PACR vs LinkedIn: A Network Built for Researchers

Last March I was sitting in a borrowed office at a conference, half-listening to a keynote on the screen in the corner, when a postdoc I'd met twenty minutes earlier asked me a question I've been thinking about ever since.

She'd just finished a project on inflammatory markers in a pediatric cohort. Good data, odd result, and she needed someone who actually understood the assay to tell her whether the odd result was biology or batch effect. "I've been searching LinkedIn for two days," she said. "I keep finding people who list immunology as a skill. I have no idea if any of them have ever touched a plate."

That's the whole problem in one sentence.

What LinkedIn is genuinely good at

I want to be fair here, because I think a lot of academic commentary on LinkedIn is lazy and slightly snobbish.

LinkedIn is very good at a specific thing: it tells the world you exist, in a format that people outside your field can read. If you're a PhD student thinking seriously about industry, LinkedIn is where recruiters live. If you're a clinician who wants to be visible to health systems, device companies, or policy organizations, it works. If you're spinning out a company, it's where your investors and your first three hires will find you. It handles the "I am a professional, here is my trajectory" job well, and it handles it at enormous scale.

I've told students to set up a LinkedIn profile and I'll keep doing it. A colleague of mine got a consulting contract that paid for a year of her child's daycare because a program officer found her there. That's real.

But notice what all of those use cases have in common. They're about being found by people who aren't researchers.

The hard part of an academic career isn't being found by recruiters. It's being found by the four people on earth who understand your method well enough to collaborate, challenge, or review you.

Why the generic model breaks for academics

General professional networks are built on a shared assumption: your career is a sequence of jobs, and your capability is a list of self-declared skills. That assumption is fine for most of the workforce. It's a terrible fit for research.

Think about how you actually evaluate a stranger in your field. Somebody emails you asking to collaborate. What do you do?

You don't look at their job titles. You go find their papers. You look at where they sit in the author list, because third author on a big consortium paper means something completely different from last author on a tight methods piece. You skim their methods sections. You check whether their preprints turned into publications. You look at whether their work cites the people you'd expect, and whether anyone you trust cites them back. If they're clinical, you might check whether their trial is registered and whether the registered outcome matches the published one.

That whole evaluation takes maybe eight minutes, and not one step of it touches a skills list.

A second-year PhD student emailed me last spring with a profile she'd drafted. Under skills she'd written "data analysis, scientific writing, teamwork." I asked her what she'd actually spent the previous eighteen months doing. Turns out she'd built a reproducible pipeline for harmonizing three cohorts with incompatible phenotyping, and she'd written the documentation so well that two other labs had adopted it. None of that fit in a skills tag. It was invisible exactly to the people who would have cared most.

That's the structural failure. Generic networks flatten the things that distinguish researchers from each other, and amplify the things that don't.

The endorsement problem

Peer endorsement on a general network is social, not evidential. Your office mate endorses you for machine learning. You endorse them back. Nobody involved has evaluated anything.

Research has its own, much older endorsement system: citation, coauthorship, review, and replication. It's imperfect and it has well-documented biases, but it's grounded in work that somebody actually did. A network that ignores that system in favor of mutual back-patting is throwing away the only signal that matters.

The expertise-versus-interest problem

On a general network, "interested in single-cell transcriptomics" and "has run single-cell transcriptomics experiments for six years" look nearly identical. In a research-first network they shouldn't, because the evidence is right there in the publication record.

I once watched a lab lose four months because they recruited a methods collaborator based on a profile and a friendly call. He'd written about the method. He'd never implemented it. By the time that became obvious, the grant deadline had moved.

What a research-first network does differently

The design principle behind PACR's research workspace is simple: your profile should be a byproduct of your work, not a separate marketing document you maintain on the side.

Nobody updates their profile. Let's be honest about this. You update it when you're on the job market, and then you forget it for three years. Meanwhile your actual research identity is changing constantly, paper by paper, preprint by preprint.

So the useful move is to stop treating the profile as something you write and start treating it as something your output produces. If PACR is where you're searching literature, reading, drafting, and working with coauthors, then your expertise profile can be drawn from what you've published and what you're working on, rather than from adjectives you chose about yourself at 11pm before a deadline.

A few things change when you build it that way.

Publications are the unit, not jobs

Your record is the record. Authorship position carries information. A methods paper and a commentary are different objects. A preprint that became a peer-reviewed article tells a story about how your thinking developed.

Keep your ORCID current regardless of what platform you use, by the way. It's the one identifier that will follow you across institutions, countries, and career changes, and it costs you about ten minutes a year.

Expertise is inferred from work, not declared

If your last nine papers involve mixed-effects models on longitudinal clinical data, that's your expertise, whether or not you ever typed the phrase into a box. A system that reads your corpus can surface that, and it can surface it at the right granularity. Not "statistics." Not even "biostatistics." Something closer to the actual thing you do.

This matters most for the people who are worst at self-promotion, which in my experience correlates alarmingly well with being good at research.

Finding collaborators starts from the literature, not from a search box

Here's the workflow I actually want, and it's the one that connected tools make possible.

I'm reading a paper. Something in the supplementary methods is exactly the problem I've been stuck on for a month. On a general network, the next step is: open a new tab, type a name, hope they have a profile, hope it's current, hope there's some way to tell whether the person who built that pipeline is the first author or the senior author or the technician who isn't on the paper at all.

Inside a connected workspace, the next step is: the paper is already in front of me, the authors are already identified, their other work is already linked, and reaching out doesn't mean leaving the context I'm in. I can see what else they've done, whether it overlaps with my question, and what I'd actually be asking them for.

When you're searching and exploring the literature in the same place you write and collaborate, the gap between "this person's work is relevant" and "I've contacted this person" collapses from a week of procrastination into a single session.

The part most career advice gets wrong

Standard advice says: build your brand, post consistently, grow your network, be visible.

I think that advice is mostly aimed at the wrong target for academics, and I've watched it waste a lot of early-career time.

The researchers I've seen build durable, genuinely valuable networks didn't do it by posting. They did it by being useful in specific, traceable ways. They reviewed carefully and the editors remembered. They shared code that worked and documented it properly. They answered a methods question from a stranger in detail, and three years later that stranger was writing a grant and needed a co-investigator.

Every one of those is an act of work, not an act of marketing. A research network should be built to capture and surface exactly that kind of contribution.

Which is also why I'm skeptical of follower counts as a proxy for anything. I know researchers with enormous followings who struggle to find a reviewer for their own paper, and I know quiet methodologists with no social presence at all who get three collaboration requests a month because their name is on the right four papers.

Reach isn't the metric. Relevance is.

A realistic scenario

Say you're a clinical researcher planning a trial of a behavioral intervention in heart failure patients. You need three things: a sense of what's already been tried, a statistician who's handled stepped-wedge designs in this population, and eventually a writing group.

Working the generic way, that's four systems. Literature search in one place. Notes in a document. Collaborator hunting on a professional network where nobody lists "stepped-wedge." Writing in a shared drive where the version history is a crime scene.

Working inside a single workspace, it's one thread. You run the literature search on stepped-wedge trials in heart failure, and the summaries you generate stay attached to the papers they came from. The authors who keep appearing on the design-methods papers are visible as people, not just names in a reference list. When you start drafting the protocol, the evidence you gathered is already there, so you're not re-finding the paper you read on Tuesday. And when you register the study on ClinicalTrials.gov and start thinking about reporting, your CONSORT-aligned structure isn't a separate exercise, because the EQUATOR Network guidance and your draft live in the same workspace.

No step loses the context of the step before it. That's the actual argument, and it's an argument about workflow more than about networking.

The PACR workspace is built around that continuity, and the network sits inside it rather than beside it. Your collaborators aren't contacts in a separate address book. They're the people you're working with on the thing you're working on.

What I'd actually tell you to do

Keep LinkedIn. Spend twenty minutes a year on it. It's your front door for the non-academic world, and sooner or later you'll want that door to exist.

But stop expecting it to do the research-specific job, because it was never designed to and it isn't going to start.

For the work itself — finding the people who understand your method, keeping your record of contribution current without maintaining it by hand, and getting from a half-formed question to a draft without losing what you learned along the way — you want a system that knows what a paper is, what authorship position means, and what a methods section contains.

The postdoc at that conference, by the way, eventually found her assay person. Not through a network. Through a friend of a friend who remembered a talk from 2019. It worked, but it took five weeks, and it shouldn't have taken five.

If you've been running your literature, your writing and your collaborations in four disconnected places, try consolidating one project — just one — into a single workspace and see what it feels like when nothing gets dropped between steps. You can set up a PACR workspace and start with whatever you're already stuck on, or sign back in if you've got one sitting idle.

The people you most need to find are already in the literature. The question is just how many tabs it takes to reach them.