I'm using feedback on my last post to dogfood radical open review. Come play!
Why traditional startups struggle to disrupt the academic publishing industry.
At first glance, the staid academic publishing industry seems like a perfect fit for disruption in the form of an enterprising startup. Its total addressable market, or TAM, (~$19b revenue/year) is more than big enough to support a unicorn or two. It relies on centuries-old processes based in the limitations of print that have been proven to be ineffective and inequitable. It is dominated by a few large mega-corporation incumbents who, like the newspaper industry before them, have become used to extracting enormous profit-margins for activities that produce questionable value.
The market has a nice shape to it, too: it appears to be a two-sided marketplace with researchers on one side, institutions and funders on the other, and publishers in-between — a classic chicken-and-egg situation, ripe for disruption. Researchers need to adopt a new publishing platform that abandons the worst parts of the status quo. To do that, universities, labs, and funders need to give those researchers credit for those alternate modes of publishing so they don’t lose their jobs by opting in.
So, an enterprising startup should be able to succeed by raising enough money to build a slick new publishing platform and pay to subsidize researcher usage of it until their institutions are forced to recognize their contributions and pay to support the platform. Thanks to the rise of a new generation of private labs like Arcadia Science, a KFG1 member and partner, and Focused Research Organizations (FROs), the job actually appears to be less complex than ever, because a lot of the initial bootstrapping of the platform can be cross-subsidized by a new type of institution without the constraints of universities, rather than paid for directly by the startup.
In theory, I think this should work. And to be clear, I think private labs and FROs are a key part of the solution, because they can help reduce the risk of adopting new forms of publishing by proving the models outside the status quo. But in practice, what we’ve learned building KFG over the last 7 years is that the startup approach to building disruptive academic publishing technology is often doomed by a failure to understand the complexity of the market.
There are lots of different visions of what a disrupted academic publishing industry would look like, and in general, it’s not clear that “disruption,” as understood by many, is desirable. So, for the purposes of this essay, I’ll keep the goal simple and work backwards from there. I (as many before me) imagine a future where:
Anyone can publish their work to the web freely and easily, in the form that best suits the work.
Anyone can discover that work freely and easily.
Anyone can publish feedback about work they find freely and easily, in the form that best suits their needs and the author’s needs for evaluation.
Anyone can find and use this feedback for their purposes.
Money is distributed equitably and fairly to everyone involved in the publishing and evaluation process, based on the quality of the work and the evaluations.
Getting to that end-point is going to be far more challenging than in other, simpler markets, because academic publishing is a byzantine, multi-level enterprise-ish market. The customers who pay most of the bills for publishing directly (libraries) are not the same as the customers who pay for research (foundations and governments), who are not the same as the customers who pay researcher salaries (universities and labs). And none of the payers are the key stakeholders, the researchers themselves. Despite being different customers, their decisions are tightly coupled. Institutions who pay researcher salaries evaluate them based on the success of research that the funders pay for, which is measured as an output of the publishing that the libraries pay for.
Because of this complexity, the academic publishing market imposes enormous social switching costs on its customers. For an individual researcher to change their behavior, they need to know that the university or lab which pays their salary, the funder which pays their research costs, and the library which pays for access to their field's top journals won't penalize them for it — or, in the best case, that they’ll be rewarded for it. Many startups in the space build publishing products that compete on price, ease-of-use, time, etc., but none so far have been able to overcome enough of the social switching cost to win the number of customers required to build a viable high-growth business. Instead, to survive, most startups end up capitulating to the status quo, building features that their customers demand that reinforce the worst parts of the current system. Over and over again, once-radical publishing startups find themselves unable to say no, and compromise on their mission to, e.g., build direct connections to journals, sell data to analytics tools owned by their competitors, waste money supporting esoteric formats, and still find themselves without a viable high-growth business.
As a result, a long list of independent publishing tech startups (bePress, Atypon, Editorial Manager, eJournalPress, Hindawi, Publons, F1000, Authorea...) have been acquired by publishing conglomerates, which has damaged the reputation of for-profit startups among the very people they need to win over. Private labs and FROs are a great new market entrant with fewer of these constraints that can help push the ecosystem forward by demonstrating that the social costs are lower than anticipated. But there's not enough of them for startups to build a viable business on, and they still need to interface with the productive parts of the status quo (DOIs, citation graphs, metadata tags, discoverability tools, etc.) to build the communities around their science which they believe will bring about change.
To truly disrupt the market, we need a coordinated, wholesale switch in the way academic careers are evaluated by three different customers: research institutions, libraries, and funders. Doing this will require the product instincts of a startup, the specialized industry knowledge of an incumbent, the coalition-building capabilities of a foundation, and the resources to survive what's likely to be a long, iterative transition. A startup may be able to succeed in that environment, but I believe the path is much more narrow than in most markets, which is why we built KFG the way we did — to build great products that can compete on a traditional cost basis while offering services that help users understand what they really need from the status quo. We’ve learned that great technology is not enough in this market. You also need to help users overcome switching costs by providing the examples, education, guidance, and confidence that allows them to let go of their assumed needs and embrace more radical forms of publishing without worrying about not getting credit for their work. On top of this software and services offering, we’ve built a revenue model that can keep us sustainable during the transition without requiring us to chase the for-profit-level returns that would likely force us to make concessions to the status quo that would compromise our mission and deplete our hard-won credibility.
That doesn’t mean startups and others shouldn’t try as well. But if they do, they would be wise not to make the same mistakes so many others (including earlier iterations of KFG) have. Here are some of the common mistakes I’ve seen startups make in this industry (including my own!).
A core bad assumption many technologists make when entering the academic publishing market is that the market is analogous to another one they’ve seen, often the newspaper market that was so successfully upended by Craigslist, Facebook, Twitter, Google, LinkedIn, etc. The truth is that it is far more complex than most markets well-suited for startups.
The main inefficiency in the academic publishing market is indeed similar to the one that doomed the newspaper market: misalignment between the value proposition of the industry’s core products, the needs of the people who use them, and the needs of the people who pay for them. In the news market, newspapers subsidized the cost of their main subscription product with advertising. Over time, they became complacent, and began assuming that because their readers paid a subscription fee, they valued their core news product, when in fact they mostly valued ancillary services like classifieds that were easily disaggregated by digital upstarts. The true consumer value of the news product isn’t 0, but it is likely too low to support thriving newsrooms anywhere but in the country’s largest cities, which is why most local news is transitioning to nonprofit models.
In the academic market, the core product produced by publishers is peer-reviewed research. It is subsidized primarily by university libraries, who pay enormous subscription fees so that their faculty can access them — or, alternately, pay enormous article publishing charges so their faculty can publish work open access. This is where the similarities between the two markets end.
Unlike in news, academic publishing conglomerates have not been complacent. Learning in part from the news industry, they realized that there is little intrinsic value in providing access to peer-reviewed content, which is why they have been more than happy to gradually acquire every promising startup and switch to models based on publishing charges, services, and analytics while seeing record revenues.
What they realize, and what many startups do not, is that the value of publishing lies in the prestige and opportunity for career advancement an author earns by being published in high-profile venues. Changing the status quo will require researchers to change their behavior in ways that risk their livelihoods. To break through, we must change the way researchers are evaluated and compensated for their work so that they have an incentive to publish in new ways.
Shifting compensation also sounds like a classic problem for a startup, but here, too, startups often fail to understand the complexity of the market. In different markets, like hotels or transportation, startups like Airbnb and Uber can quickly change once-hardened behavior at scale by creating alternative markets that provide superior experiences to existing ones. This works because those markets are fundamentally consumer-driven: individuals can vote with their wallets.
The academic market, as I discussed above, is a multi-level enterprise market. Institutions and funders pay the salaries and budgets of researchers, and largely demand that their researchers publish in prestige journals to earn tenure, promotion, and future grants. Because of that demand, researchers in turn demand that their libraries give them access to those prestige journals, so they can learn how to be successful at publishing in them. Worse still, the same companies that publish prestige journals also own and sell the analytics platforms universities and funders use to evaluate research success, and those companies deliberately limit what they will include in those metrics to publishing outputs that look like the status quo.
Thus, even if a startup developed the perfect open publishing platform, and every researcher wanted to publish on it (and many would), they would risk their careers doing so until their institutions and funders decided to change both the method and tools they use for evaluation. Particularly in light of the pandemic, momentum is shifting towards recognizing more contributions like preprints that don’t look like traditional journals. And new types of institutions are pursuing a strategy of delinking compensation from publishing activity to great effect. But even these new institutions have limits to their progressiveness. To get quality feedback from researchers at traditional institutions, and to have their output recognized as legitimate science, even the most progressive labs must make their work legible to the status quo by minting DOIs, adding journal-like structures for aggregators, etc. In a best-case scenario, a progressive lab working with a mission-driven technology builder (again, this is what we are attempting at KFG) could provide just enough ties to the status quo to help outside researchers come along for the ride.
As discussed above, however, traditional high-growth startups need to build features quickly to win customers in the 12-18 month timelines set by their investors. The vast majority of training and research still happens at universities, and the vast majority of funding still comes from foundations and governments that demand adherence to the status quo. As a result, the features built to give progressive users just enough of a link to current system have a tendency to snowball into selling points for more conservative customers that unwittingly reinforce the status quo, compromise the company’s mission, and hurt its reputation among its core users.
Startups who understand this reality are faced with a narrow set of choices. Some, like Authorea, Sci.pe, Libero, etc., choose to work within the system by creating better technology to support traditional journal or preprint publishing, in the hopes that if they can attract enough adopters, particularly from established journals, they can eventually use their platform’s scale and network effects to nudge people towards more progressive modes of publishing.
The problem with this approach is it targets publishers as the customer, rather than the place most of the revenue comes from: libraries. Top-tier journals are published on technology owned and operated by their publishers. They have no incentive to switch to better technology they don’t own — at best, they may (and often do) acquire those startups and adopt them internally. Independent publishers like societies and university presses have very limited budgets, and, with some notable exceptions, are often fairly conservative, only reluctantly embracing even open access, much less new publishing formats. New, progressive independent publishers who don’t want to publish with the large conglomerates often do choose to publish on new platforms. But they have even smaller budgets and quickly find that a better technological experience alone is not enough of an incentive to attract authors to publish with them instead of prestige outlets. Worse still, the long history of independent publishing platforms being acquired by large conglomerates discussed above has effectively poisoned the well for new startups, causing most independent editors to value mission alignment above technology and further reducing the capacity of the industry to innovate.
University libraries are incentivized to switch their subscription budgets away from rent-seeking conglomerates, but they ultimately answer to faculty. As several high-profile failures by large libraries to renegotiate their contracts to send less money to publishing conglomerates have shown, until those faculty are comfortable with both not having access to prestige journals and not being able to publish in them, libraries can only shift very small amounts of budget to new upstarts. Librarians are also among the most progressive actors in the space, and feel the sting of startup acquisition more acutely than most.
Taken altogether, this means that until the social switching cost problem is overcome and libraries can devote more of their budgets to paying for novel publishing experiments, the actual TAM of the academic publishing market is likely a tiny fraction of the oft-quoted $19B figure. In resource-scarce environments, even the best startups will struggle to stay afloat on the short timeframes they’re typically funded to prove traction within.
Other startups, like Publons, Pubpeer, ScienceBlogs, and ResearchHub, attempt to more squarely address the social cost problem by eschewing the journal system altogether and inventing simple, highly structured, alternative forms of publishing that can be recognized and compensated for differently. No dominant platform exists to publish work outside of the current system, they argue, so if they can build a simple system with a great UI and novel compensation and recognition schemes, they can convert enough users to build traction until they became a de facto platform for academic publishing that incumbents can’t ignore, much like Facebook or Twitter in social media.
The argument has merit: most academic publishing technology is truly horrendous to use. But there are three problems with this approach: first, it incorrectly assumes that ease-of-use is the key barrier to adoption, when, as we’ve demonstrated above, it is just one of many. If ease-of-use were the main problem, blogs and social networks would have long solved it. Instead, platforms like ScienceBlogs fizzled out, and even successful, purpose-built social networks like ResearchGate and Academia.edu have been unable to change the publishing behavior of the masses, despite numerous attempts.
Second, the approach creates a very narrow success path. Because tech startups are driven by lean principles, they tend to over-learn from early success and underestimate the heterogeneity of the publishing ecosystem. A platform that works for one sub-discipline is often inadequate for an adjacent one — sometimes just because the cultures are different. But these disciplines are so small that even winning 100% market-share of one doesn’t look like meaningful traction to a startup’s funders. Thus, startups are forced to either spread their focus too thin (deadly for any startup), focus on a large discipline that gives them early traction but can’t be easily replicated in other spaces, or produce technology that is too generic to be a meaningfully better alternative than the status quo, despite UX improvements.
Finally, the approach underestimates the technical complexity of the ecosystem, which is itself designed to impede disruption. As discussed above, even the most progressive research producers will need some connections to legacy systems for their work to be considered at all. Publishing a single minimum viable research output requires, at minimum, knowledge of JATS XML, the DOI system, multiple metadata tagging schemes, and direct relationships with dozens of aggregators and indexers. Learning all of these esoteric requirements and developing those relationships takes years for even the smartest team.
It may be possible for the right team with a background in the ecosystem to strike the perfect balance with the right series of pivots, but it’s a much narrower path than exists in most markets.
With the caveat that I’m incredibly biased by my employer, I believe the solution, as I wrote above, lies in creating a new type of knowledge institution that combines the best of startups, non-profits, and expert consultants. These institutions must be capable of producing innovative, trusted technology that allows anyone to experiment with new approaches to publishing. But that’s not enough. They must also be able to reduce switching costs by combining those tools with services that help users maintain the right ties to the status quo that give them the credit and credibility in the current system without reinforcing the worst parts of it. And they must find a way to become sustainable on their own merits so that they’re not reliant on grants, of which there simply aren’t enough to support technology organizations over the long term. KFG won’t be the only one of these institutions, and our approach won’t be the only one that works (if it does). But until we acknowledge the complexities of this market, and the challenges new entrants face, we’re going to see a lot of startups come up empty in their attempts to disrupt anything except their own bank accounts.