By @edhagen.net

I gave my new preprint its own Bluesky account:

And you should too. Why? The short version is that the account, which has a unique ID analogous to a doi, and which under the hood is a database that can hold the main text and all files related to the study, becomes the publication.

Goodbye Elsevier and the rest of the scientific publishing oligopoly!

The longer version:

The problems facing social media & scientific publishing are similar: both are dominated by powerful oligopolies. The AT Protocol technology underlying Bluesky that aims to solve the social media problem might also help solve the scientific publication problem.

The social media oligopoly traps users in walled gardens, owns their identities and social graphs, and shapes feeds to benefit themselves. The scientific oligopoly owns research paid for by the public, charges them to access it, and increasingly publishes junk science to increase profits.

To free social media from the oligopoly, several decentralized technologies have been developed, e.g., ActivityPub (Mastodon), nostr, web3 (blockchain), and ATProto (Bluesky).

The ATProto solution works like this: each account gets a unique ID that points to an account-specific database (literally sqlite). The database, termed a personal data server (PDS), can be hosted anywhere, yet still interact seamlessly with the network:

All user content -- posts, replies, likes, follows -- lives on the PDS. Importantly, third party apps can read PDS data and (with permission) write it using existing or new data models (termed lexicons). Large compute-intensive services, run by Bluesky or third parties, aggregate data from all the PDSs and generate customized feeds for each user of client apps.

If Bluesky turns evil, folks could migrate their accounts to other services, retaining all their data and their social graph (although this vision is not yet fully realized in practice, there are currently over 2000 independent PDS's on the network). The open protocol and credible exit will hopefully create a healthier social media ecosystem.

How could this technology help fix scientific publishing, and why should academic papers each have their own account?

First, the scientific publishing oligopoly maintains its hold on scientists because they mint the currency in our prestige economy: Our careers depend on having papers accepted in prestigious journals, such as Science or Nature. Scientists will only adopt a new publishing paradigm if it affords a similar opportunity to gain prestige, and social media provides

it: likes, reposts, and followers! Striving for prestige might sound crass, but prestige is simply public recognition that one is providing value to one's community.

Second, prestige requires eyeballs. Bluesky helps solve this problem: many scientists are already here. Follow my preprint account! 🙂

Third, papers typically have multiple authors, and authors contribute to multiple papers. Each needs a unique identity. The open AT Proto infrastructure assigns a unique ID to each account, termed a decentralized ID (DID):

For papers, the DID is analogous to a doi. For researchers, the DID is analogous to an ORCID (Open Researcher and Contributor ID). Giving papers and researchers separate accounts allows each to be clearly identified and linked to each other via their DIDs, in a many-to-many graph.

Fourth, paper accounts give discussions and news about papers a natural home. By following a paper account, interested users will get updates about it in their feeds. Moreover, papers could follow relevant papers. Creating accounts for papers that follow other relevant papers generates a citation graph.

Fifth, lexicons (data models) and apps specific to academic publications could be developed, allowing all data about a paper -- preregistrations, text, data, code -- to be stored together on the PDS. Many researchers already create individual GitHub or OSF repos for each publication. Creating a Bluesky (ATProto) account for a paper is similar to these, both conceptually and technically.

Sixth, peer reviews could happen "on protocol", stored on the reviewer's PDS but linked to the paper via the DID. Because the protocol and network are open, anyone can develop custom lexicons and apps. chive.pub has already implemented many of these features and more.

If Bluesky turns evil, paper and researcher accounts can easily be moved to better hosts, yet still interact seamlessly with the network. Better yet, existing paper archiving services like Zenodo.org and arXiv.org could create a PDS interface (api) so that each paper submission would also automatically create an ATProto account.