Cornell computer scientist Cristian Danescu-Niculescu-Mizil analyzed messages sent between players of strategy game Diplomacy to tease out early signs of future betrayal. A computer algorithm then predicted betrayal correctly 57 percent of the time, which is way better than the players themselves did. Read the rest

Research says "to-do" lists don't work, writes Daniel Markovitz at Harvard Business Review. That's not exactly what he means, though. Instead of condemning the very idea of "to-do" lists, Markovitz piece makes an interesting case for re-thinking how you use those lists. If you're throwing a jumble of stuff to be done onto a page, that's probably not going to be terribly effective. A better solution involves breaking down how various tasks fit into allotted spaces of time on specific days, and setting up that more realistic list as a part of your routine, rather than just magnetizing it to the refrigerator. Basically, it's not that "to-do" lists suck. It's that some people probably aren't using them effectively. Read the rest