When Robots Attack Online Polls: A Report on Ourselves

"This is an unscientific, informal survey for the interest and enjoyment of TIME.com users," goes the disclaimer below the Kurdish statehood poll, "and may not be indicative of popular opinion." But that's overly apologetic. Online polls -- all online polls -- are unscientific in that they're not random samples of the population. Instead they self-serve: They measure how many people cared enough about the issue to take the time to find the poll and pick a side, and they're dead-on scientific about that... until robots attack. "The more popular the poll, the more trouble we have defending it from hackers," says Time Inc. New Media applications developer Anthony Martino.

Martino breaks down TIME.com poll popularity into three categories of voting results: the thousands, the tens of thousands, and the hundreds of thousands (and up). It's simple statistical penetration: How many web users come to a posted poll before one of them happens to be a serious -- and seriously skilled -- hacker? A mild-mannered TIME Daily poll such as "No more Independent Counsels?" fell into the first category -- fewer than 2,000 votes in its first week -- and it tantalized no intruders.