LinkedIn Is Doing What Bluesky Was Supposed to Do

Rebuilding a public square on the platform you least expect

For a brief moment about a year ago, it really did look like Bluesky might work. Researchers and left-of-center intellectuals were flooding in, swapping starter packs, reassembling what felt like a nostalgic reunion of old Twitter. Then everyone arrived, and the center could not hold. It turns out that people can disagree even when they are all on the left, and that without strong social norms of free speech and civil conduct, humans—including university professors with allegedly high IQ—remain tribal and quick to pile on. Meanwhile, LinkedIn (yes, LinkedIn!) has quietly been doing the job Bluesky was supposed to do.¹

Like most researchers and public intellectuals I know, I had a LinkedIn account for years that I barely used: I would accept the occasional connection request and otherwise ignore the platform. At some point I noticed that people were actually talking on it, often substantively, in a way I had previously expected only from old Twitter. So, for the last year or so, I've been cross-posting essentially the same content to Twitter, Bluesky, and LinkedIn. At this point, the pattern is consistent enough to feel like an A/B test. Pieces that read as pro-immigration get cheers on Bluesky and silence on X; pieces that read as "anti-immigration" get the reverse. Pieces that carry the most nuance get nothing on either platform. But LinkedIn has been the exception.

It is not hard to specify what functional public-facing research discourse should look like in theory. You post something new and informative that you care about, like a working paper on a policy-relevant topic. People who find it interesting or useful say so and add nuance. People who disagree explain why in a respectful, substantive way. People who find it uninteresting ignore it. Nobody calls you names, impugns your motives, or turns a technical dispute into an accusation unrelated to your argument. If they do, they are in the minority, and they get called out.

These principles once described Academic Twitter. On Bluesky, they have mostly collapsed. Bluesky has fundamentally failed as a venue for public-facing research and, more importantly, as the channel through which serious research reaches policymakers, journalists, and the broader public. The platform that has quietly taken over that function is LinkedIn. If a lot of sane people left Bluesky or X for LinkedIn tomorrow, or at least started reposting their content there, the research internet would be better for it, and so would the rest of the internet.²

What the public square was actually for

Public-facing research had a simple social media function for most of the last decade. It was a cheap pipeline from research to policy, to journalism, and to the reading public. Old Twitter did this job unevenly but at scale through the famous posting-to-policy pipeline, as well put by Daniel Golliher. It was where a congressional staffer bumped into a political scientist’s thread and turned it into a briefing, where you learned about a working paper before it was published, where a fight over whether an RCT had identified the right thing played out in real time for an audience that partly understood and partly just absorbed the norms of how serious researchers talk.

When Elon Musk bought Twitter and rebranded it as X, the pipeline broke: the algorithm got more chaotic, external links got suppressed, spam and reply-guy slop proliferated, and blue-check monetization created a new economy of rage-farming accounts. The new regime did loosen some of Old Twitter’s speech restrictions around heterodox positions on public health, gender, and race, and for a minority of researchers on those topics, post-Musk X is genuinely freer than its predecessor. But the net effect was negative on balance, and a large plurality of social scientists, think-tank writers, and left-of-center journalists migrated to Bluesky over 2024-2025.

Why Bluesky failed

The short version is that most of what people used to blame on Twitter’s algorithm turned out to be problems with users and norms. Bluesky removed the engagement-maximizing amplification but, through aggressive self-selection during the Musk-era migration, kept a left-of-center user base disproportionately invested in policing speech and unwilling to treat ideological deviation as anything other than a moral failure. The predictable result is that any post touching a politicized topic (migration, AI, racial disparities, or whatever gets politicized next) draws volume rather than argument. For most researchers most of the time, the rational response is to say less rather than more, and the deterrent scales faster than the discourse.

This is not just something right-wingers complain about. As Max Read argues from the center-left perspective, Bluesky’s discursive norms are almost designed to repel outsiders. Nate Silver also argues that Bluesky functions less as a political movement than as a tribal affiliation with a narrow demographic profile; and Noah Smith observes that progressive commentators on Bluesky, having lost their mainstream audience, now spend much of their energy trying to cancel each other.

I have often experienced this all firsthand: academics respond to even my less controversial Bluesky posts by email, because they don’t want to risk getting piled on for saying anything in public. The numbers bear out this diagnosis: Bluesky peaked at around 40 million registered users by late 2025. Pew’s 2025 survey finds just 4 percent of U.S. adults have ever tried Bluesky.

Why LinkedIn

LinkedIn is not a perfect platform, and I will get to the downsides in a moment, but the case for it rests on a handful of things that have quietly become decisive for anyone doing public-facing research.

Scale and distribution. LinkedIn reports 1.2 billion registered users globally, roughly 30 times Bluesky’s total user base and orders of magnitude higher than its daily active count. A reasonable skeptic asks why an account number should translate into actual reach, since having a profile is not the same as reading a feed. Two things answer that. First, LinkedIn’s algorithm explicitly pushes posts with early engagement outward to your 2nd- and 3rd-degree connections, to followers of relevant topics and hashtags, and to professionals in the same industry, regardless of whether they follow you directly. A strong LinkedIn post travels to strangers in a way a Bluesky post simply cannot.

That said, LinkedIn engagement is notoriously hard to measure externally, because LinkedIn closed its public API in 2015 and restricts analytics data to approved Marketing Developer Platform partners, which is one reason the migration of research-adjacent discourse to LinkedIn has been largely invisible to researchers who expect to measure platforms the way they measured Twitter.

Audience mix. Think of LinkedIn as a kind of Switzerland of the internet: higher-status people from different camps show up under their real names and engage with each other because the professional cost of behaving badly is real. This is also the point most researchers have not fully absorbed, as I can attest, and it explains the apparent paradox of LinkedIn being the dominant platform for research-adjacent discourse without most researchers noticing. Bluesky is full of other researchers, the journalists who cover them, and activists.

The people on LinkedIn are the people we should be trying to reach: policymakers, congressional staffers, civil servants, industry analysts, foundation program officers, and journalists at general-interest outlets. A 2025 Avoq survey of DC policy insiders found that 81 percent of Democrats, 84 percent of Republicans, and 78 percent of MAGA-aligned respondents use LinkedIn. Good representative data on LinkedIn compared to other platforms is notoriously hard to find, but this looks like a bipartisan footprint no other platform comes close to matching. Researchers have not clocked the shift because the people reading their LinkedIn posts are not the people they spend time with online; they are the people their work is supposed to reach.

Format favors substance. LinkedIn’s format (longer posts, real names tied to real careers, a less snarky default register) does a lot of the work of civilizing discourse without needing heavy moderation, because when the poster is visibly accountable to an employer and a professional reputation, the median comment tone shifts correspondingly, and bad-faith quote-dunking becomes rarer. There are also no anonymous accounts and almost no sub-tweeting; the median post reads more like a memo than a hot take.

People can still disagree or criticize you heavily if you post something provocative, but they are much less likely to do it in a mindless or righteous way. In some ways, LinkedIn feels like an academic conference: people are civil, sometimes too nice, and not always willing to criticize a colleague openly. That conference-style politeness can smooth over real disagreement, but it is a much better failure mode than ad hominems and pile-ons.

Discussion that actually moves understanding. The clearest evidence I have for all of this is my own cross-posting experience. I have often shared the same piece, including the more controversial ones, simultaneously on Bluesky, X, and LinkedIn, and the pattern has been remarkably consistent. On Bluesky, the reaction is usually either silence or a small pile-on when the piece challenges prevailing consensus, and substantive engagement is rare. On X, responses are a mix of real engagement and the usual ratio of slop, bad-faith screenshotting, and reply guys.

On LinkedIn, the pushback I get is both the most civil and the most productive: named professionals who actually work on the topic, often from perspectives I don’t share, who write multi-paragraph responses that engage with the argument rather than perform outrage about it. This holds even for pieces and takes I expected to trigger the most hostility, because people disagreeing under their own name with their employer looking over their shoulder have strong incentives to be reasonable.

No, LinkedIn is still not perfect

A pragmatic case for LinkedIn has to be honest about what the platform does poorly. I’ve seen anonymous accounts on X complain that LinkedIn has been quietly suppressing links to right-coded publications for some time. I haven’t been able to verify this, but if you write from the political right, or many of your sources are outlets that LinkedIn’s filters treat as low-quality, your reach may be cut in ways that are not transparent. Even if this is true, for a left-leaning researcher writing about migration or AI, this is a non-issue, while for a right-of-center one writing about the same topics, Bluesky’s ideological environment is more hostile, and LinkedIn’s algorithm is not neutral either.

The interface is also genuinely clunky. The composer is awkward, threading is mediocre, search is bad, and basic features that X and Bluesky get right out of the box are either missing or buried. This is a fair complaint, but it is also a complaint that gets answered as the platform’s serious-discussion userbase grows: the more prominent intellectuals show up expecting a real public square, the more pressure LinkedIn faces to build the tools they need (I’m looking at you, Glen Weyl).

And yes, LinkedIn has its cringe: the AI-generated inspirational posts, the three-emoji bullet lists, the humblebrags about how humbling it was to be invited to speak somewhere, the AI-generated summaries of papers the poster obviously has not read. None of this is enjoyable, but it is easy to filter past, and the cost of ignoring the slop is low compared to the cost of staying on a platform that actively punishes substantive engagement.³

Come build the bridge

Every sane researcher who migrates to LinkedIn takes one more useful voice with them. The tipping threshold is probably not high: a few hundred visible researchers moving their primary public-facing work to LinkedIn, as many have already been doing quietly, would shift the center of gravity enough to make the migration self-sustaining.

If you are skeptical, you don’t have to abandon Bluesky (or X for that matter) tomorrow. Just start cross-posting whatever you’d normally write there onto LinkedIn too for a couple of months, and watch the difference in who shows up in your replies. The on-ramp is low-friction: post a short summary or excerpt of your latest piece with the link, tag a few people whose work it actually engages with, and see what comes back over a week or two. Based on two years of doing exactly this, my bet is that the LinkedIn version will draw substantive engagement from people whose opinions actually matter to your work, while the Bluesky version fades into silence or a small pile-on when you’re even slightly challenging the prevailing consensus. You won’t need another essay to convince you.

I am on LinkedIn here. Follow me, post your research and writing, and let the audience decide. The public square gets built wherever serious people choose to show up. Right now, that place is LinkedIn.

1. Most of what I'll say about LinkedIn here applies to Substack Notes too, but writing a Substack post about how great Substack is would be both vain and ineffective. So this piece is on LinkedIn. For readers unfamiliar, Substack has built-in social-media features (Notes, restacks, cross-posting) that work similarly to microblogging. They are underrated for writers going against the grain of conventional wisdom on the left or the right; for everyone else looking for actual reach, LinkedIn is simply bigger.

2. I should be open that this piece is partly motivated by my own experience. I try not to be driven by emotion, but the best time to push for a change is exactly when the case on the merits lines up with a reason to care. Something happened to me on Bluesky recently that suggests other researchers share these frustrations. These people have stayed on the platform because they can still have productive conversations in a narrow technical corner, but they are probably underestimating how much the platform has already closed.

3. The AI slop complaint is also overrated. As Stefan Schubert notes, people who say their feed is overrun with it should probably revise their feeds.