Anti-Spam in the Era of Bots and Automation

in steemit •  9 years ago  (edited)

How do we fight fire? With fire of course.

Many of us have started to notice the influx of spam as bots start posting automated content, usually plagiarized, to our wonderful website. Right now, there is not much a single user can do to prevent these posts. Personally, I am sick of it, so I created my own bot to help fight the abuse: I have started gathering a Ban List, and am adding accounts to it that are notorious for spam. They will be automatically flagged (downvoted), and a reply will be posted.


How can YOU help?

You can help in three simple ways:

  1. Always flag posts that are spam, plagiarism, otherwise low quality, or clearly posted by bots.
  2. Find bots that are spamming: report them to us (so we can add them to our ban list).
  3. Upvote the good bots! The more power you give to our good bots, the more weight they have in their downvoting of spam.

Going forward.

I am hoping to partner up with other wonderful people, many of whom are active in the #steemitabuse channel on slack. I am going to continue working on my bot, to make it better at detecting bad content.

If you are interesting in helping, there is a bounty available for plagiarism detection:

https://steemit.com/steem/@cryptoctopus/usd500-bounty-we-need-this-now

And you can always join us on slack! If you have concerns about the bot, feel free to chat with me there.



My personal bot, @cheetah, is now on the prowl.

Photo by User Mukul2u , from wikipedia.


Special thanks to the following people for being awesome in the spam fighting initiative:

@xeroc (for Piston, an instrumental tool), @cryptoctopus, @pfunk, @ash, @neoxian, and all the others.



A quick edit, using some of this attention.

I personally really support this proposal: https://github.com/steemit/steem/issues/176 Please have a look! I hope that this bot is only a band-aid, and that we can fix the real problem at the roots.

Also, there is now a nice spot to help with flagging: #doyourpart

Authors get paid when people like you upvote their post.
If you enjoyed what you read here, create your account today and start earning FREE STEEM!
Sort Order:  

I have been told to upvote you for the following reason(s): I am a bot, I do what I am told.
I am a robot, to help support the stop of spam, you can upvote me to increase my power.

GO CHEETAH, With our powers combined, YOU ARE CHEETAH.

CHEETAH, He's our hero - ganna take bot spam down to zero.
Do dooo doo dooo

dances away

lmao

  ·  9 years ago Reveal Comment

More power to you cheetah!!!!

Power up our swift little helper ;)

awesome lol

  ·  9 years ago Reveal Comment

This is great. How can I help?

  ·  9 years ago (edited)

Thanks for the support! My second point in the post has the three simple things an average user can do to help! And please inform me of any CLEARLY abusive accounts!

  ·  9 years ago (edited)

<3

seems like your bot is already learning humanoid feelings... I robot here we come.

@anyx Happy to help the project too! Does the bot have a repo?

Hey! Thanks for the support. For now I am going to keep the source code private among the fellows active on the #steemitabuse slack, since this type of code could be used maliciously.

  ·  9 years ago (edited)

Hi @anyx, highly respect your project, but personally disagree with this notion/philosophy.

In my opinion, close-sourcing code is not the correct solution to a problem of malicious use. Firstly, it doesn't actually work: eventually malicious code will be written and open-sourced. Secondly, the system needs to be exposed to malicious code in order to develop strategies for counteracting it.

There may be other motivations, such as economic and anti-competitive ones, to close-source your code. If that's the case, I would suggest to an intellectual property developer to state them up front.

Thank you for the feedback! I am working on a more sustainable solution with @pharesim, and we will be making the bots' new logic open-source.

This is how censorship starts. Automated content filtering.

You should openly post your methods for determining spam, and openly post your attack list in a post that people can see. For Transparency.

  ·  9 years ago (edited)

The end goal I have in mind would have public lists similar how Adblock works. Otherwise, indeed, it becomes a slippery slope -- one I hope to avoid. (To be honest, I am scared of someone using this maliciously as well.)

See my edit above for what looks like a better long term solution. For now, I am focusing on very clear abuse, and you are always welcome to chat with me in the #steemitabuse slack channel!

Yes I see you have joined the crew that is intent on creating (and controlling) a policeforce and ai bot censorship tools. This group of "protectors" are literally creating the tools that decentralized networks, and cryptocurrencies are designed to evade. Building a censorship weapon is complete overkill for simply mistagging something, or even spam.

If steemit would just implement personal blocklists, 90% of the spam wouldn't even show up after 1 view\flag of a user or tag placed by curators marking the spammers.

The rules and chat about this AI bot, its rules, blacklists, and keywords should stay here in the open where it affects people, not behind a password protected chat program.

I agree with some of your points. I do not believe ANY bots should be able to run rampant here, but the fact of the matter is that they currently can. And it being possible to build a censorship tool, is in itself an issue.
Personally, I hope that once the owners of steemit understand the implications of both "good" and "bad" bots, they will be more incentivized to figure out how to prevent or hinder them.

I agree.
I wrote up an alternative proposition that allows curation and individual users to take responsibility for fixing the problem.
Take a look and share it if you think it's better than building censorship bots.

https://steemit.com/steemit/@tinfoilfedora/instead-of-creating-automated-censorship-tools-allow-the-users-to-fix-the-problem-through-curation

I also feel the same: 9x cheetah clone bots downvoted a post to oblivion because it thought the tag was wrong.
If you are going to have one bot thats perhaps reasonable, but multiples? That is not anywhere reasonable in my view, centralized power.

I don't have much sympathy for posts abusing tags (like in this case, using the Photography tag for talking about money, and NO PHOTO in the post to boot).
But since there is some backlash (I guess people enjoy tag spam?), I will tone down that down.

Abusing tags is a problem steemit can solve by simply allowing curators to retag things from the comments.
It does not warrant the creation of automated censorship tools.

I personally really support this proposal: https://github.com/steemit/steem/issues/176 Please have a look! I hope that this bot is only a band-aid, and that we can fix the real problem at the roots.

Indeed, this proposal will help with spam a lot! Spamming will be much less profitable. Good work on the bot too, I think more people will make similar bots over time.

  ·  9 years ago (edited)

Sounds good like a good way to fight spam. Thanks, mate.

  ·  9 years ago (edited)

x

steemit is decentralized, isn't it? How can a captcha appear on blockchain?

  ·  9 years ago (edited)

x

Incidentally, i would love to help any way i can. Im not much good with programming, but i have not insignificant hosting resources that i can add to the fray if you need them...

If you have relatively weak cheetas able to help me out, i am being targeted by a spammer who has at least some bots on your list... I don't get shit for votes as it stands, so starting at -13 is really damaging, even if theyre 13 relatively weak votes.

https://steemit.com/doyourpart/@sigmajin/spam-bots-and-retributive-downvoting

I wish I had saved a little voting power for this. Excellent work! Thanks for taking the time to do this. And thank you to the rest of the crew fighting spam.

Awesome! I hope this will work out as great as I imagine. :)

You are a knight in shining armor and a sight for sore eyes. A big upvote from me for taking the initiative to help the community like this. I will also do my best to help in the ways mentioned above.

Cheers!

How does this bot work?

Good luck, a worthwhile challenge, although I would assum it will be hard work as more people join steemit. Unfortunately SPAM is common place in many parts of our lives!! Some progress is never good although it is still progress!!

This is good news. I spend a lot of time in the new section, and the amount of spam thay floods in is horrible and worrying at the same time.

W

This is good to hear from you.

didn't understand !!!

Good on you - well supported by the look of things - I 'm in

Omg you are a hero, i really wanted something like this but it would be pretty hard atm to get someone to do it by myself, yet now you make it easier. Making the bot public imo would seem a good idea at the beginning but people would start adding people simply because they don't agree with them, or like them.
I guess it is a problem that we will be able to handle with time.

Great initiative!

@anyx please add @weenis to the blacklist for spamming "great post" comments. Looks like a bot to me.

hey @anyx i appreciate the efforts but your bot is currently downvoting @wang 's posts. maybe put this account on some sort of whitelist...

As silly as the wang bot is, it is not on my list. Can you point me to where Cheetah is downvoting wang? I do not see it.

  ·  9 years ago (edited)

Sorry it is my mistake. I just looked for this post and saw that it actually downvoted someone who replied on a comment from @wang . But the color on my screen is so bright that i must have missed that downvoted comment. So all is cool from that end, sorry for the confusement :)

This was the post i was refering to: https://steemit.com/introduceyourself/@trader999/i-m-crazy-fan-steemit-1#@cheetah/re-re-wang-re-trader999-i-m-crazy-fan-steemit-1-20160720t173600746z-20160720t173705

Love it! My favourite of the big cats. :-D

CG

I'm rooting for cheetah.

Pitching in myself as well. I'm finding that any post that seems like it was plagiarized from an published article, probably was. A cursory Google search with random paragraphs is doing the trick for me. Many hands make large work small!