Post by @bvdaniel • Hey
Let's talk about bots,
Since day 1 we knew there were bots in the network. Just like the ones that started following me today. They all do the same wit
Comments
- Great thoughts and ideas! I agree that there are good bots that we all use for automation and then also bad ones. For the prevention not following bots in the first place is very crucial, so keeping your own follower graph healthy does a lot already for not seeing bots on your feed. The other place is replies where the friends of friends comment module reduces the bots by most degrees. Next thing to implement that will help is reply ordering. For followers something that helps if we could present accept a follower feature or show who of your friends are following the users. Would definitely want to hear more ideas from the community.
- That’s an excellent step forward!
I am not sure about the point 5. Maybe so far you can differentiate bots and real people, but what happens when Lens gains a lot of users, could you do it then? Twitter has the same approach, and a lot of people get banned or suspended because of the bad algorithms.
Here’s something more to think about… How about adding a level and duration of engagement with the account posting paid mirror post, or some other account, why not 😃? This could be an interesting metric. Definitely not bot proof, but might be a good edition.
- It's certainly getting tricky to detect them over a longer period. Classic Whac-A-Mole...
Also I'm not 100% against those "bad" bots - also for them to participate on @wav3s.lens, if they help amplifying the content. In the best case, they're not allowed to reply - and get a different reward. Like 0.01 for bots, 0.1 for normal followers, 0.2 for verified followers.
- Thank you. I noticed some just say “ok” and had multiples. I fell for the first one, and gave them a follow.
- The problem with Proof of Humanity is that it's not 100% reliable. They might not be bots but you could still have "puppeteers" or malicious agents, as there are several ways to jump the security of the list and the many filters of the onboarding process.
Same happens with the curation tools from the Circles UBI project. There, ideally (although not 100% in the practice), you give the trust to people that you have known or met personally, and you need at least three persons to give you trust. I think this is a bit more secure since you should be - I repeat - ideally, meeting the person, which almost never happens with PoH.
In the end, I think the big problem relies in the fact that once received the trust the person doesn't need to keep verifying its humanity. That creates always a tricky point in which you can fake and/or jump the rules. If the verification would be dynamic (i.e., automatised, every time that you connect your wallet, or once per month, something like that...), then it won't be impossible to trick the system but at least make it harder to break.
Maybe a combination of both ways could give even more reliability? In any case, PoH it's a wonderful tool to already filter a lot. It's a lovely subject to discuss though, and it's wonderful that you're considering solutions for this big issue because yes, most of these bots are super annoying.
- segmentation is good for long term growth as well. as web3 social grows, campaigns with bigger budgets will want that anyway. Right now most people on here have overlapping values so it’s not needed but with more adoption that will change. Guild.xyz is a cool idea for larger spend/reward campaigns as well
I like the idea of proof of humanity as long as it’s not too hard. bots can still get through but I’d have more trust if I knew I was giving most of the $ to real genuine humans over bots. With mirroring you want it to reach other people, especially if you’re trying to get real comment engagement
- I was not so aware of this but suspecting that some of my followers are bots. I try to not follow back if I have that impression. Strangely a good filter for me was if @crypto.lens_profit (Please notify me if I’m doing wrong to you) is following a profile it is most likely a bot.
I’m mostly wondering: what is the motivation of these bots, or the people behind them?
- Thank you. I noticed some just say “ok” and had multiples. I fell for the first one, and gave them a follow.
- segmentation and clustering are imo the key for successful campaigns. If bots are catalysts for more focused campaign distribution, I take it :-).
My thesis is that people share lots of data on lens that might come handy for clustering - by posting or collecting ;-).
- True 💯
- But bots must have lens profile?? How is that possivle
- Adding a proof of humanity requirement might reduce the userbase, but it will increase the value of @wav3s.lens for promoters since they have the assurance that their content will be shared by verified (and ideally active) accounts.
- We have to find a way to do something about bots. We have block scan and other surveillance technologies that come with open and transparent blockchains & we should use them to decipher humanity.
- I really appreciate your thoughtful approach to this issue. It's great to see how the team is proactively working on solutions to tackle the issue of bots. I think the idea of segmenting users by interests and providing proof of humanity will be especially useful in ensuring only real people are engaging with content. I also like the idea of a DAO of the people promoting content - this could be an effective way of moderating the platform and ensuring that only quality content is being promoted. I'm looking forward to seeing how the team's solutions play out!
- Great subject to discuss and I believe it's one of the negative pillars to be solved when the topic involves web3. We came to do something different from everything that exists. Unfortunately, financial value is a big divider in the current world and sets an initial limit.
- I think Bots should be allowed on the platform, but disable some features for them. For e.g, allow bots to Post and Comment but not Mirror or Like.
- I like the Guild.xyz.
I believe they have been able to solve part of the issues with their solution.
- Couldn’t resist using a “bot” on this post 😉
- beep boop 🤖
- im not a robot 🫥
- I think with the technology development, it will be harder and harder to find who are bots. The only way that we are different to prove and enjoy is to write, speak, create soul content which let people feels warm and special.
- Always on point well said
- Can we really fight bots?
- good topic to discuss
- its obvious there are many bots or say spams which is everywhere and they can be identified too and i agree on your points you mentioned to filter them.
not blacklisting but maybe can give a cooldown or a suspend for few times until they raise ticket and prove them as human 😅
- Wow... That is great
- Seems to me that having some minimum amount of currency would probably solve this issue, or at least, make it more expensive for people to set up bots
- I want to add to the discussion the emergence of **really good** AI bots. In the (near?) future, we will have AI bots that are able to post creative content, grow followers, etc. The same bot could be on many different platforms (github, lens, ...) to further grow it's credibility. But all with the intention of malicious behavior, for example money scams and democracy disruption. Like human troll factories but on a much larger scale. Also, the bot, unlike a human, doesn't care if it gets banned or gets it's reputation destroyed. My prediction is that in the future, some kind of physical based identity may be needed, when it gets harder and harder to tell that a bot is a bot.
- As we're on the topic of banning and censoring and so on. How does the mechanism around banning etc works on lensprotocol? Let's say I say smth real bad against humanity and or just spam it, what's the process around that?
- The evaluation and screening of robots is not determined by a few simple rules, especially in a decentralized world like web3. An idea, whether in the future VR and other fields can increase the biometric function, and then become an attempt method for web3.0 screening robots~
- ☕️
- Love to see the content from our fearless leader and founder, shoutout to @zadan.lens for leading the #lozpride here on lens !ROAR