Kill the Bots

“We'd all become unpeople undoing unthings untogether”
— The Doctor, The Mutants

“Wouldn't know the difference between a real blonde and a fake”
— Bob Dylan, Highlands

A picture I used for one of my Instagram bot honeypot posts

For probably the majority of its user base, Instagram is a quick and fun way to share photos of what you’re doing with your friends.  Furthermore you can follow celebrities and artists you admire to see what they’re up to.

For musicians, as well as other artists and businesses, it’s a way to engage with your audience and help new people discover what you do.  This can be done by simply putting out information about upcoming releases and shows.   You can also share the day to day aspects of your life or other facets and interests you have to build up a rapport with your fanbase.

But whatever this platform is in its ideal state, it’s no secret that it’s being co-opted, gamed, and abused.

The plain semantics of a “like” on Instagram should mean that someone saw your photo or video and enjoyed it enough to take the level of effort it requires to press a little heart icon and thereby let you know they think what you’ve shared with the world is something worth at least that level of appreciation.  It’s a nice little dopamine hit for you; a nice little reward for your effort; and it acts as a reinforcement for your use of the platform.

But that’s not necessarily what a “like” actually means in practice.  If it’s not a user that you're familiar with, it’s less likely that they’re someone who happened to stumble across your photo and wanted to let a stranger know they enjoyed a post than it is someone or something that is seeking your attention.  So now it’s not “good job” it’s “hey, look at me!” — a total inversion of the intention.  And since this is a digital platform it’s not even necessarily a human — in fact it’s more likely than not to be a bot acting on a human’s behalf.  In all likelihood no human actually ever laid eyes on your photo to produce that supposed interaction.

The primary way the bots discover content to “like” is through the hashtags you add to your post.  If you ever want to see this in action, simply create a bunch of related tags (30 is the max — beyond that Instagram will make your post blank) and use them with a picture of literally anything (it could be television static and you’ll still get likes — I literally did this).  There are even tools that will help you create  such a collections of tags.  Once posted, you’ll be pretty much instantaneously swarmed by bots giving you likes.  In fact, you can get more “likes” than people your post supposedly reached — “reach” being measure of how many people have seen your post in their feed one way or another.  Short of discrepancies in how Instagram’s backend updates these counts, logically this sort of mismatch doesn’t seem like it should ever happen (you couldn’t like something that hadn’t reached you).

I did a spate of such posts recently (in the name of science!) and they became the frontrunners for “engagement” compared to all my other posts for the past 2 years — engagement being a catch all for likes and comments. After playing around a bit by creating posts with tags of different subjects (fashion, food, photography) and I’ve concluded musicians — particularly singer-songwriters — are the some of worst perpetrators of this sort of nonsense.  It’s frustrating because some people could be legitimately discovering your content, but to know who is doing that as opposed to who’s actually only interested advertising to you requires more time and effort than its worth.

I mean, I’m pretty sure the random chiropractor’s office that’s liking my post is not on the up and up.  But a random songwriter? — well, honestly, chances are, they’re not either — and how exactly you’d determine otherwise I’m not exactly sure.  I assume the idea is to target non-musicians who are posting about music they enjoy in hopes of attracting them to the ostensible “liker’s” own brand, but as a musician who’s just starting to build audience, this means the majority of accounts “liking" my posts are probably more focused on building their own followers than they are in any content I’m creating.  Maybe this just becomes all so much noise as you grow your number of followers, but I think it's much more insidious than simply being a nuisance.

And the bots don’t necessarily simply ”like” the post.  Some will add some supportive but generic and inane comment that at first glance might seem relevant but is really meant to be applicable in an all purpose way.  Some bots even have the gall to directly call for you to look at their own work.  Others will actually start liking a handful of your other posts as if they had just discovered you and started looking at your feed.   Still others will follow you — only of course to quickly stop following you even if you naively start following them in return.  In my experiments, even after the bots swarm a heavily tagged post, my overall number of followers pretty much remains constant over time.  This is an insidious way to build up a follower base without having to do any actual engagement oneself.

All of this activity will no doubt sound familiar to anyone who’s spent much time on Instagram.  I find it to be disingenuous and manipulative.  Furthermore, I would also contend that all of these machinations hurt the value proposition of the platform and I’m surprised that Instagram even allows it to happen.  The fact is though, the functionality that allows this is part of their web API which means that far from making it difficult to do, Instagram has actually enabled it.

Now, according to Instagram’s terms of service, automated activity along these lines would appear to be a violation.  According the "General Terms" of the API usage, client applications are not to “confuse, deceive, defraud, mislead or harass anyone.”  As I said, it’s inherently deceptive to say you “liked” something when you never laid eyes on it according to what I think would be the generally accepted idea of what “liking” should mean.  To post a comment on it that suggests a person looked at a photo when no person did is obviously misleading.  So I would say the powers that be at Instagram are a bit lax about enforcing this rule.

The good news is apparently the API functionality to post “likes” automatically is being removed on December 11th of this year (though this doesn’t quite jive with the list of features that were supposedly deprecated on April 11th).  Why this was ever made API in the first place astounds me to be quite honest.   This is an open invitation to automate “likes” by a client application.  It feels like the developers came up with general purpose API that exposed all the platform’s capabilities but who ever was minding the store as far as the overall system's functionality didn’t think through the implications as to how it could be used in practice and what that would mean to the end user experience.  And given the level of activity one sees from bots, whatever measures they might have in place to govern bot activity are clearly insufficient.  Aside from making the numbers of likes and followers suspect, it also casts a pall on a business that wants money to promote posts and sows doubt in the authenticity of activity generated by such promotions.

It remains to be seen whether or not this change in the API leads to a serious decline in the activity.  People may wake up on December 12th and see their apparent engagement plummet.   Other platforms, (i.e. Twitter), have had issues with fake accounts, and some users have seen follower counts dwindle when such accounts are culled.

We do live in a day and age where we have be wary of whether news is “fake” or not — is it legitimate content or is it being shown to us because an entity that is hostile to our interests wants to sow dissent — or is someone is simply "trolling" us?  In the larger political arena, making it more difficult for voters who are likely to oppose your positions is common place, not to mention diminishing their representation through gerrymandering.  And the consequences of being deceptive or telling falsehoods seem to have fallen by the way side in general.  So it shouldn’t be a surprise that we see this “bending the rules” elsewhere.

I’m cynical in general about people’s motivations and willingness to adhere to rules in the online world.  It doesn’t even surprise me that some segment of the creative community online would behave in an unethical way for their advantage.  Not so long ago I was in an internet poll that to my mind was clearly being manipulated (if you look at my Instagram history, it’ll be apparent what I’m referring to) and given its nature it was quite an easy thing to do.  I discouraged anyone voting for me from doing so, but some folks didn’t seem to regard there to be anything wrong with not playing along with what should be the general understanding of what a vote would mean in that context.   IP Addresses don’t listen to music, so they shouldn’t be the ones voting and having more than on IP Address shouldn’t entitle you to vote multiple times.

In the context of promoting one’s self as independent artist, one might see these tactics used on Instagram and elsewhere as simply being scrappy and resourceful — doing what it takes to get ahead in a dog-eat-dog world is a tried and true tradition and a little violation of the spirit of a platform’s intentions is minor.  But I’ll stick to my guns and say that using bots is a perversion of the system and makes Instagram less useful to its users — both artists and casual users alike.  And in the realm of the singer-songwriter where authenticity is good currency, using these tactics or others undermines any claim you might have to that authenticity.

I hope other artists and creatives would consider this and how ethical anything they are doing to build their fanbase actually is.

Kill the bots.

— E.G.

---

E.G. Phillips is a singer-songwriter based out of San Francisco.  You can follow him on instagram as @duckswithpants — this also works on the Facebook and the Twitter machine.