Before you go, check out these stories!

Hackernoon logoI’ve just solved Twitter’s fake news and bad bot problem overnight. by@pastaduck

I’ve just solved Twitter’s fake news and bad bot problem overnight.

Author profile picture

@pastaduckPaul Anthony

No. not really. Sorry. I’ve just finished reading this. research shows that Twitter’s platform is crawling with bots, bots that have potentially influenced the U.S. election. In its critique of Twitter on the matter, it’s not alone.

Now here’s the thing. This in itself is not news. Twitter has, and always has - been crawling with bots.

For us technical folks that understand how these things work, they have this magic thing called an API which has given Joe Schmo developers like me the ability to create bots in the first place. This is what makes Twitter’s ecosystem what it is today. It’s also provides some of the secret quirky sauce that makes it appealing to me as a user versus their arch nemesis. Bots are an integral part of Twitter’s appeal.

Delightful bots, that tell us when someone in U.S. congress edits a Wikipedia page or Parliament Edits for us Limey’s across the water. There’s GIFs from Wikipedia, Dear Assistant or any number of articles on the web extolling their virtues. Oh, did I mention, there’s even a bot for finding bots. Meta.

When I read that pretty much everyone is pointing the finger at Twitter for this whole Russia-rigged-the-presidential-election thing, I can do nothing but feel sorry for them. The problem isn’t bots per se; instead, it’s when bots go rogue, impersonate human interaction and trick the masses. Is that Twitter’s fault? With machine learning and NLP technology becoming more and more advanced that problem is going to get more difficult, not less.

Twitter for all their faults, are caught between the proverbial rock and a hard space. On one hand, they’ve tried to be more than just A fight they don’t want to go head to head with Facebook on. Focus instead on the strengths of the platform and the things that make it unique. With the return of Jack Dorsey to CEO who promised to try and bring more developers back into the fold you have something of a tale of two wives. The bots go hand in hand with the whole opening up an API thing and being developer friendly. On the other, they are being hauled across the coals for not doing more to prevent bad bots. If they turned off API access tomorrow, the same journos that are currently beating them over the head for the whole election debacle would be taking them to task for no longer being able to see their favourite instagram pics turned into paintings. Ho hum.

Blocking the bad bots.

Almost ever article I read on this subject cracks me up. ‘Block the bad bots’ we hear them cry! ‘Just block all the Russian bots’, or ‘Block bots from IP range’. Usually followed up with ‘The smartest minds in the business work at Twitter, they musn’t care’ or similar rhetoric. Au contraire mon ami — this only highlights the complexity of the problem.

Just blocking bots is one of those things that sounds like the right answer. The answer readers want to hear. Good old common sense - a simple solution to a complicated problem. Now I don’t work at Twitter, but I’m pretty fucking sure for every article that is written that suggest to ‘Block the bad bots’ or ‘Block bots from Russian IP addresses’ there’s some poor engineer at Twitter headquarters punching a pinata of a non technical Journo that doesn’t understand exactly what it takes to achieve that goal.

Sorry, my pedigree chum. This isn’t a simple problem that can be solved overnight, and every time you write about you are showing a distinct lack of understanding of the problem.

Even I, who doesn’t have half a clue on Twitter’s current internal infrastructure knows the following:

  1. I don’t know what the real technical challenges look like, so me suggesting a solution is idiotic.
  2. Blocking IP addresses is fruitless. Lots of the bots don’t originate in Russian IP space to begin with, secondly. Proxies and botnets around the web make IP solutions something of a digital wack a mole.
  3. Detecting bots is easy. Determining good bots from bad bots that are already an engrained part of the Twitter ecosystem is practical impossibility.
  4. You are asking someone to perform a turing test across millions of users and get it right 100% of the time and not throw the baby out with the bathwater.

So, if you are planning on writing about this topic, please, stop. Talk to someone on the Twitter team. Hell, find some sort of knowledgable source that will at least half explain to you what the challenges are before you start. This is not your average tech challenge please stop writing as if it is.


The Noonification banner

Subscribe to get your daily round-up of top tech stories!