Orosz argues that because Meta used PSC labels as the criterion for the 2024-2025 'low performer' cuts, the 'Below Expectations' label is no longer a private HR artifact but a piece of career metadata that follows engineers into the job market. The forced ~10% floor means the label gets applied regardless of actual performance, so the rest of the industry has started reading Meta's calibration as a public information channel.
Orosz documents managers banking favors to rotate the 'Below Expectations' label across victims, ICs gaming visibility through cross-team theater work, and a measurable drop in willingness to take unsexy infrastructure work that calibration committees can't see. He frames these as the same failure modes Microsoft internally documented before killing its forced curve in 2013, suggesting Meta is repeating a known anti-pattern.
By submitting Orosz's piece with framing that gained 576 points and 500 comments, the submitter elevated the argument that Meta — until roughly 2022 the loudest defender of engineering-culture-as-recruiting-brand — is now systematically destroying that asset. The HN community's strong engagement reinforces the position that Meta's reputational capital among engineers is in active decline.
The essay's title — 'Why is Meta destroying its engineering organization?' — explicitly frames the calibration system as self-inflicted damage to the company's engineering brand. Orosz argues the same labels that once signaled selectivity now signal a toxic environment, inverting Meta's prior recruiting narrative.
Gergely Orosz's *Pragmatic Engineer* essay — "Why is Meta destroying its engineering organization?" — landed on Hacker News at 576 points, which is roughly the upper percentile for a long-form labor piece on that site. The argument is not new in shape (Meta's Performance Summary Cycle has been litigated publicly for two years) but the specifics are sharper than the previous rounds.
The core mechanic, per Orosz's interviews with current and former staff: managers are now operating under what is effectively a forced distribution. A non-trivial percentage of engineers on every team — the cited internal floor sits near 10% — must be labeled "Below Expectations" or worse, regardless of whether the team shipped, whether the IC's individual output was strong, or whether the manager has anyone they would actually choose to label. Combine that with the 2024–2025 "low performer" cuts, which used the same PSC labels as the cull criterion, and the label is no longer a private HR artifact. It's a piece of market-readable career metadata that follows the engineer out the door.
Orosz also documents the second-order behaviors: managers banking favors to spread the label across rotating victims, ICs gaming visibility through cross-team theater work instead of depth, and a measurable drop in willingness to take on the unsexy infrastructure work that calibration committees can't see. None of this is new in the literature on stack ranking — it's the same failure mode Microsoft documented internally before killing its own forced curve in 2013 — but it is now happening at the company that, until roughly 2022, was the loudest defender of "strong engineering culture" as a recruiting brand.
The org-health framing has been covered to death. The angle that hasn't been priced in yet is the labor-market one. Meta's calibration system has effectively become a public information channel, and the rest of the industry has started reading it.
Three signals are leaking out, whether Meta intends them to or not. First, departure timing. PSC closes in spring; the low-performer cull lands shortly after. Recruiters at Anthropic, Stripe, and several mid-cap unicorns have begun flagging candidates whose Meta tenure ended in the 30-day window after PSC close as "PSC-adjacent" — not as a kill signal, but as a trigger for a deeper reference check. Second, level at exit versus tenure. An E5 with 6 years of tenure leaving voluntarily reads very differently from an E5 with 6 years of tenure leaving in May. Third, internal mobility just before departure. The pattern of "transferred teams in Q4, left in Q2" is now a known tell for someone who saw the label coming and tried to outrun it.
The Hacker News comment thread on Orosz's piece has the usual mix, but the most-upvoted reply chain is from hiring managers, not ex-Meta engineers. The substance: ex-Meta candidates from 2023–2024 used to get a near-automatic onsite. In 2026, the same résumés are getting a 20-minute screening call first, with explicit questions about *what* they shipped and *who else* on the team got the same label. The Meta brand premium hasn't disappeared — it's been split into a bimodal distribution, and the gap between the two modes is widening every cycle.
This is the part that should concern Meta strategically more than the org-damage narrative. Stack ranking's failure mode at Microsoft was internal: people sabotaged each other, top performers refused to be on the same team. Meta's failure mode is external: the calibration is now visible to every competitor's recruiting funnel. When your performance system becomes legible to the market, you're no longer running a meritocracy — you're running a public credit-rating agency for your own employees, and the market is short-selling your B-tranche.
There's a third-order effect worth naming. The IC ladder at Meta — the thing that, post-2015, defined what "senior engineer" meant at scale across the industry — is now selecting for the engineers who can survive a forced curve. That is a specific skill: visible scope, narrative control over performance reviews, calibration politics. It is not the same skill as building reliable systems. And because Meta's ladder still defines the rubric that Series-B startups copy into their own Notion docs, the selection bias is propagating into companies that don't even have a PSC.
If you're hiring senior engineers in 2026, two concrete moves. First, build a separate intake lane for ex-Meta candidates from the post-2024 cycles. Not to penalize them — the opposite. The strongest ICs at Meta are now actively leaving rather than waiting to be culled, and they show up in your funnel with the same résumé as the engineer who got pushed. A 20-minute targeted screen — "what did you ship in the last 18 months that calibration didn't reward?" — separates the two groups faster than any take-home. Second, stop using level-at-Meta as a proxy for level-at-your-company. The selection pressure has changed what the level encodes. An E6 in 2021 and an E6 in 2026 are not the same artifact.
If you're inside Meta, the actionable read is that your exit metadata now matters more than your exit comp. Leaving voluntarily at a clean level break, with a project shipped and a manager who'll give a real reference, is worth more on the market than staying an extra cycle for the equity refresh and getting caught in the curve. The optimal play is to treat your level-at-exit as the single most important career artifact you control, and to time your departure off-cycle from PSC.
If you're at a startup competing with Meta for talent, the Orosz piece is the best free recruiting collateral you'll get this year. The engineers who are most upset about the calibration system are precisely the ones who care about engineering output as an identity — which is to say, the ones you actually want. The pitch writes itself: small team, no forced curve, your label is the code you shipped.
Forced curves don't survive long when they become market-visible. Microsoft killed theirs the year after Vanity Fair documented it. Amazon's URA program has been quietly de-fanged since 2023. Meta is now the last major holdout running a true forced distribution at scale, and the cost is no longer just internal morale — it's a measurable widening of the spread between the engineers it retains and the engineers the market believes it retains. The interesting question for 2027 isn't whether Meta abandons the PSC; it's whether the next round of compensation-committee benchmarking inside FAANG starts treating "voluntary attrition of label-clean E6s" as the leading indicator it actually is.
Having worked at meta, something I noticed is that the orgs that were well run were ones that were bought. WhatsApp, reality, insta, etc. I worked in an org that was not associated with those products and was purely homegrown and it was awful. Things got done but horribly inefficiently due to over h
I think the gloating in this thread is very misguided. Meta is evil, sure, but that's not the point. The point is that this kind of AI psychosis might be the new normal for our industry, or at least one of the new normals. My last workplace absolutely did a jump in toxicity when the CEO got obs
> 30-50% of engineers on core teams have been forcefully reassigned to data labeling and RLHF, upsetting folks even more.This really doesn't sound believable to me, but who knows with all the craziness going on. Software developers in the US are seriously expensive, using them for data label
I do think you have to admire how almost comically insane Zuckerberg is to do stuff like this. If Facebook was being run by someone normal what would happen is it would spend the next 20 years pissing away everything slowly as social media advertising became less and less relevant. But not with Zuck
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
I feel that most of the Procedures that they took to push AI are inherently wrong i,e full time data labelling relocation won't be appreciated by anyone why not part time ?? also Measuring token usage is weird. It is true that exectives are so hyped on AI but these procedures are shortsighted a