Meta's Tale Becomes a Ghost Story
Its founders are gone, its traffic slight, and its presence a misty apparition
Every start-up hopes to live a long and profitable life. Not all make it, perishing from one cause or another. Then there are those which linger among us, as if their very souls cannot comfortably depart this earthly realm — a ghostly spectre of their once vibrant, hopeful early selves.
It looks like Meta has entered the spectral zone.
It was just over four years ago when Meta, the start-up planning to text-mine all the scientific literature and produce AI-driven insights, was acquired by the Chan Zuckerberg Initiative (CZI). Sam and Amy Molyneux, siblings from Toronto, started Meta in 2010, and it was viewed by some as transformative, with one board member writing at the time of its acquisition:
The acquisition of Meta by the Chan Zuckerberg Initiative (CZI) promises to transform scientific investigation. As a byproduct of this, it will likely transform scientific publishing as well.
We don’t know how much CZI spent — Meta had raised $9-10 million prior to the acquisition — but we do know that Meta seems to have faded ever since that peak moment. You just don’t hear about it anymore. I asked around earlier this week, and the perception is the same generally — once the founders departed, Meta stopped contacting publishers or platforms. Since then, nobody has seen any publisher-facing tools, services, or applications.
As a result, it looks like Meta is . . . a ghost. Sam and Amy have departed — Sam in late 2019, Amy in early 2020. Instead of a standalone startup, Meta is described by its owner as a “project” inside of CZI, one run by a Director of Science Product Management (Patricia Brennan) and a Director of Programs (Jennifer Lin). Both names may be familiar to readers, as they have held multiple positions over the years in scholarly publishing.
The company is located in Redwood City, California. Judging by appearances, it’s been absorbed into Facebook’s facilities, because if you search Google Maps for it, you only get metal fabrication and metal work shops, and one unrelated startup using the name that is listed as “Permanently Closed.”
The company has no jobs posted that I could locate.
Since October 2019, anyone can sign up for a free account. This change occurred about the same time Alex Wade, a Technical Program Manager at CZI, appeared at the Force 11 conference to discuss how societies might help Meta attract sign-ups. Not much was heard from Meta after that, according to people I contacted.
In retrospect, October 2019 may have marked the end of Meta as a commercial player.
Despite being free, Meta seems to be underutilized compared to commercial content aggregation sites. For example, its visits and traffic ranking are far lower than Ovid.com, according to SimilarWeb (Ovid is blue, Meta orange):
Meta’s still active in a way — it has an active blog on Medium, for instance — but contact with the press person at CZI during the development of this piece was a little odd. Reporters seem to have trouble with CZI, with one story from 2017 about Meta noting, “Due to a CZI edict, Sam Molyneux was not available to comment.” Sure enough, I messaged Sam via LinkedIn, and had no response.
The press office at CZI seemed spooked by the outreach, and it took a while for the responses to feel natural. Even then, it felt a little unearthly. For instance, a publisher I contacted suggested that one publishing person who had worked for Meta was no longer working for CZI or Meta. I pinged this person on LinkedIn to confirm, and within hours received an email out of the blue not from this person, but from the press office at CZI, confirming this person still works for Meta.
Spooky.
I also asked for a list of current initiatives or programs at Meta, and received no reply. Overall, exchanges confirmed the impression that Meta is not a top priority for CZI, or something those currently or formerly involved are comfortable discussing.
Even projects seem to vanish. One researcher — Andrew McCallum, Director of the Center for Data Science at the University of Massachusetts Amherst — received $5.5 million from CZI for the Computable Knowledge Project. The idea was to find ways to convert the information from papers into knowledge using machine learning, McCallum’s area of expertise. On his projects page, the Computable Knowledge Project only links to a 2018 video on Facebook, circa the time the grant was announced, in contrast to a project he’s done with Google about OpenReview, which links to what an active URL and related project. I tried numerous times to reach McCallum to find out where the project stands, as nothing has been written about it that I could find since around the time the grant was made in 2018. He didn’t answer my calls or reply to my emails.
In press coverage at the time, McCallum says one of the main draws for him was Molnyneux’s involvement:
“I felt like I had met a scientific soul mate,” McCallum said of his first meeting with Meta’s founder Sam Molyneux. “We just shared so many of the same goals for using artificial intelligence and machine learning for revolutionizing science. So I just knew that I wanted to work with them.”
Coverage at the time said that the Computable Knowledge Project would eventually reside within Meta, with McCallum focused on the AI research and Molyneux focused on product development. Molyneux left Meta less than a year later.
From a brand management perspective, Meta has a jumbled morgue, as the collection of logos below shows. In some treatments, “Meta” is capitalized; some have the “t” fully crossed, some not; there are different network patterns or renderings; some use a superscript alpha character; there’s a haunting serif M version; and there’s one with the URL. The third one from the left has the “eta” slightly cut off at the bottom — this is not anything I introduced, but another sign to me of someone not paying close attention as a brand designer or manager:
It’s not as if Meta was on a path to success before CZI swooped in. As Rick Spence covered it in the Financial Post in 2017, Meta was about to run out of money, and at one point was at risk of being unable to pay salaries.
Of course, CZI itself may not be up to the task, as I’m guessing CZI isn’t Facebook’s main priority. Around the time of the Meta acquisition, CZI’s high-profile CTO departed. He now runs Roku.
In 2018, Bill Gates critiqued CZI for over-committing — taking on too many things to realistically achieve results.
The story of Meta may not be over — but a once-promising startup appears to be but a shadow of what it once hoped.
Update on Preprint Estimates
In early January, I examined how a popular estimate of Covid-19 preprints is likely overstated by as much as 58%. This week, in a Twitter exchange with one of the people behind these estimates — Bianca Kramer — I was told that there are no plans to rectify the situation, even to the point of Kramer defending the entirely redundant entries from RePec, an index of preprints, which added 10% to their estimate for no discernibly good reason.
I think this goes to show that even when you call something “data,” egos and incentives can still drive its collection, correction, or curation.