This fascinating paper (sub req) in Nature on tipping points should have made a huge splash in the media. Hell, it should have at least prompted one of those 10,000 word posts on The Oil Drum. In her NSF press release, Cheryl Dybas does a nice job of summarizing the key findings. Here's two snippets directly from the paper to whet your appetite:
In systems in which we can observe transitions repeatedly, such as lakes, ranges or fields, and such as human physiology, we may discover where the thresholds are. If we have reason to suspect the possibility of a critical transition, early-warning signals may be a significant step forward in judging whether the probability of an event is increasing.