Welcome to Science with Shrike! This week we talk about the potential future for academic publishing, what existing decentralization means for the rest of the economy, and what decentralization might look like for science in the future. Last week, we covered the first big shift to decentralization, Open Access, and states’ attempts to hijack and control it. This week, we look at other potential solutions in the publishing marketplace, which may leave publishers out of a business.
Right now, academic publishing is already more decentralized than most industries. Publishers and journals compete for authors, and the marketplace of journals is huge (~4000 journals in the top 4 tiers alone, plus thousands more Journals of Nobody Cares). On the other hand, the top tier publishers all have a reasonably strong reputation and prestige moat that keeps them on top. Aside from initiatives like Plan S(hame), authors currently have the choice of journals to submit. The days of being unable to publish are gone. The only question becomes the same one asked during limbo: “How low will you go?”
With all that said, tenure and promotion committees keep score, as do grant review committees, and at the top universities, you need to have a regular stream of top publications. If you already publish in big name journals, it’s easier to continue publishing in big name journals. If you’re not in the club, it’s a lot harder to break in. This is why there are a limited number of people to work for as a postdoc if your goal is academia—you want to work with people who have the political power to get your work into the top journal, or at least get your work *reviewed* by a top journal. If your PI gets you connected when you are a postdoc, and you have excellent stories later, you will be able to keep publishing in top journals. You’ll also get the grants and the talks, and careers increase exponentially.
However, if you’re not one of those top 1% of academics, life is much harder. This aspect gets bandied about and discussed, which leads scientific societies and some journals to try new approaches. Eventually the big publishers have to try something too, just to keep up, and if it’s popular, you have progress.
This is how the Open Access movement started. PLoS was one of the first to launch, and their success (driven in part by getting buy-in from well-known PIs), got others thinking about the process. PLoS One, with it’s ‘accept anything technically correct approach’, also forced big journals to respond. Scientific Reports was Springer-Nature’s response to PLoS One. Like other industries, the big players have to buy, ape in and/or compete with the new ideas, which is part of how the more popular ideas become mainstream. Academics are always full of ideas, so there’s a never-ending supply of new things to try. Some catch on, others not so much.
Journal-independent review is the next step in decentralizing publishing even further. The idea, embodied by initiatives like Review Commons (LINK), is that the review process can be disconnected from the journals themselves. Once a paper gets reviews, journals would then select the papers to publish. This idea is attractive for playing in high profile journals, because priority is often the barrier to acceptance (or even being sent out for review). Journal-independent review ensures your paper gets reviewed, and the journals can then decide which papers they want. Some publishers are implementing this within their own family of journals. For example, Cell Press has what they call “Community Review” (LINK). The pros are that your paper is more likely to get reviewed (avoiding the gatekeepers at the top journals), and review at Cell Press becomes one round instead of several. The cons are that it is easier for the editors to hand it off to a lower journal, and that you may end up with a lower overall impact publication. You might have a lower impact publication because by staying within the journal family, you miss out on other journals in the higher tiers.
Overall, journal-independent review is a net positive for peer reviewers, because it reduces the amount of time spent in peer review. Instead of waiting on review at 3 or 4 journals, it’s all done at once. The downside is that one or two lethal reviewers might sink the work for that entire family of journals. It will also likely change the author-publisher-reviewer dynamic, and new opportunities to get your papers published will arise. Specifically, it will put more of a sales angle to your paper, since journals need to be sufficiently interested to opt for the paper. Journals will be able to shop around for papers, and probably will attempt to use some metric to judge engagement/interest prior to acceptance. It’s unclear how this will work, and an analytics company aimed at improving journal impact by selecting “winning” pre-prints (i.e. will get cited a lot) has the potential to end up a huge winner. Many publishers will likely develop something along these lines themselves, so if you can create a convincing metric, or scrape existing ones, you might be able to do B2B with them.
Another fad that mostly failed to catch on is “double-blind” peer review. Typically peer reviewers are not identified to the author, at least prior to acceptance/rejection. Note that some journals are moving to permit reviewers revealing their identities, and Frontiers has an open reviewer policy AFTER publication. “Double-blind” peer review is the opposite, where the reviewers officially do not see the author list. Aside from making it hard to catch potential conflicts of interest, it also doesn’t work well in practice. The methods section will almost inevitably give the research group away, and if you are building on previous papers, you might have to repeat half of the controls in those papers to satisfy the reviewers.
To give you an idea of where “biomedical science” currently is, crypto is not even on the horizon for “reimagining biomedical science”. A recent “Reimagine Biomedical Research” essay contest (LINK) had no crypto solutions in the top seven entries, though three focused on diversity. One of the winners proposed decentralizing grant recipients, which was promising. However, if the NIH Small Business grant awards are any indication, it will be very poorly implemented.
Instead, we’ll explore how the crypto space can transform publishing and how publishing can be further decentralized.
One key roadblock in publishing is peer review. On one hand, peer reviewers need to be vetted, and certified. On the other hand, peer reviewers would ideally be incentivized to review papers. Journal-independent review is a step in the right direction, but it still doesn’t fully solve either problem, and it also has the problem of soliciting qualified reviewers. It’s also challenging for reviewers to get credit
Credentialing non-fungible tokens (NFTs) may be one solution. Reviewer expertise could be built into an NFT that is used with relevant credential information. Scientific societies, journals or even universities could mint these and award them to reviewers and outline the expertise present. This would provide relative pseudonymity as the reviewer would just be an address. Anyone could verify the review activity based on tracking that wallet’s interactions with journal websites. One challenge to this approach is the changing nature of expertise. Scientists build skills and expertise, so some of the expertise areas might need to be updated. For early career individuals, this would especially be the case.
Paying peer reviewers (and editors) or providing them with actual incentives will also potentially improve peer review. Academics don’t value their time well, so this has not been a serious challenge so far. However, journals have recently had challenges getting enough good peer reviewers to serve for their journals, so this might finally be changing. Also, awarding coins becomes very easy, so long as there is a source in the first place. Potentially authors could lock up editor/reviewer fees in a smart contract along with their paper. They would turn this over to an editor, who would use the reviewer fees to solicit experts to provide timely reviews. So long as the reviewers provide a timely review, they get paid out of the smart contract. The editor’s job would be to ensure a fair review and check that the appropriate credentials/conflicts are managed as needed. Potentially, fees for hosting services (possibly using Interplanetary File System (IPFS)) would also be locked up in the contract (along with a small fee for the dApp creating/administering the smart contracts). Now instead of paying journals publication fees, authors would be directly paying for the services that they need: peer review, someone to manage the peer review (editor) and file hosting.
If most of the paper submission process is replaced by a smart contract, it would further ‘journal independent review’ and fragment it into ‘editor-based review’. At this point, the question is what value the journals provide. The value proposition for journals (and publishers) would be aggregators. Aggregators are important, but it’s unclear if the Nature editors could outperform an AI developing custom keywords for you and/or flagging articles being read/cited by lots of people. Or someone could make a virtual Nature editor with machine learning, and possibly perform the same.
The only thing this alternate system currently lacks is monetizing the articles. If the articles themselves are minted as NFTs, it becomes easier to control access to the article. This is where authors could sell their NFT/articles to one or more aggregators (currently you are only allowed to have an article under consideration at one journal at a time). Aggregators would provide access (possibly also via NFT) to subscribers for a custom set of articles, based on which ones they had purchased. Ideally, authors would receive a percentage of the value they brought to the aggregator (eg highly cited/read/spread articles should accrue value) This would flip the dynamic from authors paying publishers to publishers paying authors. Or if the authors preferred Open Access, they could make their article open to everyone. The value of being included in an Aggregator would depend on how well the Aggregator consistently brings value to their readers. This would also let institutions and other organizations break into the publishing/aggregating business. This would be a mixed blessing for scientific societies. On one hand, the expertise value goes up “Recommended by the American Association of Immunologists” is stronger than “Recommended by Aggregator X”, so they could make more on subscription fees. They might also be able to get lower rates from authors who want to support their society.
The challenge to this level of decentralization will be rigor. If all it takes is 4 randos with a PhD to accept an article, how good is the article? Sales becomes more important than being right, and for reproducibility, and rigor, that’s problematic. Also, with thousands of articles published daily, how do you sift out the gold? Now the Aggregators seem like they have a more valuable service, because the one place an AI likely underperforms is assessing rigor. Overall, a setup like this would help drive money to those elements directly providing value, and it would increase transparency.
We’ll talk about the current grant process later, but it’s worth noting that bringing grants onto a science-specific chain (or possibly just a science-specific token) as well could integrate everything from grants to publishing on one chain. It would also make it easier for the community to invest in projects that they like—just bridge to the science-chain and/or buy the tokens and invest in the projects that you like. If universities provided mining operations, they’d be able to drive value to their researchers beyond lab space. Of course, one current key limitation is off-boarding. Paying researchers in a science token is easy; they could exchange for ETH. Buying supplies would require an easy way to convert back to fiat. Some vendors would likely refuse to adapt, and if they are essential, researchers have little choice.
This is the rough blue-print; making these ideas reality will take a lot of work, sales and a funder willing to bet on the idea. Howard Hughes Medical Institute or Gates Foundation would be my choices; they both need to spend money annually, and they love high risk/transformative ideas like this, if sold well (and in a way the Boomers running the organizations can understand). Maybe you could convince the NSF to spend some money on it, but you’d need the background and an institution.
From what we’ve discussed, how do YOU think crypto could improve the publishing process?