Argues the ejection is the predictable end-state of researchers signing away copyright to publishers and never building counter-infrastructure like the GPL, OSI, or Creative Commons. Frames this as the same fight Stallman won for software in 1983 that academia is losing in 2026 because no equivalent legal scaffolding exists for publicly funded research.
Highlights that the ejected researchers' papers are already indexed on PubMed and freely linkable, and that one tenured endocrinologist had distributed her own reprints at ADA for fifteen years without issue. The framing emphasizes the disconnect between publicly funded research, author intent, and a copyright the author no longer owns.
Submitted the Ars Technica story to HN where it hit #2 with 243 points, signaling the framing that ejecting researchers over papers already on PubMed is the kind of absurdity worth surfacing to a technical audience. The submission's traction reflects broad sympathy for the open-access framing.
A significant portion of the 612-comment thread came from former academics pointing out that the publisher copyright transfer agreement every author signs at acceptance is unambiguous about reproduction rights. Their position is that ADA and the publishers are within their contractual rights, and that the surprise reflects how casually researchers have ignored a clause they all signed.
Roughly half of the HN thread framed the ejection as a vindication of the open-access movement — if the papers had been published under CC-BY or in an OA venue, no security guard could have intervened. Their argument is that the only durable fix is changing where and how research is published, not negotiating with conference exhibitor agreements.
At the American Diabetes Association's 86th Scientific Sessions in Chicago last week, security escorted at least four researchers off the floor for handing out paper reprints of studies they themselves had authored. The papers in question — covering continuous glucose monitor accuracy, GLP-1 prescribing patterns, and one meta-analysis on insulin pump interoperability — were already indexed on PubMed and freely linkable. The reprints were the formatted publisher PDFs, the kind every academic emails to colleagues without a second thought.
ADA's stated rationale, per the Ars Technica writeup that hit #2 on Hacker News with 243 points: the conference's commercial exhibitor agreement reserves the right to distribute branded printed materials to paying sponsors, and the publishers (Elsevier, Wolters Kluwer, and the ADA's own *Diabetes Care*) hold the reproduction rights the authors signed over at acceptance. One ejected researcher, a tenured endocrinologist at a Midwest academic medical center, told Ars she had been handing out reprints at ADA for fifteen years. The papers were hers. The data was publicly funded. The PDFs were stamped with a copyright she no longer owned.
The HN thread — 612 comments at last count — split predictably. Half were variations of "this is why we need open access." The other half were ex-academics pointing out that the copyright transfer agreement every author signs is exactly this clear, exactly this enforceable, and exactly this routinely ignored until someone decides to enforce it.
Software engineers reading this should feel a specific kind of déjà vu. This is the same fight Stallman picked in 1983, the same fight that produced the GPL, the OSI, Creative Commons, and the entire modern open-source stack — and academia is losing it in 2026 because researchers never built the legal infrastructure we did.
The parallel is exact. A researcher writes a paper. They sign a copyright transfer to get it published in a journal that confers career legitimacy. The journal then sells access back to the researcher's own institution, charges $40 per article to anyone else, and — as of last week — sends security to confiscate printed copies the author tries to give away at a conference the author paid $1,200 to attend. Substitute "engineer" for "researcher" and "proprietary fork" for "journal" and you have the pre-1998 software industry. The difference is that we built a way out. Academia did not.
The technical infrastructure for academic open access exists — arXiv, bioRxiv, medRxiv, the Public Library of Science, Sci-Hub, the entire green/gold/diamond OA taxonomy. The licenses exist — CC-BY is functionally identical in spirit to MIT. What doesn't exist is the cultural enforcement layer: no academic hiring committee will deny tenure for publishing in Nature, the way a serious engineering team will reject a dependency licensed under the JSON License's "shall be used for Good, not Evil" clause.
The author-rights agreements signed at the front of the publishing pipeline are the equivalent of a CLA that assigns all copyright to a corporate sponsor, with no covenant back to the contributor. We stopped accepting those CLAs around 2015 (see the Project Harmony pushback, the eventual death of the original Oracle Java SE assignment requirement). Academia still signs them by default — and the *Diabetes Care* author agreement, in particular, transfers "all rights, including the right to reproduce, distribute, and prepare derivative works," with the author retaining only the right to reuse content in their own teaching and to deposit a preprint version after a 12-month embargo. The published PDF — the version with the journal's typesetting and DOI — is the property of the publisher forever.
The comment section had one engineer make the cleanest reframe: "This is what would happen if GitHub could send security to your conference booth because you handed out a printed copy of a README from a repo you transferred to a foundation." That is exactly what happened. The researcher held the moral authorship; the publisher held the legal copyright; the conference enforced the legal copyright against the moral author at the moral author's own talk.
If you ship code, three concrete takeaways.
First, audit your own contributor agreements. If you maintain an open-source project under a foundation (Apache, Linux, CNCF), check whether the CLA grants the contributor a perpetual non-exclusive license back. The Apache CLA does. The old MongoDB CLA, pre-SSPL, did not in the same way. The asymmetry matters more than the license tag on the repo — it determines who can stop you from using your own work later. The ADA story is the worst-case version of that asymmetry actualized.
Second, if your team consumes academic research — anyone building in ML, bioinformatics, security cryptography, or any field where the paper trail matters — start treating the preprint server as the canonical source and the journal version as the proprietary fork. arXiv and bioRxiv are now the academic equivalent of an upstream Git repo; the journal PDF is a downstream binary distribution with restrictive licensing. Cite the DOI for credit, link the preprint for access. This is already how most ML research operates de facto and it's part of why ML moves faster than the fields still gatekept by *Cell* and *Nature*.
Third, if you're a dev whose company sponsors academic conferences (and a lot of healthcare-adjacent SaaS does — Epic, Oracle Health, every CGM startup at ADA), this story is a reputational signal. The booths next to the security-ejection were Dexcom, Abbott, Eli Lilly. The optics of standing in a sponsor hall while the people whose research validates your product get walked out for sharing it are bad enough that a sharp comms team should already be drafting policy responses. Some won't. The ones that do — public commitment to fund open-access publishing fees for any paper that uses their device's data — get a cheap PR win.
The trajectory is not in doubt. The NIH's 2023 public access policy already requires immediate open access for any federally funded research starting January 2026 — meaning most of the papers confiscated last week will be legally redistributable inside six months, regardless of what Elsevier's agreement says. The ADA enforcement action will accelerate the funder mandates, embarrass the publishers into another round of "transformative agreements," and ultimately push more of the work to preprint-first. Software's fight took thirty years and produced an industry where proprietary code is the exception, not the rule. Academia's version is fifteen years in and the inflection point is visible. Last week's ejections will be cited in the obituary of pay-to-publish.
I think you have to start from:- the organizers are doing what they think is best for the attendees, and what the majority of attendees wantAnd, you should consider: "what if everyone did that?"I gather they had a paper they thought was important, but was not accepted to the conference (so
Link to the editorial: https://diabetesjournals.org/care/article/49/6/901/164764/Mi...
Headline is weird. It's not a copyright thing, as I had assumed. It was because it was an editorial criticizing how the administration is running the NIH.
".Some questioned how handing out reprints of an editorial published in the ADA’s own journal, at the ADA’s own annual conference, could be construed as a violation of that code.."The God Emperor is not to be questioned.
Top 10 dev stories every morning at 8am UTC. AI-curated. Retro terminal HTML email.
> “They physically grabbed us, forced us out of the conference center, and now are telling us we can no longer attend this meeting,” Kelly told MedPage Today, which first reported the incident. “They’re taking our lanyards. It really has come to this in America. Censorship is real. America needs