Twitter spam

I don’t know how Twitter handles spam internally. They’re probably devoting a lot of time to fighting it.

But I don’t think it’s unreasonable to observe so much repetition in the still-visible spam techniques and conclude that Twitter is being extremely conservative about deploying automated heuristics, relying heavily on the “Report Spam” feature instead.

Spam-fighting is always a tricky balance: if it’s too aggressive and automated, it’ll prevent some legitimate messages from reaching their recipients. But if it’s too conservative or manually triggered by user reports, a lot of spam will get through.

The operators of spammable services need to decide where their priorities are on that spectrum: severely annoy a small number of your users by not delivering some legitimate messages, or moderately annoy a large number of your users by showing them too much spam.

Twitter seems to have chosen the latter. At this point, given their resources, it’s almost certainly a philosophical choice — e.g. “every message must be delivered” — and not because of a lack of spam-fighting abilities.

There are three big problems with this approach:
A lot of spam is shown to users before it’s cleared away by the few that report it (and whatever actions result from that). The spam succeeds. And if only, say, 1 in 100 people report spam that they’re shown, the spam is annoying quite a lot of users before anything is done about it.

It appears to users that the service is taking a passive, almost neglectful approach to spam, which diminishes the motivation to use that “Report Spam” button. If the ratio of spam views to reports gets worse — say, if only 1 in 500 people report it — then spam starts to anger even more users before anything is done about it.

Report-and-respond-later systems are far less effective when the barrier to posting new spam is extremely low. In Twitter’s case, who cares if they ban a spam account after it has spammed 500 users, if the spammer has hundreds or thousands of other accounts that it can keep creating at nearly zero cost?

Fundamentally, I believe Twitter’s priorities here are wrong. Twitter needs a far more aggressive, automated, proactive, heuristic-based anti-spam system. And if someone has trouble legitimately tweeting a link with no text to 100 people in a row who don’t follow them at precise 1-minute intervals, that’s just the price we’ll have to pay.

In the meantime, I’m never using the “Report Spam” feature again, because it just seems like I’m wasting my time.

Marco Arment

@marcoarment New York, NY
Creator of Instapaper. Amateur writer. Coffee nut.
http://www.marco.org/