The Guardian has a great interview piece with Twitter’s resourceful (and ever-upbeat) Del Harvey, who in her two-and-a-half years at the social platform has gone from tackling a moderate spam issue on Twitter all by herself, to becoming the lead of a rapidly-expanding and continuously-challenged Trust & Safety team.
The biggest problem? As her team has grown, so has the size – and intellectual evolution – of the spammers themselves.
An early highlight is Harvey’s revelation that, just one month into her job, she accidentally suspended the accounts of Twitter co-founders Biz Stone and Evan Williams, as well as (then) head of product Jason Goldman.
Biz Stone and Ev Williams discovered pretty quickly that Del Harvey was serious about the task they’d set her of putting a lid on Twitter’s spamproblem. Within a month of her joining in late 2008, she had suspended both of their accounts on the grounds that they seemed to be acting suspiciously … like spam. Oh, and for good measure she also suspended the head of personnel and Jason Goldman, the head of products.
Just over two years later, Harvey, a 29-year-old with a laconic line in humour, can look back on it and joke about it. But it still gives her chills. “This message went around on the internal email saying ‘Hey, er, is anyone else seeing this ‘Account Suspended’ thing?” she recalls. “That was not my best day. But they didn’t fire me, so that was good.”
In January 2009, Twitter’s Trust & Safety team was made up of just one person – Harvey – but in those times of old, which Harvey refers to as “the good old days”, Twitter’s spam problem was close to zero. Just a couple of months later, things changed significantly, and for the worst.
“In the beginning, the spammers weren’t very sophisticated,” she says. “It was fantastic. The heavy influx wasn’t until we hit about 40m users. It was probably in March 2009 that we started seeing an uptick in spam. It’s not like a bell went off, that – ding! – we hit 40m, but that was the sort of time and scale. And then by God they evolved.”
Diet pills and porn quickly followed, using tactics such as following huge numbers of people and then unfollowing them if they didn’t follow back; those who remained would be bombarded by direct messages. The porn turned out to be easy to repel: because they contained affiliate links (where a middleman was getting paid for each clickthrough and signup), “Trust & Safety” assigned someone to contact the porn sites, getting affiliates’ accounts cancelled for bad behaviour. Turning off the money tap worked surprisingly well as discouragement.
Then in 2010, Koobface appeared on social networks, dumping malware on the machines of people unwise enough to click links to overly enticing offers. Spam and malware could go viral. Koobface became part of the wider problem for “Trust & Safety”, along with spam, and spam accounts. Basically, says Harvey, if it’s a problem with the website itself, it’s one for support; anything else, her team looks after. Copyright, impersonation, API problems, bad applications; all belong to her team.
As Twitter’s roster grew, so did Harvey’s – at an even greater pace – and her Trust & Safety team now encompasses some 30 employees. And they’re still hiring.
As for how big Twitter’s spam problem has become, Harvey is typically philosophical. “We’re so small,” she says, “Maybe some day we’ll be able to pull off the thing like the Rustock spamkill. You know, that’s fine. I’m fine with that. I will totally commit to doing that if we get as big as Microsoft.”
There’s a lot more detail about the challenges Twitter has faced and that lie ahead in Charles Arthur’s excellent piece, including false-positives, malware and the limitations of CAPTCHA, which you should read here.
And once you’ve done, head on over to Twitter and shoot @delbius a nice message to say thanks for her persistent (if increasingly up-against-it) work effort.