Facebook Beacon

» see facebook.com as it lived, on the wayback machine

~ the obit ~

the first time i heard about beacon was through the overstock-engagement-ring story. an american user named sean lane bought a diamond ring on overstock as a surprise for his wife. facebook posted the purchase to his news feed. his wife was on his friends list. that was the story that traveled across the world, and that was the story that ended up filed in federal court two months later.

beacon launched on november 6, 2007 at a new york event where mark zuckerberg announced facebook ads as a three-piece platform. beacon was the third piece. forty-four partner sites had agreed to ship a small javascript snippet that fired whenever a logged-in facebook user took an action on their page. buy a movie ticket on fandango or rent a film from blockbuster, and the action would appear on that user's facebook feed as a default-on broadcast.

i was in turkey in 2007, finishing university, mostly running msn messenger and the early facebook profile pages that turkish friends had started signing up for that year. beacon was a us-only ad platform tied to us merchants. it did not touch me directly. i read about it the same way most of the non-american web did: as a privacy scandal in english-language coverage, mostly through valleywag and the new york times. the shape of the story was already familiar by the time it reached us. an opt-out feature that should have been opt-in.

~ the rap sheet ~

~ how it landed ~

the technical design of beacon was the part that made it indefensible. a partner site would call facebook's beacon endpoint with the user's purchase data. facebook would match it to a logged-in account and queue a feed story. the user would see a small toast in the corner asking if they wanted to publish the story. if they did not click it, the story would post anyway after a brief window. and if the user was not even on facebook at the moment of the purchase, the data still arrived at facebook's servers.

that last detail is what later civil liberties writers pointed to as the lasting problem. beacon was sending data on logged-out users to facebook regardless of consent. by the time the lane v facebook plaintiffs were filing in august 2008, the security researcher stefan berteau had already published proof that the tracking worked without an active session. facebook's own statement at the time admitted that the system was live for about a month before the opt-in change went in.

~ the cracks ~

on december 5, 2007, four weeks after launch, mark zuckerberg posted a public apology on the facebook blog admitting the company had built a bad product. the post said beacon would become opt-in. by then moveon.org had collected fifty thousand signatures against the program and large partners including coca-cola had quietly pulled out. the new york times had run multiple cover pieces on the privacy backlash through november. zuckerberg's apology was the first one of his career that the press treated as serious.

the apology did not end the platform. beacon stayed live in opt-in mode for almost two more years, generating limited revenue and continuous lawsuits. the lane v facebook class action consolidated the major civil claims by mid-2008 and settled in september 2009 for $9.5 million plus a permanent shutdown of beacon. judge richard seeborg approved the settlement on march 17, 2010. most of the settlement money went into a privacy foundation rather than to individual class members, which the ninth circuit later affirmed in a 2012 opinion that became a frequent citation in cy-pres-class-settlement law.

~ what got buried with it ~

beacon is the cleanest pre-cambridge-analytica example of a social platform tracking off-site purchases without meaningful consent and treating the feed as an automatic broadcast surface. it is not the most famous facebook scandal anymore. it is the earliest one that produced a settled federal class action with a written remedy. the cy-pres foundation that the lane money funded ran for several years before winding down.

the larger frame is that beacon was the moment facebook decided what a default ought to be. the company chose default-public broadcast as the answer and then walked it back after public pressure, but spent the next decade and a half repeating versions of the same choice in different products. by 2018 the cambridge analytica disclosure made beacon look quaint by comparison. it was not quaint. it was the first version of the same pattern, fully visible in 2007 and written into a settlement two years later, and apparently not enough of a warning.

@}-->--

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

« previous  ·  random  ·  next »