LinkedIn, in 2026, is the de facto standard for employment. Opting out costs more than staying, especially if your career crosses a border or your next job needs to land within a few months. So most of us keep an account, post the appropriate amount, and never quite say what we mean.

The effect of writing on the platform is that of a glass office. You can be seen at any time by people who might one day hire you, so you write as if you can be seen at any time by people who might one day hire you. The rules are not posted anywhere. They are enforced by whoever happens to be reading, which makes them invisible until you break one. The job market is the moderator the platform never has to be.

The tacit agreement

There is a quiet contract that comes with the account. In exchange for proximity to recruiters and the world’s largest searchable directory of professional humans, you agree to behave as if you are always slightly auditioning. You write updates that could plausibly be screenshotted by a future employer. You translate your private experience into the dialect of the platform.

That dialect has a name in linguistics. Jürgen Habermas drew a clean distinction between communicative action, where two people speak to reach actual understanding, and strategic action, where speech is a tool aimed at producing an effect in the listener.1 LinkedIn is structurally incapable of the first and almost entirely built on the second. A genuine loss has to be framed as “the leadership lesson I learned this quarter.” A promotion has to be “humbled” because the alternative reads as arrogant. The interface is not the problem. The incentives are.

Byung-Chul Han gave the related, sharper version: we now live inside what he calls a transparency society and an achievement society, where private interiority is smoothed flat into marketable surface and the worker is an entrepreneur of their own self, performing productivity not because anyone is forcing them to but because they no longer notice the difference between work and identity.2 The platform does not need to surveil you. You surveil yourself, in the third person, in real time, on a public timeline, and you call it personal branding.

How it got like this

LinkedIn used to be the place you uploaded a CV once every five years, accepted a connection request from a colleague you had forgotten, and closed the tab. It was boring on purpose. The boring was the point.

The market for software engineers is more precarious now than at any point in my working life. Q4 2025 was the turn, and it has not turned back. AI displacement and the broader economic weather around it are pulling the ground out from under a wider slice of professionals than before, which means the platform that mediates the hiring side of all this carries far more weight than it was ever designed to.

I am, for the moment, lucky enough to have enough client work to keep the lights on. That does not stop me from worrying about the prospect of landing a stable position as a remote worker abroad once that runs out. It would be dishonest to pretend it does.

Microsoft acquired the company in 2016 for $26.2 billion, the largest acquisition in their history at the time.3 In the decade since, LinkedIn has slowly converted from a Rolodex into a content platform with a Rolodex attached. A boring directory does not retain daily active users. A network does, especially once the network is allowed to behave like a feed. What we ended up with is the world’s largest professional directory wearing the skin of a social media platform. The directory part is still useful. The skin is where the damage is.

Welcome to the jungle

If you spend any time on the platform, you start picking up the local vocabulary. None of it is in the official help docs and all of it is real.

Broetry: writing in which.

Every sentence.

Is its own paragraph.

For dramatic effect.

The algorithm rewards posts where readers have to click “see more,” and one-sentence-per-line is the fastest way to manufacture that click.

Engagement bait: posts engineered to provoke a reaction regardless of whether the post has anything to say. The “unpopular opinion” opener followed by a wholly conventional opinion. The fake poll where one option is obviously correct. The screenshot of a DM that may or may not have been sent.

Ghost posting: jobs posted to the platform that are not actually open, either because the role is filled, was speculative, or exists to harvest applicants for a future cycle. Also, separately, posts written by ghostwriters in the voice of an executive who has not personally written a sentence in years. Both are common.

Vulnerability porn: personal trauma, mental health disclosure, or loss used as load-bearing material for a leadership lesson. The vulnerability is the hook. The generalisable insight is the product. The line between disclosure and instrumental confession is thin and almost always crossed.

The humblebrag pipeline: a small fake-rueful complaint that reveals itself as a flex and closes with gratitude. “Tough day rejecting six job offers, can’t believe how blessed I am.” The structure has effectively replaced straightforward bragging.

The dialect is so well-codified at this point that the paid search engine Kagi ships a “LinkedIn-speak translator” as one of its built-in assistant tools, the entire job of which is to convert phrases like “humbled and excited to announce” back into “I got a job.” Take a moment with that. A search engine built an opinionated piece of software whose explicit purpose is to undo the way a generation of professionals has been taught to communicate. That tool exists because the demand for it is real.

The Suggested feed

The feed is no longer a network feed in any meaningful sense. It is a recommendation surface trained to surface whatever makes people pause for two seconds: layoffs as personal branding, carousel sermons about leadership, AI-generated founder wisdom, and posts where four hundred verified executives agree just a little too enthusiastically about something a first-year MBA student could have phrased better.

The AI slop layer has matured faster than the platform’s tolerance for it. The structure is unmistakable. A confident opener (“X is outdated”). A “BUT.” A newline-heavy listicle with emoji and verbs. A close that demands you comment “YES” to receive a free PDF. The whole thing reads like an LLM doing an impression of an executive doing an impression of a thought leader. Nobody is fooled. The engagement numbers stay high anyway, because the bar for engagement is a thumb pausing, not a mind being changed. The existing browser extensions that try to clean any of this up either do not work or get patched out within a couple of releases. There is no extension that can solve a problem of incentives.

The AI layer

I have written at some length about what AI has actually done to my work. Hiring is where its weirdness lands hardest, because both sides of the table are now using the same tools against each other.

Recruiters lean on AI for the application funnel, so applicants ATS-optimise their CVs with AI to pass the filter, so the version that finally reaches a human is then sometimes shredded for being “too AI-assisted to feel sincere.” Three layers of automation stacked in a row, each one annotated by humans who blame the previous layer for the result. The candidate’s only job is to guess which round of the loop will finally have a person in it.

The interviews are where the charade becomes most visible. You are recruited by people who are not exactly recruiters, asked questions that were obviously assembled with the help of an LLM, and asked at the start of the call whether you consent to Gemini Notes being turned on for the duration. You can usually see your interviewer’s eyes track the auto-suggested follow-up before they ask it. After the call, another model is fed the transcript and asked to assess whether you are “top-notch for the role.” A system that cannot, by any honest definition, cognate now has decisive influence over real economic outcomes and over how strangers behave toward each other on video calls. It is a novel hellscape in its own right.

The job hunt

Then there is the actual job-hunting, which compounds the cultural problem with several specific failure modes.

The listings are where the product is hardest to defend. You can pay for Premium and still get sponsored results with no relationship to your filters. “Remote” does not mean remote. “Hybrid” does even less work: a hybrid role in Marquês de Pombal is a thirty-minute taxi from where I live; the same role in Oriente is a two-hour expedition each way on a public transport network whose punctuality is something the regulator publishes annual reports about.4 Both go in the same filter bucket. Tick Hybrid and you will still see sponsored ads headed “REQUIRES RELOCATION TO REMOTE VILLAGE IN ROMANIA.” For a company owned by Microsoft, that is a long way from acceptable.

Recruitment on the platform has split in two. There is real recruitment, where someone has read your profile and can explain why the conversation is worth having. Then there is what a LinkedIn post by Matthew Weeks called ambush recruitment: the verified call from someone who names a job title and asks whether you are interested before giving you anything else to go on. The pressure point is timing. They want a decision before you have had time to check the company, notice the missing salary, or discover the role is on-site two hours away and that “remote” was doing a lot of dishonest work.

There is also the ghost job problem: listings that look real but never resolve into a human conversation. The same role sits there for weeks, reposted and promoted. After a while you start overvaluing apparent activity because it gives the illusion of movement. A site with many openings feels healthier than a site with fewer. It is not necessarily true.

The hardest part of the loop is the silence afterwards. You send the application, tailor the CV, write the cover letter, and then nothing happens. No reply. No rejection. No sign that a human saw the thing you spent an evening shaping. LinkedIn makes the silence feel public because it keeps showing you the same company while refusing to tell you whether you are still in the running. The practical workaround is the same one everyone in the industry already uses: referrals. A warm introduction does not guarantee anything. It just clears away the ambient sludge. That is also why the platform is so hard to abandon. The product has rotted. The relationships on it are real.

Why nothing else has saved us

It is not as though LinkedIn lacks competitors. Indeed exists, and acquired Glassdoor in 2018.5 StepStone holds meaningful share in Europe. Welcome to the Jungle and Otta have built thoughtful UIs for the kinds of roles software engineers actually want. XING is still the dominant network in the German-speaking world. None of them have meaningfully threatened LinkedIn.

What LinkedIn has, and what nobody else has cloned at scale, is the social graph. Even when its product is worse, the directory underneath is the part everyone needs and nobody can replicate without spending another decade convincing people to upload their CVs. That is the lock-in. The obvious upgrade has to be either an entirely new graph nobody has the patience to build, or a federated alternative that does not need a single graph at all.

When someone tried

Jonathan Tensetti spent a chunk of 2026 building exactly that: a European, ActivityPub-powered job platform. Federated, open, free of the engagement-bait logic, and built in clear-eyed response to most of what is wrong with the dominant product. I was lucky enough to beta-test it. From inside the product I never saw what the public discourse around it suggested I should be seeing. The platform worked. The filters were honest, language preferences were respected, listings carried real metadata about location, remote eligibility, and employer verification.

He shut it down. The reason he gave was that the steady drip of feedback across the build had stopped feeling like criticism and started feeling like hounding. From the outside, nothing I saw in the actual product justified the volume of it. That disconnect is the part I keep coming back to. Online, dismissing a thing is cheap and shipping a thing is expensive, and the gap between the two is wide enough to bury a solo project. Whatever the real ratio of fair criticism to noise turned out to be in his case, the noise was loud enough to make a competent developer walk away from a good idea.

It is a real loss. Not because the platform was perfect on day one, but because it was the right shape, in the right place, with the right protocol underneath it, built by someone who had clearly thought through every complaint in this essay before I had finished writing it.

Rooting for the next attempt

So for now, LinkedIn it is. I keep the account. I search for jobs somewhere else first and come back, because the market is still routed through this place and pretending otherwise is a different kind of fiction. What I have never managed to do is perform on it. I tried, or rather I tried to try, and could not get myself across the line. The broetry stays unwritten. The engagement bait goes unbaited. The AI slop scrolls past unread.

I do not think this is a permanent state of affairs. The frustration is too widespread, the gap is too obvious, and the alternative is too clearly buildable for the current situation to be the end of the story. Someone will pick up where Jonathan left off, and the next attempt may catch a quieter wind. Europe in particular has every reason to ship a credible competitor. The bar is not high. A serious alternative would only have to do what the twentieth-biggest competitor to Booking.com already does as table stakes by comparison: filters that work, language preferences that are respected, verified employers, listings that are actually open, locations that are actually locations. None of this is innovative. It is just the floor.

Until that floor exists somewhere else, I will keep working behind the glass. The corridor never empties. The corridor never has to. I am genuinely hopeful, though, that this is one of those professional rituals my future self will struggle to explain to anyone younger than thirty. I would like to be wrong about how long that takes. I am rooting for whoever proves me right.

Footnotes

  1. Habermas, J. (1981). Theorie des kommunikativen Handelns. Suhrkamp. English: The Theory of Communicative Action (1984, 1987, Beacon Press, trans. T. McCarthy). The distinction between communicative and strategic action is developed across both volumes. ↩

  2. Han, B.-C. (2012). Transparenzgesellschaft. Matthes & Seitz. English: The Transparency Society (Stanford University Press, 2015). See also Han, B.-C. (2010). Müdigkeitsgesellschaft. English: The Burnout Society (Stanford University Press, 2015). The argument about auto-exploitation under the “achievement society” runs through both. ↩

  3. Microsoft. (2016). Microsoft to acquire LinkedIn. Announced 13 June 2016; the $26.2 billion deal closed in December of the same year. ↩

  4. Autoridade da Mobilidade e dos Transportes (AMT) publishes annual quality-of-service reports for CP Comboios de Portugal and Metro de Lisboa, documenting punctuality, cancellations, and infrastructure incidents. Lisbon-area commuter rail in particular has been the subject of repeated coverage over declining reliability in 2024 and 2025. ↩

  5. Recruit Holdings. (2018). Recruit Holdings to acquire Glassdoor. The $1.2 billion acquisition closed in 2018; Glassdoor’s job-search functionality has since been progressively merged into Indeed’s listings infrastructure. ↩