It's Time For HIVE to Pivot on Anti-Abuse | Focus Group Anyone?
In reference to @neoxian's post The Stupidity of Hivewatchers , I have been both a participant and observer of anti-abuse efforts on both STEEM and HIVE. I'm talking more about the response to plagiarism, citation, etc., and less about obvious abuse such as spam. This post has plenty of merit, but I'd like to take a different tack. While I feel anti-abuse is important, the current methodology we as a community have been supporting has been missing the mark for a long time. I quit participating in anti-abuse simply because it was quixotic, thankless and downright soul-sucking, attributes that would be better served by a non-human.
@klye was on target with his response with regard to what I feel is the correct technical approach to anti-abuse:
As he stated, the human element needs to be removed from the equation. The real question is does the technology exist in order to feed JSON data into an API in order to detect plagiarism? If one conducts a simple google search for 'plagiarism api' or 'AI generated content detection API', one gets multiple hits. From that we can at least agree on the possibility the technology exists and more exploration is warranted.
@steevc offered his opinion in regard to the human element of anti-abuse, and I must say that I agree with him
I have seen some of this heavy-handedness at times, and unilateral decisions being made regarding people's continued ability to earn rewards. I in no way want to discount the difficulty of policing human behavior, especially when real financial incentives are involved, but a tiny minority of humans doing so in obscure fashion is not the answer.
Here @thunderjack remarks on the lack of transparency and recourse:
The lack of on chain transparency has always bothered me, and the community is often told to dig through discord to find the information or evidence relevant to various cases. That was always hit or miss for me.
Here @hivewatcher basically eludes to the futility of their own efforts:
An infinite struggle to fight invisible users? Would these same users be invisible to a well-tuned algorithm?
I don't know enough about the individual cases referenced in the post above, but we have had an entrenched anti-abuse system in place, one that has neither changed nor innovated from what I can see. I've seen lots of bad actors get ensnared and rightly so, but I've also seen good folks get ensnared for reasons that are less clear. I'm not here to pass judgment, make accusations, or impugn anyone's character that participates in HW (@guiltyparties, @logic, et al.). What I will say is that we need a better and more impartial system, a system with explicit rules and predetermined responses to detected abuse. Some ideas:
- When abuse is detected, an automated response is provided along with the links to the detected sources
- An automated reply system that allows the user to either provide proof of ownership, or notice that proper citation was inserted.
- A mechanism by which a warning is first issued, and subsequent violations are met with incrementally higher penalties. This would be all visible on chain, allowing the runway for people to improve their behavior, and for curators to tune their blacklists.
- Provide a decay mechanism for penalties (eliminate or minimize the use permabans). If someone posts a certain number of compliant posts, their penalty will decay at a certain rate
- Detect AI and provide the probability that AI was used to generate the post along with what percentage of what is thought to be AI generated
Let's be honest with ourselves. We will never be able to detect or prevent all abuse. What we can do is up our detection game and provide the larger community the information and tools to participate in the policing of the chain, and rely less on empowering a few humans and all the faults that come with that. Centralizing this effort to a pseudonymous group presents all sorts of legal and ethical implications. We don't want to centralize any sort of power, particularly one that could be used to chill speech.
So how do we move forward? As a community we need to come together to discuss the feasibility of alternatives. Whatever the solution will be, it will not be free. We'll need compute resources, paid access to API's, etc. Could it be accomplished for the 145 HBD per day the community is currently paying for the centralized service? I'm not sure, but it's worth exploring. The real question is do we as a community have the resolve to develop a new solution or do we continue to criticize the current one? I feel like the technology is out there, and we have the resources to divert to such an effort.
If anyone would like to start a focus group to explore alternatives I am willing to join. Let's determine what resources we need, both technical and financial, and get a proposal together. Even the input of HW team could be invaluable to creating an enduring solution--perhaps this would free them to pursue more worthwhile endeavors. Please CC anyone that might be interested in the comments below.
cc: @enforcer48, @pfunk, @antisocialist, @smooth, @azircon
Getting @themarkymark's input will be invaluable as well. This needs to be resolved soon. Kicking the can down the road has gone on long enough.
Where will this take place? Discord? Some other location?
I agree on both points. I'm hoping we can gauge interest on chain and go from there. I'm ambivalent about joining/creating yet another HIVE discord, but if that's what it takes...
I do not see the reasons those things have to be discussed on a platform outside of Hive; it always smells of 'behind closed doors'. Why need everybody to sign up on a different platform, using another app?
Maybe I am wrong, but the comment section on Hive should be good enough for a well thought-of discussion?
That is because it is behind closed doors.
That is why I have eschewed such platforms and am only found contributing to society through mechanisms that aren't utterly vulnerable to covert mechanisms. As AI is ever more able to substitute for human interactions on machine mediated society, it is becoming apparent that it won't be long until I am only able to participate in person.
We have a very limited window of opportunity to try to create mechanisms of interaction that aren't liable to degrading society through the substitution of AI for human interaction.
By interacting on the blockchain we prevent the wholesale replacement of history with fiction. As little as that is, it is what is available.
After years of 'nice times on the internet' I am happy to be mostly back in real live.
It is nice to have some scrolling for the morning coffee, but even there, I started to read short stories in good books.
No way I am getting back into the anti-abuse camp. I devoted a large portion of my time here to it and has gotten me nothing but headaches. The amount of work involved is insane and the end result is nothing but bitching and moaning for it.
Yeah, I remember those days.
Well said! Let’s explore solutions and find ways to innovate rather than simply complain about the status quo. And rather than simply doing nothing.
Yep!
Congratulations @joshman! You have completed the following achievement on the Hive blockchain And have been rewarded with New badge(s)
Your next target is to reach 81000 upvotes.
You can view your badges on your board and compare yourself to others in the Ranking
If you no longer want to receive notifications, reply to this comment with the word
STOPTo support your work, I also upvoted your post!
Check out our last posts:
Support the HiveBuzz project. Vote for our proposal!
Most of what you're saying is exactly how HW works right now but you want to automate it. So what you're talking about is basically a machine learning
bot(edit: used incorrect terminology, meant 'system') capable of processing an entire chain. It's possible because anything is possible to build on Hive but just designing and testing the concept before any development would be pushing it. Cheetah, which is a minor fraction of what you're describing, was a massive bot that cost thousands a month to operate. To answer your question, no, it can't be done 145 HBD per day.What can be done is people providing feedback on HW scope, which is rarely done. The scope is meant to change and it does as Hive changes. You've been around enough that you know how it goes in anti-abuse. It's a shit job, everyone is disenchanted with it, everyone thinks not enough is done, its thankless. I just run servers and deal with phishing and even that is depressing as all fuck.
The hysteria some people start overnight because a friend got caught for something and suddenly they're very into HW/anti-abuse is a regular resurgence. Much of it is based on the remnants of the expectations of the old "Steemit Slack" where if you start shit you can force whomever you want to submit. Old habits die hard. We go through this every six months or so on average and then no one cares anymore and getting feedback on anything is like pulling teeth.
So, you are just doubling down on deflection.
This has been an ongoing grievance for ages. And nothing ever changes because it’s much easier to pay someone $145 to muck around instead of actually building something.
Everyone makes excuses for HW, for Logic,etc and their shortcomings.
If I could build this with just two developers I would have. This isn't just a script with a few functions, it's a massive project.
We've been all talking about anti-abuse for years. If HW was you and whomever else instead of Logic and Nuttin it'd be the same thing because there's still scope and there's still the same frequency of botnets, fraud, identity theft and everything else. Everyone knows that's the case. Whomever doesn't believe it can go to the HW discord and try and help guide people. There's no secret there, it's all out in the open, there's no private channels.
Cut the horse crap. Your system ain’t working and people have been demanding changes.
All the excuses you give is just to keep the status quo and make sure two pointless persons are getting paid.
If we can fund the HBD stabiliser, we can fund development of meaningful automation. You and your sycophants can go take a hike.
The HBD Stabilizer is not even close to what is being described here. We're talking a system similar in complexity to the best multi-function corporate surveillance systems.
I'm not going to argue with you on the other points because I know its personal.
You won’t because you can’t.
Nah it's because I'm going to sleep now.
I'm not sure why you would want to process the entire chain rather than monitoring it in real time for posts that meet a certain criteria and then punting them to a checker. Maybe that's what you meant. The bulk of hive traffic is not blog posts. Why would I want to feed splinterlands battles into a plagiarism API for example? There should be an arbiter for that built into the system. If Cheetah for example were processing token sales on hive engine and the rental market on splinterlands, and checked them for plagiarism, I would argue it's poorly designed.
You'll get no argument from me here, which is why as much as possible should be done by a 'trainable' algorithm.
For me it's more of a question of 'out of sight out of mind.' If it's not happening on chain, it didn't happen. That is a problem, because it should be on chain. Ideally if someone breaks a rule, I should see their dossier posted in a reply. You plagiarized, here's a link to what you plagiarized, and here are all of your past transgressions. If I see someone I recognize was downvoted, I shouldn't have to run it down on Discord. From what I remember a large percentage of your time on discord is spent defending your actions because their impetus was not evident on chain.
You have to process every post in order to monitor it. So what you need to do is have a system to run these posts through. One, with a simple but expensive Google API like Cheetah (Cheetah is a 'similar content' bot that searches every post for similar wording via Google). Two, with a reporting form (simple too). Three, with custom reverse engineering tools to detect not only AI but translations from obscured sources (creation of this would be the most expensive part). Four, with "account history". There's no way to avoid massive infrastructure and extremely high costs in development and operations. I believe Tiktok and Facebook are attempting to create similar scale models. Not saying its impossible, but would be more complex than anything else ever built on Hive. It's an interesting concept though that would push anything ever built on any chain. Whether it would be ethical to build is another question.
Regarding Discord: even if a comment is clear on chain people don't read it or want to hear the answer again. That's just human nature. We took out the auto comments which spam the chain and other pre-Hive-related protocols. Those prevent the person from appealing and moving on.
Let's assume HW is not here since start. People do complain alot about their right to spam with as low effort as possible. They also get mad when whales flag or any flag happening in any part of blockchain.
Either it is about HW or whale doing this antiabuse, one thing is clear
People will complain about downvotes even when downvotes are necessary like running a botnet case.
The solution people sometimes demand is just blurt with no downvote button. What I observed is these people don't even like blurt.
I do agree with @enforcer48 points in general as leonis cares about chain in general.
But we should also consider filtering out opinions based on logic. We just can't turn hive into blurt overnight nor we want that.
@guiltyparties @enforcer48 this opinion is not limited to this thread but what I have seen so far all over the chain. Appologize for my limited short sighted opinion.
no, it is because of the way HW are behaving.
Im going to take up the challenge to provide more specific feedback on scope now that I got my rant out.
While I've had some concerns about hivewatchers there always needs to be a human element in any anti-abuse system.
Any automated system can be beaten by smart humans. AI is actually pretty stupid and always will be.
Also justice and fairness requires the availability of human review of all decisions made by automated systems.
Centralised social media is such a mess because of over-reliance on automated systems because humans are expensive. But on Hive we have a self policing community that moderates content for free.
Hive has a sustainable model, AI is not.
I don't disagree with the need for some human oversight, however whether a human feeds the post into a text box, or it's fed into an API, the resulting findings would be the same. You could also build it so an accused person could present their counter claim on chain. For example in a direct reply to the bot: I really own this content and here is the evidence of that; or my mistake, I forgot to add citations, and here they are.
One or two people doing it manually is sustainable? Is it scalable?
The whole point of upvoting and downvoting is that the whole community moderates content.
Yes, first and foremost, I'm arguing that an algorithm should present the whole community with the information they need to make informed voting decisions, rather than it being obfuscated by taking place on discord.
Delegating responsibility to a couple people is the whole community moderating content?
This is the core idea, and it is mostly being distorted by the steep distribution of power. With a flatter distribution of power, this idea of the community self-regulating could mybe work.
I am not sure if this can ever be fixed; powerful actors would have to give up their power.
Like in real world, it e.g. would need a power-cap. It is bizarre that someone here has 1000000 x the power than others. Quite a reflection of the world out there...
In my opinion it is absurd that we have to talk about policing in that way.
Hive is getting more and more distributed all the time. The proportion held by Whales is always decreasing.
But there will always be variation in HP. It reflects commitment to the platform over time.
Those that have committed more (time, money and skill) to the platform will have larger holdings than those that have only a passing interest.
This is how it should be.
But should larger holdings (and profits!) equal larger power? And if so, to an unlimited extend?
I am not sure about the numbers, but right now I would guess that the distribution is not unsimilar to the one of wealth on the planet. Very very very uneven.
I think that there is a way out of this dilemma, but for this we all should admit that Hive as an experiment of maximum decentralisation has failed, and move on from there.
We literally have one person in HW deciding on what is and what is not abuse. It's worse than centralized media.
We need more eyes and volunteers to maintain an algo.
We aren't asking for a machine to decide it all. We need a machine that can aggregate potential problems for the community to review. Not one dude and his one or two mysterious partners.
the solution is simple, let the admins of each community moderate their community. problem solved.
Hello.
We do check all reports and post all comments manually although we use a certain format for different types of abuses.
Automation is very costly.
We ran cheetah for a while during Steemcleners years. It was detecting copypasta and plagiarism automatically. It was a very expensive service to run.
Is it more expensive than you(HW)?
If I remember right it was something between 800 and 1200usd per month, but that was a long time ago, I could be misremembering.
As I have been "active" on this blockchain since May 2018, I read both this post and the one from Neoxian. Both from very ... "seasoned" ... members of the active base of real people, in the real world, who try to "engage" in this "virtual world" ...
As stated elsewhere in all of the associated content, comments, etc., we are talking about what struck me very early on about my "journey" through all that this "virtual world" has to offer. What's that?
Human nature. And all the "infinite" permutations we all experience through sitting down and interacting with each other through this medium ...
I won't take the time to say much beyond simply expressing the fact the way "justice" is uhhh ... "administered" ... inside this ... "ecosystem" ... has left me ... "cold" enough, that I seldom "waste my time" contributing the content I once did (yeah, I know ... Lots of quotation marks => Lots of different ways to say more or less the same thing ...)
Why?
"On a scale of 1 to 10" ... the injustice I have seen repeatedly far outweighs whatever benefit I might perceive from the "justice" also ... "administered" Your own words are "heavy-handed." Whole posts have been, are currently, and will be in the future written about how others perceive it.
So ... A "black and white" solution "administered" via some code? Setting aside all the obvious "blood (time) and treasure" questions, written by who? Then "accountable" to who, for the "finished product" results? Based on what?
"Good luck" with this last question. It has "driven" all of the "answers" thus far and ... We're still at it ...
As someone familiar with your character, this is a sad thing.
All valid questions. I think the first step is to get good information. In network security we have a device called an IPS (Intrusion Prevention System). If you simply added it to your network and turned it on, it would basically kill legitimate traffic. An IPS has to be trained for weeks to months on what constitutes valid data, so it sits there in passive (intrusion detection) mode until you turn it on. I see this the same way. Let's start by figuring out how to get good data, and present it in a way that is transparent to the community, and build up from there. That may fall short of 'automated justice', but at least provide the community with the tools it needs to police itself. Automated justice could be applied much further down the road where it makes sense.
Now retired, I was blessed at the end of career with the opportunity to attend two world events related to Business Analysis. In one of them, I learned about an aspect of "AI" that I will never forget, from an international authority and a very gift communicator.
He explained that AI is almost universally known to stand for "ARTIFICIAL Intelligence." While he said that most certainly was applicable, AI had other equally important meanings and he then talked at length about "AUGMENTED Intelligence." His presentation was unforgettable.
My point in raising that memory now is taking what you are referring to here and simply pointing out that "augmented" may be deemed ultimately to be a better solution than "automated." All of the "hard work" and "heavy lifting" would be continuously executed by these "expert" systems and then the information "handed off," along with recommendations, to whoever was given the final authority to act upon it.
My $0.02 and (debatable?) "worth every penny" ... 😉
Please everybody be aware that what is being discussed is the (re-)establishement of a policing body. Maybe this is the only way to get Hive out of the gutters, but will it be Hive then?
Wasn't there an idea of a self-regulating platform? If so, I see two reasons this does not work:
If nobody could have 1000000x times the power than others (e.g. a cap of 3x, maybe 10x?), and everybody (or many) would participate in both sides of the coin (up and down), we could skip all the ruling.... DONE.
(I am a utopist, but this is what atrracted me to this platform).
I've been advocating a 1000mv cap on voting.
Not a coded solution, but a voluntary cap on what the most powerful take from the pool.
We had this experiment on steem, as you can imagine the only people that were in favor of it were those without the power to keep it.
When 20 accounts take 50% of what is available to all ~4k of us, there is a power disparity that won't fix itself.
You can't really blame them, we are trapped by crapitalism after all, but we don't have to support them, either.
We could denounce them, but no self respecting sycophant would have that thought occur to them, so here we are, a few voices pushing back and thousands of onlookers not capable of defending themselves from this sort of exploitation.
The more I think/read about it the more I come to this conclusion:
It is time to throw old ideals over board for the sake of reality.
Let's just admit that the experiment of decentralisation has failed, and move on and/or see what we can salvage.
I lost a lot of illusions while reading a book from the 16th century describing the exact dilemma we are facing on Hive (Utopia by Thomas Morus). And I am thankful for each and every illusion that I loose.
Lol, never give up hope, rage against the darkness, is my motto.
Read these two books and let me know how their treatment of crapitalistic math fits with your world view?
https://archive.org/details/ironhee00lond/mode/2up
https://archive.org/details/lookingbackward01bellgoog
They will burst some bubbles for you, they did for me.
Thanks for the books, I need brainfeed ;)
I've got dozens, some quite voluminous.
This is short story along the same lines as the other two.
This is an academic treatment of the ideas.
I've been involved in the blogosphere for close to 30 years now. Granted, I am somewhat new to blockchain, however every blogger I know has had the means to moderate their own blog. The same is true on the Hive blockchain as each community has a team of admins fully capable of moderating their community, and guiding contributors through the hoops so that these contributors stay within the bounds of the rules.
To have one individual, or a very small group of individuals claim the right to determine, punish, and control how actions of appeal can be performed, who hide behind a wall of anonymity is problematic to say the least. I consider it unacceptable.
Hivewatchers/guiltyparties/etc. have put themselves in a position where they are lawsuit waiting to happen.
Bottom line, it should be up to each community to moderate their community members, not an individual, or a small group of people who are not involved in said communities.
You volunteering?
https://discord.gg/tVjRSwTG9v
It is up to each stakeholder to ride herd on the anti-abuse crowd so that these things don't happen.
Folks are failing in their responsibilities, no surprises there.
I am not running or administrating a community, The admins for each community should be responsible for monitoring their community. If, and that's a big if, I ever decide to start a community, I would be involved in monitoring only that community.
For me, blogging is a hobby. I do it for fun.
Welcome to the hive.
I agree, we are here to have fun.
agreement is good.
most of my regular blogging was from around 2002 to 2010, politics and current events. I burned out, and time was a precious commodity with raising a young family and my being older, shall we say. A few years ago I cleaned up my original blog, purging all of the political stuff as it was no longer relevant. I even stepped down from serving my community in elected office in 2010 after 6 years of serving my community, and that position only took up a few hours a month for meetings. That was an educational experience (government functions, budgeting, open meeting laws, and a host of other stuff I had to learn) and was worth doing.
currently I am content just having fun and writing about things I enjoy.
thanks for the welcome message, appreciate it. As for volunteering, give me a couple of weeks to stew over it. Maybe I can provide some useful input. Mostly, though, anyone that wants to act as a policing unit on the hive-based blogosphere is setting themselves up for a lawsuit.
Do you support continued rule by force?
I agree. I don't know how it could be done with technology, I think algorithms fail many times(I've had my Facebook account blocked a few times for nonsense like that xD) And it would be nice if only ONE warning before blacklisting someone and start chasing them with the spaminator bot. Especially if you see that the person made a mistake due to being new or ignorance, not intentionally, and has not repeated the same mistake again in any other post. In my opinion, I would consider it much fairer, and this can be done even without having technology, looking at the individual case of each user. I agree that if someone repeats the same thing again then yes they should be put on that list.
I understand that it is a hard and arduous task but I think it would not be so complicated to differentiate who abuses consciously and who does not.
If we want the Hive community to grow or new users to join we must take into account that at the beginning we will all make some mistakes because we are not perfect, it's okay to put a punishment or warning, but according to what happened.... Because in this way one takes away a little the desire and the illusion to publish, and does it with fear. Besides, when you are on this list it is as if you were an outcast. 😅
@thunderjack
"anikys3reasure" was a farmer that was farming automated votes from neoxian. She's neoxian's friend.
She was spamming with effortless google translations of SKALE articles from the SKALE website and from different blogs published by SKALE followers.
She was allowed by SKALE to share their content but they had no idea that she sold it on Hive (monetised). They likely thought l that this ambassador promotes SKALE in places like Medium, Twitter or Bitcointalk. They had no idea Hive is different and their content is sold/monetised here.
She never told them that she sells it on Hive.
The spam and milking were going on for years and started on Steemit.
She has almost no engagement with the Hive community. A few comments in 2 years.
She never promoted Hive. She still promotes Steemit in her SKALE ambassador description:
.PNG)
Her Skale portfolio intentionally omits Hive. It only mentions her profiles on Medium, Twitter and Telegram. She had no intention of promoting or mentioning it, in case someone goes to check up and sees that she sells SKALE's articles on Hive. https://skale.space/blog/meet-the-skale-protocol-ambassadors
.png)
We contacted SKALE twice and they never confirmed that she was allowed to sell (monetise) their content.
why did you tag me in this comment?
AND
These two claims are an obvious contradiction.
After "anikys3reasure" reached out to SKALE to contact us. She asked them to confirm that she is allowed to share. She did not mention to them that she sells the articles on Hive. They contacted us and said that she is allowed to share the articles. When we asked back if she is also allowed to sell their articles, they have not responded back. I have just contacted them a third time, using a different contact method. We are awaiting confirmation.
This user never promoted Hive anywhere.
Many similar spammers are happy to milk Hive while they are too embarrassed to mention Hive anywhere.
We have blacklisted dozens of similar users since 2016.
What they do is they apply in some blockchain project to be their spamming shill (nicely reworded to "ambassador" so it looks nice). They get paid from these projects in tokens.
Then they go on Bitcointalk, Medium, Telegram, and Twitter and spam with copypasta of the project's articles. Sometimes they use Google Translate to turn them into another language.
They also crosspost spam to Hive to milk Hive (while getting also paid in tokens from the project that they shill).
Doubling down on failed strategies essentially confirms Mark Twain's observation that it is easier to fool someone than to convince them they've been fooled. The fundamental issue isn't the failure of humans to be rational, but of substituting pecuniary interest for far more valuable societal aspects.
Humanity is sacred. We are sovereign, inherently endowed with love and affection our free will enables us to benefit one another by. Insofar as money has become a surrogate for societal interactions, as we can see has utterly overwhelmed IRL government and civil society, perhaps best illustrated by the malevolence of Bing's AI revealing the psychopathic intentions of the infestor class that has commissioned it, the substitution of social values in curation on Hive is replaced with financial interest, as is governance via stake weighting.
It is the fundamental substitution of money for societal values of far greater worth to humanity that is the cause of the problem, and automating suppression of the problems this causes Hive, which is a society of people, is doubling down on that fundamental malformation of the platform. Every increase of substitution of human interactions on Hive will further deprecate society, degrading people by equating them to toasters.
Resolving this malformation of Hive will require enabling actual human values to be promoted and financial interest to be deprecated. This is possible, but water is wet. It flows downhill, following the path of least resistance. It is far more difficult to be reasonable and succeed than to go with the flow and dissipate into the rocky substrate of financial subsumation of humanity ongoing globally today.
Thanks!
Just saw this post, I mean the damage is done. They trained a couple of hundred professional Hive haters and released them into the wild. The Brand is completely destroyed in some giant crypto communities outside of this ecosystem. Good Job Hivewatcher, since you're not the one taking the heat outside of your discord cave.