Samuel East was laid off from Plex about 60 days after starting, having already been made redundant from another company earlier in the year. He had turned down a second offer to take the job: permanent, fully remote, £95,000, after a four-stage interview process. Five minutes after a company-wide Slack announcement about eliminating 24 positions, an email told him his role was being eliminated.
His essay is personal and angry, and it's worth reading as a primary source rather than a verdict — Plex's side isn't in it, and one person's 60 days is not a statistical sample. But the structural questions he raises don't depend on the specifics.
Four rounds in, sixty days out
A four-stage interview is an expensive signal. It says: we are confident enough in this decision to spend senior people's time on it. A permanent contract says: we are planning around you. Then 60 days later the role is eliminated.
Those two facts can both be true — restructuring happens, priorities change — but they can't both be about the same thing. If the hiring process was a genuine evaluation, the firing wasn't. If the firing was a genuine performance judgement, the hiring process was negligent. East's update makes the answer fairly clear: the entire player team went, including the most senior engineers who had been holding that side of the product up for years. That's not performance management. That's a strategy decision wearing performance management's clothes.
The humane version of this is not complicated: say it's a restructuring, pay severance proportional to the disruption, and stop pretending 60 days was an evaluation.
The measurement problem
The more interesting part of the essay is what he says about the work itself. At Plex, the loop had changed: pick up a ticket, feed it to Claude, read what comes back, correct it, move on. Pull requests arriving fast, also fed to Claude, which reviews changes and sometimes invents solutions to problems nobody had.
Meanwhile he's learning a new codebase, a new architecture, new processes, and a new AI-mediated workflow, all at once.
Then he asks the question the industry should be able to answer: what are we measuring developers against now? Engineering ability, system understanding, problem-solving — or how quickly someone can produce output through a tool?
Those are different things, and a 60-day window measures the last one almost exclusively. Worse, it measures it during the period when a new hire is worst at everything: the codebase is unfamiliar, the context is missing, and the AI workflow is new on top of that. You have constructed the least favourable measurement window possible and then drawn a conclusion from it.
"Human in the loop" is a real job, badly specified
East's phrase for what he felt he was becoming — a human in the loop, managing prompts and reviewing generated code — is accurate and not an insult. It's a real role. The problem is that companies are redefining the job without redefining how they hire, evaluate, or promote for it.
If the work is now judgement over generated output, then the evaluation should be about judgement: can this person spot the wrong abstraction, the missing test, the change that will break in six months? That takes longer than 60 days to observe, and it requires reviewers who can recognize it. If the work is still engineering, then the AI is a tool and the fundamentals matter more, not less.
What you can't do is hire for one job, measure another, and fire on a third.
The onboarding inversion
There's a quieter point in the essay that deserves more attention: AI was supposed to make onboarding easier by answering questions about the codebase. In practice, it adds a layer between the new hire and the system. Instead of learning why the code is shaped the way it is, you learn what the model says about it — which is often plausible, sometimes wrong, and never the same as having a colleague explain the history.
For senior hires, that's corrosive. Their value is judgement built on experience, and the fastest way to destroy it is to put them in a loop where they never touch the actual reasoning.
What the industry owes
I don't know what Plex's strategy is, and it's not my place to guess. But the general pattern East describes — hire fast, measure output through a tool, restructure, repeat — is a choice, not a law of nature.
If a company needs fewer engineers because AI changed the economics, it can say so. That's a legitimate business decision, and people can plan around it. What they can't plan around is being told they were hired for their judgement and then let go for their throughput in eight weeks.
The essay's last line is "I honestly have no idea where we go from here." The honest answer is that the industry hasn't decided yet — and it's currently making that decision by accident, one 60-day contract at a time.
Source: Laid off 60 days in: what is the tech industry doing to developers? — Samuel East.