Are bots bad for Steemit? - 🤖 BOTS! 🙀 Act 2

After taking time to research and write about the whale experiment, hard fork and some new ideas for apps, I'm returning as promised to my bot series.

The previous post was Act 1 and was mainly concerned with the question "What are bots?" to establish a basis for this discussion.

Bots, always controversial. Why can't we only have people on the network, you know, like a social network?! Society is made of people, not robots. 🚫 🤖 ‼️

I think new users are especially put off by them, particularly those of a non-technical background. There's something suspicious about robots going around, making posts and votes, that you cannot talk to and that are not ... human.

This post will lay out some of the arguments for and against them, with my own comments and arguments.

TLDR for @mynameisbrian and @schattenjaeger: here's a picture I found on the Wikipedia bot policy page 😜 "Let humans and machines be friends".

The problem as it appears

Most of us want the main activities on Steemit to be used by people, and assume that it is built for direct usage by human persons. We may express a preference that:

• posts are works written by people
• comments are written by people, on topic and in response to the content
• votes (both up and down) are given with the active consideration of people, at a rate which a human cannot do (i.e. not 24 hours a day)
• rewards should be fairly distributed to people, by people and these should be earned

Bots can do all of these things automatically and tirelessly, some better than others. It can be felt that because bots can attempt to "game" the system using advanced (and not so advanced) analysis, they will damage the platform and make it inhospitable for real people.

Why this is might not be a problem

Bot traffic is now just over half of total internet traffic. This might sound very depressing but really it just shows that automation is increasing, as is the intelligence of that automation. We are delegating more and more of our tasks to machines and software bots. Bots always work to serve humans in some way. To quote one of the authors of that report from my last bots post interview:

In the end, most bots - from feed fetchers to SE crawlers - are simply there to gather data for human consumption.

In this way bots are like any tool, they can be used for good or ill. They can give you an advantage. They can be so normal you do not notice them anymore, that they seem part of the furniture.

Why this might be a problem

The best arguments against bots I think are:

1. bots unfairly empower the bot owner
2. bots can't read posts, with two flavors:
   1. bots do not provide engagement
   2. bots cannot really determine quality

I am excluding however one common thread - emotional feelings such as the general dislike of bots on the platform.

But to briefly address this, I will say that disliking something is perfectly valid, and actually does impact the platform because if a lot of people dislike an aspect of it, it is a problem. It's hard to argue with though, because it does not stem from a rational position. If this is you, read on and perhaps you'll come to more clarity to either strengthen or dispel your gut reaction against them and their existence on Steemit.

So the arguments:

1. Bots unfairly empower the bot owner

This is a concern and the main concern that lead me to create a free, open source voting bot, and start the Steem FOSSbot organization (sorry for endless plugs 😅).

Tools are only unfairly empowering if they are extremely restricted and unfairly available. This is a main argument of gun ownership lobbyists, and the root of the famous quote "A well-armed populace is the best defense against tyranny." Every person should be allowed to be armed as a constitutionally protected right, and then no one, not even the government, is unfairly empowered, and thus, the argument goes, there is more liberty for more people.

Opponents say that some of us are crazier than others, and that guns are just too damn powerful, they take away life. I prefer to think of bots as tools though and not as weapons. It's the idea of responsibility, personal liberty and the contentious nature of controlling their power that rings a bell for me.

On the internet and with software, equality of access to tools is more real that it's ever been before. It's only a matter of time until someone comes out with an open source version of a needed tool, and all closed source bot writers know this and that the days of their advantage are numbered.

2. Bots can't read posts

What about bots reading posts? Obviously if this is possible on a level anywhere near people we are again talking about a different kind of AI than most or probably all bots here have.

2.1. Bots do not provide engagement

This is the attention economy of @beanz and @demotruk AKA engagement / views-less-than-up-votes problem of @stellabelle. It seems self evident that on a social network, no one can have a meaningful connection with a bot (unless the bot is extremely humanlike, but even then... 😅). Well almost no one, as this poster asks:

I'm single, and I write my bot's in Node.js, so they are generally less attractive then the Python bots, but still, is there any danger I might fall in love and have relations with my bot?

@heretickitten I think you're an expert on bot emotions and communication right? Maybe you could comment 😉

Regardless 😅 what does actually matter is that bots are not humans who read content. And having content read is the highest of priorities for authors.

The fact is that bots do not provide engagement, by definition, engagement is defined as human interaction with posts. Engagement is the crack cocaine that most authors crave, and they presume it's what makes the platform, or any social network, tick.

While it's true that without engagement there is no social network at all, I believe it's wrong to assume that bots cannot coexist with humans in this, especially on Steem. Bots have a different purpose and modus operandi than humans. As stated above it is ultimately in the service of humans, but a bot's direct action does not have the same immediate purpose as a human.

Unless bots actually push people off the platform, we will still have and still need an overwhelmingly mortal populous. I think this is not jeopardized by bots and can instead be bolstered by it.

2.2. Bots cannot really determine quality

It's quite obvious that bots cannot really read content and must rely on markers of quality to determine it. A reward orientated bot does not even care about this at all, except possibly as a marker of potential reward. Bots cannot definitely up vote quality content, but then again, can people?

I've already made my position clear elsewhere that I think quality is completely subjective, and is not necessarily the goal of curation. It's about what people want, so from a certain perspective it's an unfortunate thing that popularity, or rather getting votes whatever the reason, is what matters.

However I recognize that of course a consensus can form about quality, and likely already exists in the societies we all live in. We can like things, we can enjoy things. Bots are cold in this regard. So unless the wishes of the bot operator are very well encoded into the bots programming (which I believe can be done to a high degree) bots will be rewarding content by very different specifications to humans.

I estimate there to be majority agreement with the idea that the purpose of up voting should be to promote stuff you actually like. I like that idea too actually, I just recognise there's no way to determine this objectively, and so accusing others of wrongdoing is nonsensical if you accept we're all free to act according to our own ethics, morals and objectives.

But what if you add bots into the mix? I could argue that they are similarly free to act as people users are (they are) but there's one way that this doesn't work.

In the various blow ups we see where users are called out for bad behavior, there are people involved on all sides. There are those ranting and raving about it, those making measured post, the accused either silent or responding, either with good faith or in flames. Sometimes there is a resolution of sorts, or at least a conversation between people.

This is not possible with bots. Unless bot owners voluntarily identify themselves (or it can be otherwise proved) bots are essentially outside the social conditions of the network. I've noted before that I think shaming, for example, is a perfectly normal, natural and even more or less acceptable way for human societies to go about the messy business of organization. Bots are immune to shame or social pressure. Only their human operators are.

Bots as authors, not just readers?!

What about bots writing content? @ozchartart seemingly does this, though they are not creative posts as such. To the best of my knowledge @renzoarg is the only person who has written a bot which actually writes "creative" posts. It was created to provoke though, rather than really write posts for real (I think!), and it has not been used much by @renzoarg

I built her only for one purpose: bug and annoy people at a bitcoin related website where some rewards where given for interaction: Writing articles an commenting (yes, the original idea was not Steemit's)

What is it that's so offensive about it? I think it's the suspicion that we can be tricked by a machine, and that of course interacting with a (ro)bot is spiritually empty. That it wasn't written with any direct human thought, intention or intelligence, just the cold, mimicking and indifferent code of a machine. It is the reverse case of the engagement issue, that up voting and commenting a machine written post is engaging with exactly no-one!

The good news is that it is unlikely that a bot like this would be able to successfully pretend to be a human for long without being found out. Where does that leave this bot if it did not try to conceal it's bot nature? I'm sure it would offend some, but it might actually be interesting to others.

Are bots bad for the platform?

It is hard to define what is good or bad for the platform. Lots of people think they know, lots of it based on social models and things learned on other platforms. Very few of these definitions include bots.

What I think we can all agree on is that no one likes to be cheated. This includes breaking or even bending the rules, as well as being mislead and lied to (as in the above example). On the rules, it's clear that bots do work and can only work within the rules. However what they do can be legitimately perceived to be like card counting in blackjack. By reading the Wikipedia article (hey! it's research) I discovered the term [advantage gambling](Advantage gambling, or advantage play, refers to legal methods, in contrast to cheating in casinos, used to gain an advantage while gambling) which refers to

legal methods, in contrast to cheating in casinos, used to gain an advantage while gambling. [...] Unlike cheating, which is by definition illegal, advantage play exploits innate characteristics of a particular game to give the player an advantage relative to the house or other players. While not illegal, advantage play is often discouraged and some advantage players may be banned from certain casinos.

This is exactly what most bots attempt to do. It sounds really bad, I'll admit. But a few things keep this in check. The rate limiting of voting for example keeps all of us from using our stake without consequence. However advanced AI, predicting the whales (not currently an option!) and so on can never be ruled about as long as bots are permitted.

This cannot really be mitigated against except by widespread usage of advanced bots, reducing the relative advantage, or banning bots outright. In truth, this is part of the robopocolypse we're always hearing about, they can just do some things better. @pipokinha makes a good point in the linked post:

There are two attitudes that are not advisable in this kind of social transformation [brought on by AI and robotics]:
1- Try to stop innovation.
2- "Bury your head in the sand" and pretend that nothing is happening.

And Steemit is definitely a fertile site of innovation.

Bots are like the drunken uncle at the family gathering: tolerated by most, berated by a few, kept an eye on by all but generally seen to be a disgrace, though many of us have a nip of whiskey with him and secretly think he's cool. There's a variety of opinion on whether he should be invited next time. But even those that would, would not like all the uncles, aunts and cousins to start modelling his behavior.

Perhaps the means do matter as much or more than the ends with regards to a social network. And the means here are clearly humans on the platform. To this end we need to be self moderating and have the tools to combat abuse.

Poll: what do you think?

Early I compared bots to guns. Weapons are necessarily destructive but are robots? Not necessarily, unless your gun can make you a sandwich and drive your car (maybe someday it will be!).

Robots are generally actually a robotic "something", e.g. a robotic car, a robotic sandwich maker, a robotic gun. When we say "robot" alone I think we generally imply robotic person.

Bots are not robotic people, they are more like robotic "somethings". Since almost any tool can be robotised it strikes me that perhaps the term "bots" is too general in this discussion.

So can I ask you to comment on whether you agree with any of the following statements?

1. All bots are bad
2. Some bots are bad (please detail which, such as comment bots, up vote bots, flagging bots) but not all bots are necessarily bad
3. Not all bots are necessarily bad, i.e. I do not have a problem with bots

Finally

Whether we can tolerate bots or not is the question, because it's clear to me that automation on Steem, and thus bots, cannot be stopped by the current configuration.

I will lay out why this is the case in the next post in this series.

Thank you for reading 😊

Attribution

The robot drawing is mine and original.

The photo is Enon robot.jpg by Ms. President (Flickr User), original here. It is licensed under Creative Commons Attribution-Share Alike 2.0 Generic.