You can no longer watch The Pragmatic Engineer Podcast in the Spotify app. Only audio. I quit publishing video on that streaming platform.
The decision wasn't hard. After a series of reliability issues—Spotify consistently failing to process video episodes—I decided that reliability takes a back seat within that team, and across much of Spotify.
Three outages in five weeks
For the first two years, the podcast published on three platforms: the master RSS feed, YouTube, and Spotify. For eighteen months, nothing went wrong. Then, from late May, things fell apart.
The admin portal for podcast publishers ("Spotify Creators") was wonky from the start—intermittent errors, unable to remember my login, requiring a code sent to email every Wednesday to publish.
Outage #1: My episode wouldn't process for 2+ hours. The processing pipeline appeared to stop running. The Creator portal showed NaN% values everywhere. I emailed the team. They confirmed the outage and promised to do better.
"The issue was in one of our podcast publishing metadata pipelines... We identified the root cause, deployed a fix, and reprocessed the affected episodes... We're also tightening the system... Apologies again that you hit this."
Outage #2: Four weeks later, all of Spotify went down for many users when I tried to publish. Spotify doesn't maintain a status page, so it's impossible to tell how widespread the outage was. I didn't include a Spotify link in that week's announcement.
Outage #3: Five weeks later, episode publishing broke again. After waiting two hours, I sent out the announcement with no Spotify link. I emailed the team and said I was considering stopping video publishing. I asked for the incident review.
No incident review, no transparency
For the first outage, I got a vague description and promises of improvements that were never delivered. During the second outage, there was no improved communications to creators, as promised. The incident review for outage #3 never arrived, even three weeks later.
I checked my Spotify stats: stream plays had been trending downwards, while other platforms didn't show the decline. "Enough is enough."
When I switched away from Spotify, the Creators portal became buggier than ever—broken UI, missing data. A day or two later, these issues disappeared. I assume no one had tested the flow of moving away from Spotify Podcasts to an RSS feed.
A few days after offboarding, I finally received the incident report for outage #3. Something didn't add up in the timeline. My email confirmed I had alerted the team at 17:30, but the report downplayed that customers alerted them before their own automated alerts fired. I complained, and to their credit, they updated it.
But the promised improvements remained vague. One line stood out:
"During this incident, many creators learned something was wrong from their audiences before they heard anything from us. We are improving our processes and technical capabilities so creators get notified as soon as possible when things aren't working."
AI psychosis at Spotify?
The focus of Spotify's leadership is on AI, not reliability. In March, I met Spotify's Head of Technology & Platforms, Tyson Singer, who said the company puts reliability far ahead of AI adoption. So it was surprising to read this summary of a podcast Spotify did:
"Spotify now ships 4,500 production deploys a day, and 73% of PRs are now AI-assisted. Niklas Gustavsson keeps 5 to 10 Claude sessions running in tmux... Agents working in the background. All of it inside a 20M+ line monorepo."
All the talk is about AI, and none about reliability, all while Spotify's platform becomes less reliable than ever.
I've used the term "AI psychosis" before—to describe Meta's rush to develop AI models at the cost of reliability of its profitable business activities. Instagram's most embarrassing account takeover incident happened after the Trust & Safety team was slashed, and AI-generated, AI-reviewed code caused the hacking of a former US president's account.
Spotify seems to be following the same playbook.
The "suck less" principle
Max Kanat-Alexander, distinguished engineer at Capital One, wrote about how a software project can become successful just by "sucking less" with every release:
"All you have to do to succeed in software is to consistently suck less with every release... As long as you consistently suck less with every release, you will retain most of your users... But what happens if you release frequently, but instead of fixing the things in your software that suck, you just add new features that don't fix the sucking? Well, eventually the patience of the individual user is going to run out."
Personally, I got tired of Spotify's Podcasts product continually going in the wrong direction on that scale. The poor reliability, frequent errors, and the sense that they don't really care about improving things.
I don't regret the choice to leave. Video podcasts on Spotify never truly took off, so quitting wasn't a big deal. But I'm particularly disappointed that Spotify has prioritized AI usage over reliability. I know some executives there pushed against this, but I feel safe in assuming they lost that battle.