Lovely piece of informal research reported in Nature:

I investigated this idea at a meeting where speakers were given 12-minute slots. I sat in on 50 talks for which I recorded the start and end time. I decided whether the talk was boring after 4 minutes, long before it became apparent whether the speaker would run overtime. The 34 interesting talks lasted, on average, a punctual 11 minutes and 42 seconds. The 16 boring ones dragged on for 13 minutes and 12 seconds (thereby wasting a statistically significant 1.5 min; t-test, t = 2.91, P = 0.007). For every 70 seconds that a speaker droned on, the odds that their talk had been boring doubled. For the audience, this is exciting news. Boring talks that seem interminable actually do go on for longer.

That figures. As Alex Tabbarok commented,

the fundamental explanation is that a boring speaker doesn’t think about their audience. A speaker who cares puts herself in the audience’s shoes, thinks in advance about what is important, how much an audience can absorb in one sitting, where a graphic would be helpful and so forth. A good speaker plans and practices and thus ends up being interesting and ending on time.