Meta.org

» see meta.org as it lived, on the wayback machine

~ the goodbye ~

A biologist in 2015 could not read their own field. Something close to two million peer-reviewed papers were going out every year, and the standard tools for keeping up with them were a keyword alert and a journal's table of contents, both of which were designed for a literature a fraction of that size. Meta's answer was to let a machine do the reading. It parsed papers and preprints, worked out what they were about and who had written them, and let you subscribe to a subject the way you would subscribe to a podcast.

The company's description of itself was careful and worth quoting, because it is the clearest statement of what the thing was for: a biomedical research discovery tool that analyzes and connects millions of scientific outputs to give you a comprehensive view into science. In practice you searched a term or built a feed, and the feed kept filling with papers and preprints that a model had decided were about your subject. Altmetric data sat alongside the results, so you could see which papers were being argued about outside the journals.

It started in Toronto in 2010 as Sciencescape, built by the siblings Sam and Amy Molyneux. By late 2016 the company said it had analyzed over 26 million papers and profiled 14 million researchers, with publication dates reaching back into the 1800s. In January 2017 the Chan Zuckerberg Initiative bought it, the first acquisition that organization ever made, and promised to take the subscription fee off and make the tool free for everyone.

The literature I keep up with is computer vision, and it does not have this problem. arXiv posts the preprint before anyone has reviewed it, the conference catches up months later, and the field is small enough to skim between other work. Biomedicine runs on the opposite settings: far larger and far slower, scattered across thousands of journals with paywalls in between. Without machine help it is simply not readable at the rate it is produced, and Meta was built for that problem. I have never had to stand where you need to be standing to feel it.

The bet underneath it was that citation counts and journal prestige are lagging indicators, and that a machine reading the text itself could surface a result months before the field agreed it mattered. That is a reasonable bet, and it is the same one Scirus had made before it and Microsoft Academic was making alongside it. Both of those have graves here too.

~ the chronicle ~

~ five letters, two organizations ~

On 28 October 2021 Facebook announced at its own conference that the company would be called Meta. The same day, a banner went up on meta.org: "Meta.org will sunset March 31, 2022. Meta will be supported through March 31, 2022. In the lead up, we will work with you in transitioning to alternative open services."

CZI's stated reason was the transition to open alternatives, and it never once said the name was the problem. Vice reported the timing anyway, under a headline about Zuckerberg's foundation killing its science company on rebrand day, and Wikipedia's account of the episode states that Meta's assets were transferred to Facebook to enable the rebranding. Two organizations with the same backer, one of them a philanthropy and one of them a public company, needed the same five letters. Only one of them had a product people would miss.

~ the last capture ~

The Internet Archive's record of meta.org ends on 28 March 2022, three days ahead of the announced date, with a 301 and nothing behind it. There is no tombstone page and no redirect pointing researchers anywhere. The site that had read 26 million papers simply stopped being fetched.

Ask a DNS resolver for meta.org in 2026 and it fails. Microsoft Academic, which was doing a version of the same job at a larger scale, had closed on 31 December 2021, exactly three months before Meta's date. Two of the serious attempts to make the scientific literature navigable ended inside one quarter, and the field they were built for is still publishing two million papers a year.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »