A few wrong claims sent us down a trail we never expected to follow. What we found was bigger than one piece of software. Bad information can gain weight through repetition, shape AI answers and become surprisingly persistent. So we started looking at what can actually be done about it.
At some point, an organization can become so busy organizing itself that the organization starts to look like the main project. I spent part of this weekend trying to work out whether TYPO3 is getting uncomfortably close to that point.
This is the final part of a series based on my WeAreDevelopers talk A True Story About Speeding Up the Wrong Things. The first looked at speed without steerability; the second at software engineering once code became cheap. But the architectural problem does not stop at code. Models now call tools and change other systems as well. So the question is broader: what has to exist around AI before its output or actions become state we rely on?
This article grew out of my WeAreDevelopers talk “A True Story About Speeding Up the Wrong Things.” The talk eventually turned into three articles. The first looked at an old pattern: systems become faster long before they become better at steering themselves. This second part stays with software engineering and looks at what happens when AI suddenly makes one of its most visible outputs much cheaper.
The video of my WeAreDevelopers talk "A True Story About Speeding Up the Wrong Things" runs for about 40 minutes and turned out to contain more than one article. So I split the argument into 3 parts. This first one is about the oldest mistake in the story: Confusing acceleration with progress.
Over the past few months, one thing about AI has become clear to me: output was the easy part. Texts, images, concepts, and summaries can now be produced at a speed that would have seemed absurd just a few years ago. What used to take days now takes minutes, sometimes seconds, and the result is often surprisingly good. That is exactly why output is no longer the most interesting question.
At some point I learned that, as an individual, there are certain things I can reliably fuck up. Not an elegant sentence, but a very useful realization.
Governance sounds about as sexy as a three-hour agenda item with a procedural motion attached. Until it is missing. Or until it technically exists somewhere, but nobody can say where decisions are actually being made.
Some mistakes live on a screen. Others eventually take up physical space in the room. In analogue production, that was not a metaphor.
Leaflet balloons. They were real. Today, we would probably file them somewhere under “content delivery.” Back then, it was a little more hands-on: you attached paper to a balloon and hoped the wind, the timer, and the target area would cooperate.
I only meant to update my LinkedIn résumé. It sounds harmless enough, but apparently it is one of those tasks where you start sorting through a few career stops and suddenly realize that, somewhere along the way, you have ended up on the table yourself.
© Copyright 2026 Cybercraft GmbH
Coders.Care uses cookies, to provide you with the best possible service. More info Besides the obligatory session cookies we use cookies for the following providers:
