Apparently the cause was that I interacted with code that GH detected as malicious. The weird thing was that GH left this supposedly malicious code as publicly visible in multiple personal repos while hiding my unrelated organizations. These organizations were home to highly used scholarly software, research projects, and datasets. I started getting emails from users like:
> I was looking at a manubot issue and found out that the repo vanished from the internet. What happened?
I got the situation resolved through my contact at GitHub, but still no response via the GH support ticket after 2 weeks. Seems like GitHub is shooting themselves in the foot. Strange.
At some level, any system for evaluating research will have to rely on peer assessment. And some amount of groupthink is inevitable with peer assessment.
But the ResearchHub model of open collaborative assessment could have provided more incentives and visibility for research that deviated from the beta amyloid hypothesis.
It sounds like gatekeepers were a large problem here... i.e. peer reviewers for journals or grants that were insistent on beta amyloid. Disentangling dissemination from assessment (as preprints are doing) will help bring contrary findings to the light.
We also need to make all assessment public. Rather than basing the validity of a study on a few anonymous unpublished reviews, let anyone provide feedback and increase the sample size and diversity of assessment. Finally, add some incentives for contrarians who advance unpopular hypotheses that time proves correct.
Anyways, lot's of thoughts, but I agree a radical shift in how scientists communicate could help get to the root of the hivemind described in the article and by @cockatiel_day.
> As of March 2017, we find that Sci-Hub's database contains 68.9% of all 81.6 million scholarly articles, which rises to 85.2% for those published in closed access journals.
> Coverage also varies by publisher, with the coverage of the largest publisher, Elsevier, at 97.3%.
> we estimate that over a six-month period in 2015–2016, Sci-Hub provided access for 99.3% of valid incoming requests.
As an project leader on ThinkLab, I'd like to highlight the awesome [markdown citation engine](http://thinklab.com/help/writing_in_markdown). Provide only a doi and ThinkLab takes care of the rest.
`Co-authored-by: Copilot <[email protected]>`
That is not coauthorship. And if it were, every pre-commit tool that automatically reformats or fixes code should also append themselves as coauthors!