchain with finite shared resource capacity.

So I'm really not sure what is the best solution. I'm not even proposing that the curve not be removed, I'm simply raising an additional concern to be considered, and I'm reasonably sure that it a valid concern in terms of the bad side effects I mentioned (again, primarily not the rewards themselves, but rather the potentially large damage caused by spammers chasing those rewards, even if the rewards are very small)

Spam will come in anyway. The problem is easy to see on steem. The rewards doesn't matter. There will come link spam if no active curation happens.

That's what on steem happens.

IMO on hive the problem is not that big as on steem. Because the community manages it.

If the first SEO tools include hive, like gsa T2 /T3 and other ping/spam tools, there will also spam happen here.

But this is not linked to rewards. It's all because of search engine optimization. Special SINA ( China search engine and the Russian one).

The algo doesn't penalty spam links in the same way as google does.

to prevent, the easiest way would be:

Reputation under xyz = link is blank, no hyperlink, or link + low reputation = comment doesn't show up.

So downvotes matter in the case of reputation busting from spammer.

Reward farms get easily downvoted IMO. And are not worth the effort. There is more money made by spammers with affiliate spam + hide link direction via buffer side.

Noobs will always try :) No matter what we do :D

I don't think the proposed strategy would be a successful means of getting rewards.

And if someone does employ that strategy, and it is working, I think there are ways to mitigate it without another reward change, such as suggestions by frontends to new users to follow black lists and mute lists that lookout for abusive bot accounts.

Implementation of user chosen mute lists are coming for peakd so hopefully that helps (they do depend on people creating good lists that are also labeled well so that users understand the list and want to use it). If needed we could consider something like a statistically driven automated badge that indicates a user as a heavy repetitive commenter. Even perhaps possible to put that badge on comments. This gives an option to be made aware while not going as far as muting their experience. #ideas

The current reputation serves a similar function, but I like your idea based on spam level (although we'd have to work out useful metrics), as something like that could be a more objective indicator.

It was successful enough in the past that people did it. It costs effectively nothing to post comments, so if you get anything, even a little reward once out of 1000 comments, it's a gain. You need HP, but you can still use your HP for voting or whatever else (as long as it doesn't use too many RCs).

Or for that matter, even if it isn't quite all that successful but shows enough promise (occasional rewards), people may do it, which is still harmful. It's not something where you lose a lot of money by being wrong, you either make some or make none, but either way spam up the chain, cause annoyance on the social layer, and increase everyone else's cost for RCs.

As for the other mitigations, yes that could work, we can try that if needed.

it's true that gains are possible through this change, but let's don't forget the positive developments that could occur from this as well... engagement of users! I think it's worth a try

I don't disagree.

I agree on the other hand.

I agree on the other hand.

If you are looking for a way to earn some extra income every week. Look no more! Here is a great opportunity for everyone to make $82/per hour by working in your free time on your computer from home. I've been doing this for 6 months now and last month I’ve earned my first five-figure paycheck ever! Learn more about it on following LINK.......>>>>>>>>>>>>>> https://www.Nifty2.com