Looking at the covid-19 hackathon and wondering if there would be an advantage of a dedicated social network for doctors and labs. Perhaps with possibility of making data publicly readable but not submittable. Thoughts?
A dedicated tool will never reach the level of training and routine that media like Facebook have built up through all that "brightest minds of a generation" dopamine engineering. The participating doctors are basically weaponizing, against the virus, skills and habits acquired on cat pictures and farmville or whatever guilty social media pleasure they usually enjoy.
I'm not saying that dedicated tools should not exist, I'm saying that there's both room and reason to exist for both.
Not the person who wrote that but I think read only feed of the conservation where random folks like myself could hear the conservation but not speak directly == "not submit comments". which would work fine for me. Any hypothesizes I generated should go through a vetting process where the idea generates own momentum so as not to waste their time if it ever showed up.
a ... peer review process writ smaller and more distributed.
My biggest issue with pursuing this is I have next to no breadth of knowledge in the medical or medical research fields. It would be hard for me to understand what would help doctors and research labs to share and distribute information, other than being more filterable and searchable than a facebook group.
tl;dr Kicking out the non-experts won't solve the current problems and is likely to introduce new ones. It's a tooling issue, not a platform one.
This is a difficult problem. In trying to reduce the noise by choosing who gets to speak, you immediately encounter a number of issues related to gatekeeping. Moreover, filtering by source doesn't directly address what I see as the underlying problem of presenting relevant information to the user.
Competent and aggressive moderation seems to work fairly well for setting a standard of discourse without excluding anyone (HN is a good example here) but it's time consuming. And it doesn't do much of anything to address the problem of filtering for relevant information.
I think conversations themselves (ie actual back-and-forth with meaningful input from all sides) ultimately just don't scale well beyond a small group. Topic specific mailing lists seem to be the least-worst solution to date, and a Facebook group or subreddit is essentially the same thing implemented using a proprietary platform.
It seems like we still lack the models and tooling to address these problems on a fundamental level. Plenty of academic research exists surrounding knowledge graphs, ontologies, and relationships, but in practice all the mainstream tooling that "just works" is restricted to hierarchical folders (and tags if you're lucky). Zotero is very much a least-worst reference management solution in my opinion (I detailed my thoughts in an HN comment sometime within the past year if you're curious). On the literature front we have things like Arxiv Sanity Preserver and Semantic Scholar, but we're still proverbially drowning in new publications. Tools like Memex (https://github.com/WorldBrain/Memex) and Contextualise (https://github.com/brettkromkamp/contextualise) look promising, but aren't (yet?) a complete solution.
In a past life I was an academic researcher. My objections are nothing specific, I just never had any obvious uses for it and I'm deeply dissatisfied with the centralized Facebook model so to speak.
ResearchGate seems to blend Facebook with LinkedIn and arXiv. That's great for finding and applying to open positions, but not so much for casual discourse. There's an entire conversational continuum, ranging from things I want to openly share with a prospective employer to crazy ideas discussed with one or two coworkers.
For small semi-private conversations, we have email (and dozens of other services and protocols, each with various pros and cons). For large group communications, we have mailing lists and forums of various sorts. For LinkedIn, why wouldn't I just use LinkedIn itself? There's also ORCID, among other things.
