not-available-for-comment:

eclipsedarksoul:

not-available-for-comment:

softwaring:

lmfao this is so funny 😭

OH MY GOD. HE STOPPED GETTING IMPRESSIONS BECAUSE HE PAYWALLED THE API

AHAHAHAHAHA

I’d assumed his followers were mostly bots because his feed is deeply boring, but I’d never put 2 and 2 together before.

YOU KILLED YOUR BOT FOLLOWERS IN COLD BLOOD MY DUDE

HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA*deep breath*HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA

Could you explain this to me so i can appreciate how funny this truly is?

Yeah, I’ve never written a Twitter bot before but let me give it a shot based on my understanding of what happened.

APIs are kind of complicated to explain to someone who’s never coded for web before, so I’m just going to explain the bits that are relevant to this issue.

Probably the most important thing to know is that websites operate on two levels: the visual interface, which is how humans interact with a website and involves buttons and pretty colors and nice fonts and fights over how many px wide a border should be. And then there’s a more computery level where, eg, the text you typed into the interface actually gets packaged up and saved to a database somewhere and then retrieved and displayed.

There are a lot of different ways to make this background computer-doing-computer-things part of the website. For simple websites that won’t be updated much you can get away with a very simple version that’s just built from a couple of relatively simple scripts and html documents. However, once you’ve built a website that’s complex and going to be updated frequently, particularly by users, an API helps manage those actions and make internal updates and expansions easier.

An example: Let’s say leaving a ❤️ on a tweet actually requires 3 actions in the background: you have to tell the database(s) that store twitter information 1) the user id of the person who is leaving the ❤️ and the user id of the person who is receiving it 2) you have to check that the person leaving the ❤️ is not blocked by the person receiving it (they have permission to leave ❤️s on that account) and then 3) save the actual ❤️ to the Twitter database. Rather than making the user do those three actions on their own, the API has a “leave a ❤️” option that you can just send two usernames to and it does all those steps for you.

Companies can decide how much of their API is open to the public. I’ve never had to set up this but of a website on my own, so I’m not as clear on how this works (maybe someone else can weigh in). But there are ways of making an API’s actions only available to the website’s devs and the user interface. Or you can chose a subset of the API’s actions and allow them to be used by anyone who knows how to write a script that can interact with them. That free-to-use part of the API is called an “open API”.

Up until very recently, Twitter had an open API. It had a lot of different uses. Some people used it to run anti harassment tools. Others used it to scrape data. And you could write Twitter bots to use it. I followed a bunch of bots that tweeted images of art by various artists. (Mikhail Vrubel and Remedios Varo were two favorites I discovered through those bots.) And of course it was also how fake followers and harassment bot armies of various descriptions also used Twitter.

But then

On February 2, Elon announced that the Open API, which up until then had been free, would abruptly become a paid service, effective February 9. Basically everyone who used the open API was incensed: One week is not really enough time to decide if you want to (or have) the budget to suddenly start paying for something you weren’t paying for before. A lot of people who wrote just-for-fun bots had to suddenly let them die. Some peoples’ security tools were rendered useless. Elon was banking on people paying to continue using the service anyway, but it was pretty clear that what was actually going to happen was everyone would use Twitter even less than they had before.

And as we all remember, the rollout of this pay-to-play option went super well and Twitter definitely did not accidentally revoke its own API access when it first implemented the paywall. 😆

Three days later…

The Super Bowl 🏈

As we see above, Elon’s followers are MOSTLY bots. Who knows what kind. Maybe some are are purchased, some may be part of various far right bot armies, etc. How many of those bot owners decided to pay to keep the bots going? How many of them even still have control of them after an utterly disastrous pay-to-play rollout?

Of course Elon got surprisingly low engagement on a Super Bowl tweet!

He killed 70% of his followers 3 days before.

Ok so as a guy who actually regularly works with the twitter API (I’ve got multiple bots that use it, plus a fediverse to twitter crossposter, plus an app that keeps track of my main twitter account)

The original developer API never went down!

First they extended the date from the 9th to the 13th, and then extended it again to “Real Soon Now”:

Other than how shitty the website has been lately, my hourly ancient bot has gone right on posting its thing the entire time. The crossposter works fine. I haven’t actually had to fix anything (not that i would if it broke)

And to top it all off? The fake user bots don’t even USE the third party API. They apparently just fake the the twitter phone clients!

All this being said:

I absolutely CANNOT WAIT until Elon kills the real API and anything that uses it goes away and twitter is stuck in fake bot hell