Can Twitter Save Science?
Two Truths and a Take, Season 2 Episode 4
Before I found my way to the tech world, I was a grad student in the neuroscience department at McGill University. I never took the opportunity to get my PhD, and left science to do a startup instead. But I still think about it sometimes.
Basic science is important. The life sciences especially, which are the part of academia that I know a bit, are going to make or break some of the big problems we face right now over the long term. Academic research, being an old and established industry, has evolved some structural problems that are hard to fully grasp, let alone solve.
Some of these problems are fascinating examples of positional scarcity, which emerges out of abundance and ends up strangling the ecosystem it grew out of. Over the long run, I think some of these problems are fixable, and as a matter of fact, the hero we need is already here: it’s Twitter.
This week, here are three intertwined stories about positional scarcity from this strange world:
How an “indentured scientist” class called postdoctoral fellows became the workhorses of life science research,
Why scientific journals are one of the greatest positional scarcity business models I’ve ever seen,
Why Twitter is so great for science, and could lead to real disruption.
Postdocs: the Workhorses of Life Science
For most of the history of academia, the relationship between students and masters was pretty straightforward. First you trained as a student, in some sort of apprenticeship arrangement with one or more senior professors. Once you earned your PhD, which was a difficult and usually rate-limiting step, you went off and found a faculty position somewhere. It was normal to expect some attrition through this process: some candidates drop out as we move through the funnel. But there was a clear start, a clear path, and a clear end.
In the mid 20th century, which is pretty recently on the academic time scale, governments began to allocate more money into academic research and basic science. University departments grew larger, and administrative work became more of a burden on faculty members. As scientific work became increasingly funded by research grants rather than university budgets, grant writing became a meaningful drain on scientists’ time.
Grant writing and administration aren't really something you want young scientists to be spending time on during the sharpest years of their career. So a new kind of “stopover” position emerged, in between graduate school and full-time faculty, which solved this problem: the postdoc.
Postdoctoral Fellows, or “postdocs” for short, were a win-win solution for this problem. Postdocs were a new option for freshly minted PhDs: rather than go straight for a faculty position, with all of the teaching and admin it required, you could instead go take up residency in someone else’s lab for a couple years, and just only focus on doing science. Everybody wins here: you get a platform to do science, without any distractions, that comes turnkey ready for you. The Principal Investigator (PI) who hosts you gets to put their grant money to work, via your inexpensive, high-quality labor, for a couple years.
It worked. Over the last fifty years, postdocs have become the draft horses of life science research. They’ve also become the entry point for foreign scientists to enter the US system. They have more experience and autonomy than PhD students, so they can move much faster, and generate more output. If you look at the scientific work being done in academic labs today, the postdocs are directly responsible for a lot of it.
Some fifty thousand postdocs currently work in US university research labs, usually making $50,000 a year or less after a decade of higher education. It’s become a logjam, as more and more postdocs compete for a fixed number of faculty openings. The “plight of the postdoc” has become a talking point, as their career prospects and collective mental health look increasingly bad. How did this happen?
One underlying problem here is that the postdoc position, even if it makes a lot of sense for any one person, broke the feedback loop between supply and demand for academic scientists. When you finish your PhD, it’s hard to find a faculty position - but you can always find work as a postdoc. So you go do that, and during your postdoc years you work really hard to publish papers and develop a research portfolio that helps you stand out from the crowd.
Postdocs became the elastic capacity of the academic hiring funnel. The number of faculty openings remain relatively fixed, even as we keep producing more PhD students. But there’s no real limit on how many postdoc positions there can be. (So long as you can fund them, they’re great ROI!) This means that new PhD students will always find a job somewhere. It’s like airplanes circling in a holding pattern above an airport, waiting for landing slots to open up. There’s always room in the Elastic Middle.
In a world of scientific scarcity, when there weren’t that many scientists and there weren’t that many other postdocs, those few years could be great ones. But in a world of scientific abundance, where there’s a flood of new graduates and foreign scientists stuffing the funnel every year, the postdoc becomes hell. You are the elastic capacity of the system, and you are in a zero-sum competition to get out of it. But you have no choice: you have to go through this system, even though everyone agrees it’s horrible.
This is more important than just some sob story. If you care about scientific progress, well, these are the people who are doing the science. Our life science research is largely carried out by these stressed out 30 year olds, with young families and no money and horrible career prospects. They can’t really bypass this positional scarcity trap though, because of another issue: journals.
Traditionally in academia, publishing was just a task that academic societies did out of necessity. Journals served a basic but useful purpose, which was disseminating information to scientists quickly and efficiently, at a time when it was otherwise hard to do that. They were a passive vehicle to convey discoveries and discussions that were worthy of sharing. The research society had power and influence over the field, but the journal itself? Not that interesting.
The journal did one important job, which was peer review: acting as a coordinator and judge for independent scientists to review, endorse or reject your work on the basis of its scientific merit. Peer review is an old tradition, and it lives on in more or less that form today. And it’s important! Science is complex and opaque; it’s hard to establish any kind of absolute truth about anything, but peer review at least gives us a protocol for how to do our best.
Mid 20th century, as national governments began to ramp up their research spend, suddenly we had this new problem: more science was getting done than could fit in each field’s journal. So obviously we needed more journals. Publishers were happy to create new ones, which kept pace with the escalating complexity and specificity of life science research during that time. Now, instead of there being one or two journals you had to follow in your field, there were five, then somehow fifteen.
Journal publishers saw two opportunities here. First, they understood that subscriptions were pretty much mandatory for universities. Professors demanded access in order to keep up with their field, so the university libraries had no choice but to subscribe, and take whatever price was offered. The buyer (the PIs) and the payer (the library) were two different entities, so universities had no effective way to negotiate against journals without angering their faculty. So they didn’t.
Second, they understood that in a world with a lot of journals, the relative rank between the journals suddenly mattered a lot. Journals no longer had a mandate to publish whatever was submitted to them. A hierarchy emerged, where the top tier of journals like Cell, Nature and Science could be selective with what they published. A pecking order beneath them fought for this metric called impact factor, which is still how we rank journals today.
The publishers were kingmakers, and they knew it. They’d captured a double feat of positional scarcity: all of the research mattered as far as what scientists needed to keep up with in their field (or else fall behind their peers), so everybody needed to subscribe to everything. But the top journals mattered more in terms of career advancement and grant funding, so everyone had to fight for - and recognize - that ranking order.
The outcome looks like something we’ve seen earlier: the scientific research funnel, which used to flow directly from “research gets done, and some of it’s worthy for publication, so it gets published”, now has this intermediate step with unlimited elastic capacity. Research gets done, and then it all gets published somewhere, so there’s no negative feedback to limit the amount of research that makes it into journals. But only some of it really matters, and it’s a zero-sum contest for everybody to break out of the Elastic Middle and get published in Cell.
The academic journal business model is a funny one, because the journals themselves don’t actually do much work. The content is produced by PIs, for free, who apply for publication in hope of getting selected. Other PIs who review and curate submissions also work for free: it’s considered a part of academic duty, and prestigious to accept but disastrous to decline.
In short, aside from the cost of ink and postage, academic journals deal in one thing only: positional scarcity. They sell their brand to researchers on both sides of the transaction, and collect money and work from both sides of the transaction. (Oh, and they also sell ads, too.) It’s a business model unlike any other I’ve ever seen.
In the past few decades, the academic publishing market was rolled up by giants like Elsevier, Springer and Wiley/Blackwell, who now collectively own the majority of life science journals. The consolidation gave them an even better business model: selling subscription access as one giant bundle, which are pretty much obligatory purchases for any university.
The internet didn’t change much: libraries could not cancel their subscriptions on pretext that their faculty members could download the articles illegally, after all. The brief wave of Open Access Journals, which promised free article access to readers, was a lot less disruptive than advertised. Journals simply demanded payments from the authors, rather than the readers. All this did was shift which university budget the fees came out of, and the open access revolution’s first iteration stalled after a few years.
From symbiont to parasite: the evolution of for-profit scientific publishing | Peter Walter & Dyche Mullins,Molecular Biology of the Cell
In recent years, it’s become mainstream acceptable for scientists to complain about the publishing business model. The University of California library system took a brave stance by cancelling their Elsevier subscription, and US and European policymakers are advancing a few different initiatives aimed to weaken publishers’ grip on science. But all of them focus on pricing or access specifics around the science itself, rather than what I think is the real underlying moat.
The real moat isn’t talked about as much, because it’s much more of a taboo topic: the interdependence between publishing and career advancement. Journal access is one thing, and I expect we’ll keep making progress on making scientific findings more available. But they do something more entrenched, and in my opinion worse, for science: they’ve become the universally recognized yardstick for evaluating scientific accomplishment and building personal brand.
The Brand Tax
Let’s go back to our poor postdoc for a minute. Our postdoc has now put in 8 to 10 years of undergrad and graduate effort, and hasn’t yet earned a real professional salary despite being highly skilled and in her late 20s. By the time she’s finished her PhD, she has so much sunk cost and so much career reputation at stake that she can't quit now. At this point, all of her friends and peers are other grad students and other researchers, so leaving academia isn't an option.
Her job for the next two years is actually pretty straightforward: build a brand. Building a brand is how you escape the positional scarcity of the Elastic Middle, and get hired for real. So how do you do that? Well, you want to establish a few things: that you do good science, that you’re smart, that you have a point of view and a vision for the field, and that people should read your work and listen to you.
Unfortunately, you can’t communicate any of those things directly, because no one knows who you are. So you have to communicate them by proxy, and the way you do that is by getting published in Cell, Nature or Science. And the way you get into those good journals is to go work as a postdoc in a lab that can get you there.
Of course, everyone knows at some basic level that what journal published you doesn’t really determine the quality of your research, or your potential as a scientist. But we accept it as proxy. Science is tricky and opaque. As much as we try our best to identify undiscovered research and scientists, at the end of the day we mostly fall back on the journal brand names, and the labs who can get into them, as yardsticks for quality.
This is why our postdoc has no choice but to surrender the sharpest years of her career to a lab that’s not hers, which has no loyalty to her, and barely pays her. The only way to build a brand in science is to pay a tax to those who already have it. This isn’t just a trivial hoop to jump through; it is the dominant concern of the people who are actually doing all of the science.
The entire business model of academic research is leveraged on this constraint. It's why postdocs get paid so little, and therefore, why postdocs have ended up as the industry workhorses. The sheer amount of output that they have to produce inflates how much research gets published, which in turn makes getting into those top tier journals even harder.
That’s why the positional scarcity held by elite journals and established PIs is so difficult to displace. The PIs may complain about the journals’ business models, but they’re deeply aligned when it comes to what the PIs actually care about, which is preserving the proxy reputation system that keeps them in charge. Senior researchers spend hardly any time doing science anymore; pretty much their entire job is writing grants, reviewing papers, and working the system. Their priorities are pretty evident.
The real shame in academic publishing, if you ask me, isn’t Elsevier’s 35% profit margin on journal subscriptions. It’s the much larger amount of money, time and influence that is regressively taxed from the young scientists, to the old ones, in exchange for nothing but brand access. So long as journal access remains the yardstick that matters, then no matter what legislation gets passed or conventions get tweaked, I doubt that the overall structure of the ecosystem will change that much. It’s bad for science, and by extension, bad for all of us.
That being said, there is something that could actually be disruptive to this setup: Twitter.
Why Twitter Matters
I was in grad school just before academic Twitter became a thing, and if we’re being honest, if it had come just a few years earlier my entire career may have turned out differently. Twitter matters, because for the first time, young PhD students and postdocs have a way to build their brand directly, and eventually acquire peer review directly.
If you are a young student, you can go hang out in senior scientists' mentions on Twitter and be a reply guy, and if you’re smart and thoughtful and have insightful things to say, you’ll get noticed! Twitter is especially empowering for a particular kind of person who’s unique and might stand out weirdly, and in normal settings that uniqueness would be a problem - but on the internet, it’s an incredible asset.
Science is full of those kinds of people. They’re the ones who do the best science! But the journal / postdoc / brand building regressive tax hits those people especially hard, because it’s a game they probably don’t play very well. The structure of academic science today is leveraged on there being no way around that positional scarcity. But Twitter is a way around.
The first real application of Twitter in science was pretty straightforward: sharing what you’ve published in an effort to boost citations and other conventional metrics of success. Academics being academics, of course, there’s been a pretty funny (but understandable) push to try to standardize and codify Tweets as an “alternate method of scientific legitimacy”, basically trying to recreate citation rankings and impact factor but with faves and RTs:
Introducing the Twitter Impact Factor: An objective measure of urology’s academic input on Twitter | Diana Cardona-Grau et al.
T factor: a method for measuring impact on Twitter | Lutz Bornmann, Robin Haunschild
The thing is, part of the reason why Twitter works so well is because it’s not as codified and explicit as a journal ranking system. It’s informal, and that informality makes it a lot more real. People can get to know each other so much more easily now, increasing the likelihood that smart scientists who ought to find each other actually do. You can just go participate directly in real-time scientific discourse and relationship building, skipping everything in the middle. A piece in Nature last yearalmost got it right:
Perhaps the most obvious, and most important, aspect of Twitter is that the platform facilitates a closer, more informal connection between scientists. It can be difficult to see the true nature and personality of authors through the mountains of academic papers they produce. Getting a more human perspective on the big shots we look up to can be refreshing; we can learn about both their science and their wider views, hobbies and the like. By having a more personal line of communication with each other, rather than relying on e-mail correspondence, scientists can connect and form fruitful relationships more easily.
I say almost got it right because it doesn’t quite hit what’s most disruptive about this new setup: it’s not that the established PIs who already have brand recognition can be seen as more approachable; it’s so that students and postdocs who don’t yet have brand power can use Twitter to acquire it. This is sneakily dangerous! It’s a way around the Elastic Middle. And when the value proposition of the Elastic Middle starts to erode, all of that positional scarcity that’s been built up over decades may soon have less value.
The first crack in the system, which is already happening, is young researchers and labs sharing their preprint results and publications directly on Twitter - initially after peer review and publication, with stuff that’s already in journals. I bet you soon we’ll see people share stuff that hasn’t been peer reviewed, either to get there first (if you’re racing for a discovery) or, quite simply, because Twitter is peer review.
Look, for instance, at this Twitter thread of scientists going through Coronavirus research in real time. This kind of discussion used to only be able to take place in private email threads between elite scientists, or in slow, back-and-forth editorials in journals, also between elite scientists. Now anyone can participate! There is no barrier to putting yourself out there and contributing.
And if you had your doubts as to whether the scientific integrity of these discussions is less good than in a supervised, formally peer-reviewed environment, it looks more like the opposite is true: it’s like real-time Wikipedia, for research that’s happening now. This is what real change looks like, and it’s worth connecting the dots between what’s happening on Twitter and the incumbent positional scarcity that’s slowly strangled academic science. The cure may be already here. Just give it a few years.
Permalink to this post is here: Can Twitter Save Science? | alexdanco.com
Have a great week, and keep our eye out for a bonus newsletter episode: a guest post all about debt from Ali Hamed of CoVenture, right after this.