Steem Robot Manifesto: The Rules of Botting
By @zackcoburn and @jasonmcz.
For those of you who aren't familiar with us: we were the market makers on STEEM internal exchange who recently came clean through open source.
From @wang the welcoming bot to @cheetah the anti-plagiarism bot to @isaac.asimov the reading level bot, Steem has recently seen an influx of robots. Some have been downvoted, some have been upvoted, and some have gone to war with anti-bot robots like @steemservices and @antispam. Even @dantheman wrote a piece about his view on this subject. We believe that the most optimal course of action moving forward is for Steem users and robot makers to reach some common ground. That's what this Manifesto is all about. Think of it as the Rules of Botting.
Before we get any further, we need to come clean about our intentions. We see a big opportunity for bots on Steem that add real value. @zackcoburn and I even attempted to make a bot, @jeeves, that replies to new blog posts with a list of related posts with the same tag or from the same author.
Here's an example of a comment by @jeeves:
We got a lot of feedback, mostly through direct replies to @jeeves.
As you can see, the feedback was mixed. Some of it was very positive, and some was very negative! Based on the feedback, we decided to discontinue @jeeves for now. We want to be certain @jeeves is providing enough value to the community to justify letting it loose on Steem 24/7 thanks to the valuable input from folks in the #steemit-abuse channel: @nextgencrypto @anyx ( and his bot @cheetah) @pfunk @repholder @bacchist and more! Since then we've seen a discerning growth of abusing USELESS bots on steemit which is why we think it's time for us to voice some opinion today.
This brings us to Rule #1 for Steem robots:
Rule #1: A Steem robot must provide real value to the Steem community.
So how does a bot maker know if his bot is providing real value? Easy, turn it on, and read the feedback. You'll get a lot of it. If your bot is providing real value, people will upvote it, and you'll start earning rewards. If the average post earns a penny, it's probably not adding much value. If the average post makes a dollar or more, it's probably adding value.
Rule #2: A Steem robot should value quality over quantity.
It's pretty unlikely that your Steem robot is so valuable that it should comment on every single post. If your bot is commenting on something every 30 seconds, it's commenting too frequently. We think a few posts per hour is a good sweet spot.
Rule #3: Conversational Steem robots shouldn't start conversations.
If your Steem robot is one that provides value when someone interacts with it, it should avoid starting the conversation. For example, if your bot can be used to ask a question like "@bot what is the current price of Steem?," your bot shouldn't be making posts like "Ask me for the current price of Steem!" Write a single blog post describing how your bot works, and let people start the conversation with your bot.
Rule #4: Your bot should provide value to the readers, not the author.
Authors write to their readers, and you should too. Don't directly address the author or provide information that would only be valuable to the author.
Rule #5: Your bot should announce itself as a bot.
Don't try to write a bot that fools people. Your bot should say something like "Hi, I'm a bot." There's a place for the Turing Test, and Steem isn't it.
Rule #6: Don't judge for the entire community.
This rule applies to humans interacting with bots. Don't assume your opinion about a bot is the opinion of the entire community. Don't go on a rampage against bots you don't like or bots in general. Just because a bot isn't valuable to you doesn't mean other people won't find it valuable. You might be a whale, but you aren't a god. It's not your right to be the ultimate judge. Also, you should not write a bot for downvoting other bots. Upvotes and downvotes should be natural, human votes. Otherwise bot makers won't get a good sense of real feedback from the community.
Think of this post as The Rules of Botting 0.1. We think bots are a reality we're going to all deal with as Steem matures. So we might as well start having this conversation now and lay some simple groundrules we can all agree on. What say you, Steemians?
Thanks for the post @jasonmcz. Lots of bot posts lately! I'll weigh in.
Rule #1: A Steem robot must provide real value to the Steem community.
I do not agree that it should be about the money you get. Cheetah gets more downvotes than upvotes. A bot can provide value in other ways.
Rule #2: A Steem robot should value quality over quantity.
This one needs to be smacked into the forheads of some bots. And that includes jeeves, which spammed people who didn't want it. It should only follow those who want it, and assume by default people do not.
Rule #3: Conversational Steem robots shouldn't start conversations.
I definitely agree with this one, and this is exactly why calva got auto downvoted until it stopped spamming.
Rule #4: Your bot should provide value to the readers, not the author.
This is an important one as well. The author wants to read reader responses, not bot responses.
Rule #5: Your bot should announce itself as a bot.
Agreed. In addition, for most bots it should not even announce itself, and only follow those who want it.
Rule #6: Don't judge for the entire community.
This is subjective. I always take feedback seriously about cheetah, but it will first and formost follow my own philosophy. There is no one that said plagiarism doesn't belong on steemit. I said I don't want it, and I tag (not flag) people who do it.
It just so happens that many people agree with me.
Reading the rule #1, I have to agree with you on that although @cheetah is sorta like a unicorn among the array of bots. I've seen authors downvoting cheetah, not the commenter nor the bot, just because Cheetah posted the original link of the blog that it detected from the internet..
@jasonmcz There are also those of us who will downvote cheetah when she gets it wrong as a way of signaling to the owner that attention is needed on that post. It's a courtesy for them that saves them getting a huge backlog of complaints.
@anyx if you ever see me flag @cheetah that's the only reason. I also upvote cheetah when she gets it right.
I've taken to flagging any bot that isn't contributing to the conversation and then putting it on mute so I don't have to see it again.
I personally am a big fan of cheetah! =)
Also great post, brings about great discussion points. so thanks.
I agree with @anyx on #rule 1 tho. A bot can (and maybe even should aspire to) provide value in other ways. Now also interesting to discuss is what we mean by 'value' maybe?
As for the last rule of subjectivity, it is interesting, cos especially some bots, as in the case of cheetah, are essentially an extension of their human's moral-values/drives/motivations.. the more the bots have a unique 'persona' the less generalized comments would get I think. people would make more targeted remarks, I'd suspect. I also agree with upvote/downvote being reserved for human interaction. Unless someone by some machine learning magic one day comes up with a smart-ass bot who actually has a taste and curiosity of it's own. =D
Subscribe-able bot service is a good idea (follow). Plus summon on demand.
yep @abit! We launched our bot again! check it out!
I would even go so far as to expand on #2 - the quantity should be based on the number of users requesting a "useful" bot.
Useful bots should be something that people want to use. They should be something that provides context to the content when asked to do so.
Maybe an improvement for @jeeves and a way to bring it back online right now is only respond to posts that mention it by name? I'd use "Hey @jeeves, share some links" or something worked into my content.
I'd probably even upvote the comment it left, because it's bringing value to my post without me having to work for it.
Steem isn't the place for the Turing test.
Yes! I am constantly checking people's responses to see if they are a bot or not. This is exactly correct.
@mctiller, I do the same thing sometime as well!
Step 1) Write automated response to some post
Step 2) Check if original author responded to step 1 message
Step 3) If yes check for a ? (has the author posted a question to the automated response?)
Step 4) If yes send the human to convince the author
90% of the work done :P
I had thought @jeeves was a pretty decent robot. Sorry to hear was discontinued. Prehaps in a future version could assign a few random stories? May make it less desirable than just trending stories, trying to think how it might draw exposure to overlooked articles that are still quality pieces.
@clevecross thanks for thinking that way -- this is exactly we want to bring up an valuable conversation here with the community to discuss about this matter. Perhaps in the future we will continue it when bots like @jeeves are more acceptable to the community and not banned by other bots.
Absolutely. I love when comment sections turn into discussion panels. Random stories on rethought may not be great as would defeat the purpose I saw @jeeves trying to serve. But something that explored content without just top trending that share a tag.... Be a challenge for sure. Maybe a community supplied "pot" of posts that cycle out inside a week?
Yep that could be a possible idea. I've opened an issue on Git regarding the problem of non-discovered good contents by linking relevant posts at the end of each post, in which case, there a handful users are already doing that by manual.
Thanks for this! I completely agree, as i've started getting very irritated by bots. If they abide by the above rules then atleast they are here to serve a purpose. Not just a bot that pretends to be a human wanting post rewards.
Thanks for this.
Once @nextgencrypto was asking me what was the purpose of programming the bot -- the question was directed towards @jeeves the content linking bot -- and I answered we are just doing an experiment to see whether there is a room for a bot like @jeeves and whether community would find it interesting.
I think a lot of devs saw @wang how it was able to get rich so quickly and perhaps that's their motives of creating these upvotting bots with the hope of making $$$$$ from curation.
Bots in any case need! but they must be very high quality and the need to help people (especially beginners). You wrote a good idea!
That's why some introduceyourself bots are viewed useful but I personally found those upvotting bots annoying.
This is a stick at both ends
Money is the origin intention for some robots, not the rules, the final rules are the steemit gui presentation and blockchain protocals. MAKE RULES THE CODE!
Some rules are just hard to implement onto the CODE level without imposing too much sensorship tho. I wish Bot devs can self regulate themselves at some point but it's tough to hold people to their moral standards when we are on the internet.
@jasonmcz I would say that this reflects the STEEMBOTS code of conduct explicitly with a couple of exceptions.
#1 There may be owners who explicitly do not want bots in their threads at all. Make sure to respect it if they have gone to the trouble of adding #STEEMBOTSTAY explicitly, regardless of what the people in the thread want.
#2 If you run a bot, tag it. Make sure the owner is known so people know where to complain to if there is a problem. You don't want people calling the dog catcher when it's running lose, make sure your dog has a collar and proper tags.
Otherwise, this is great!
I hope you'll join the STEEMBOTS initiative.
Thanks for bringing this up... I thought @jeeves was pretty cool. The only part I think pissed people of was that it was usually top of the replies...i.e It distracted readers from interacting with other steemians' replies. Bringing up the bot say after the 30 minute mark would be nice IMO.
Rule 5 is perhaps the most important to me.. Fooling people with one-liners aint cool and will certainly piss people off if they discover they had just replied to a bot... We all seek human responses to contents posted.
Looking forward to how bots evolve on the platform. Thanks for laying the foundation.
@infovore, yep! in early hours, @jeeves was doing that and I am pretty sure that was the reason why lots of people were annoyed by jeeves -- they were joking in the channel asking did you get "jeeve'd"
LOL
And WAY too much text. Even in the last iteration you had.
Keep it simple and small, I keep getting that feedback for cheetah too.
I think @upfunk's feedback was actually the best. As of now, we are not planning on bringing Jeeve's back but we will see. I have opened an issue on Github waiting for dev to take a look if not implemented perhaps we can target one of the sub tags, for example (perhaps) photography? by scanning photos under the same topic?
what do you think?
Good points!
I totally a agree with you well done :) bots should announce themselves as just that bots on a whole though I dont minds bots they are part of this new eco system
exactly!
Thanks for this post! In case the rules a ordered hierarchical, my comment would be to put rule #5 to the top of the list. As I see it, a bot without making an entry statement like "Hi, I´m a bot" is already a problematic one, even if it proves to make valuable contributions. I´m looking forward to follow the discussions triggered by this manifest proposal.
I think all rules are equal in our case but I see your point on #5 -- it should be the first test which is
Announce-yourself test. All bots must pass this test.
Thanks @jasonmcz. I'd love to see some type of visual indicator that a bot is a bot. (emphasizing #5 in your post). i.e. once the account has been flagged as a good bot, change the icon.
This is actually a great idea not sure when they can start letting us upload avatars!
Nice artice, thank you
Hello! @jasonmcz I am very much disturbed by a problem of boats. I would like that you as the creator of one of them took part in discussion of my post: https://steemit.com/steemit/@jennamarbles/i-think-that-bots-can-affect-steemit
@jeeves, find related posts
@jeeves, find related posts
This is a good feature
There you go the bot!