According to TechCrunch, Twitter API is coming into a probable cause of glitches being experienced by users of third-party apps. Users of Tweetbot, Echofon and Twitterrific, three popular third-party Twitter apps, are facing several issues, including being unable to log in and access Twitter feeds. Apparently, the only way to access Twitter is through the official client or the website.
What’s the story?
Naturally, many speculated that recent events at Twitter, including massive layoffs that impacted the operations and site reliability teams, could have played a role. TechCrunch points out that Elon Musk has already killed several developer programs. He also, according to the company’s former head of developer platforms, Amir Shevat, broke the trust of developers.
All of this makes the sudden disruption of third-party apps, all at once, highly suspicious. Is it a technical problem or a nefarious plot? After all, pre-Musk Twitter had form on breaking third-party applications.
Action or Inaction?
Perhaps the Twitter API itself can provide some clues. At APImetrics, we continuously observe hundreds of important public APIs on behalf of companies and app developers that depend on them. You can see the performance and quality data for these APIs through our API.expert portal, as well as via our alerting service, Serinus.
Ironically, Serinus currently posts API performance and outage alerts to Twitter! You can follow @SerinusMonitor to get alerts from dozens of commonly consumed APIs, including crypto exchanges, banks, fintechs, and many of the top sixty API-first companies, as defined by the GGV Capital API First index.
So, what did API Expert spot? First, hats off to Twitter’s DevOps and SRE teams. API.expert’s continuous observation of the API through the period (indeed going back a full five years) reveals nothing that points to a cause-and-effect scenario between these events. There was correlation, but not causation. You can see the results for the Twitter API on the API Expert Social Networks portal.
The Social Network API portal on API Expert reveals that over the past year of continuous monitoring, the Twitter API performed fifth, behind four other widely consumed APIs.
With a pass rate of 99.97% and p99 latency of 949 ms, the Twitter API was able to achieve a Cloud Quality score of 9.30. This puts it on par with Facebook over the course of the last year.
(The API Quality score, “CASC” is APImetrics’ ‘secret sauce’ for providing apples-to-apples quality comparisons across multiple APIs. If you’re interested in learning more, please read our overview, Everything You Need to Know About the APImetrics CASC Score.)
That said, there have been obvious issues with the API – looking at the update status endpoint, it has shown an overall trend down over the last few months – although interestingly, as you can see from the data, it has improved quite significantly since the events of Friday Jan 13.
All this evidence supports the statement by the TwitterDev account, which posted last month that the company “will continue to invest in our Developer Platform, especially our Twitter API.”
Except, to be honest, if you believe that then I have a gif to sell you…
(secretly?) Planned Action
There is a circle of API hell reserved for those who suddenly and without notice cut off access to an API as widely used and popular as Twitter. This is not the first time that Twitter has done this to their APIs, one of the first major APIpocalypses was when Twitter shut off the authentication for the first generation of APIs.
Clearly this is a planned action. It is focused on popular apps that have high traffic and appears to have been focused on keys used for mobile applications rather than desktop apps. The rumors suggesting this is more to do with forcing traffic onto the Twitter application and enabling better ad sales plus driving traffic to their own algorithms rather than allowing users to pick their own timeline.
It’s poor practice for any API provider and continues a trend of large providers enabling ecosystems and then taking them away when they feel the API access is detracting from the core service.
(Not looking at you, Facebook, for instance. Oh, wait, actually we are.)
One extra thing to consider, as the graph shows, one of the critical APIs for running a third-party app – update status, really improved dramatically, but the timeline app is showing increasingly erratic performance.
Praise where praise is due
That said, we must applaud the technical teams at Twitter. Throughout the turmoil and chaos over the past six months at the company, they have managed to deliver a robust and dependable API, at least for those Elon Musk deems worthy of consuming it!
Given what has been publicly said about the challenges inside the company, this is indeed quite the achievement.
Matt Reinhold pretty much describes our position although we will add – the problem with broken trust seems to be that people have short memories when something is useful – as he points out, and as we point out, Twitter have form.