Fergus Brown asks a sensible question:

in the comments about Worry,

I want to know what we are meant to understand by the term 'collapse'. It conjurs images of a Larsen-type crack and float-off, but I don't think this is what is expected of the WAIS, for example. Does the term refer to a point at which sudden and massive changes occur almost immediately, or to a system-state which represents a tipping-point, beyond which certain (relatively slow) processes are irrevocable?
IOW, what does a 'collapse' of the ice sheets mean?

UPDATE: Fergus has a blog

Steve Bloom, ankh and the mice do a fine job, and Eli adds a piece, but the Rabett would just like to add something from Hansen's A Slippery Slope: How Much Global Warming Constitutes “Dangerous Anthropogenic Interference”?

Time constants: the slippery slope. Three time constants play critical roles in creating a slippery slope for human society: T1, the time required for climate, specifically ocean surface temperature, to respond to a forced change of planetary energy balance; T2, the time it would take human society to change its energy systems enough to reverse the growth of greenhouse gases; T3, the time required for ice sheets to respond substantially to a large relentless positive planetary energy imbalance. I define “substantially” to mean a total sea level rise of at least two meters, because that would be sufficient to flood large portions of Bangladesh, the Nile Delta, Florida, and many island nations, causing forced migration of tens to hundreds of millions of people. That criterion requires an ice melt contribution from Greenland and Antarctica of at least 1.5 meters, given the approximate half meter contribution expected this century from ocean thermal expansion and alpine glaciers.

T1, the climate response time, is 50-100 years, as a result of the large thermal inertia of the ocean. T2, the energy infrastructure time constant, also is perhaps 50-100 years. Although new technologies that reduce or eliminate greenhouse gases might be developed rapidly, these need to replace a huge fossil fuel infrastructure, and this technologic task is preceded by the time required to achieve world-wide agreement on the need for replacement.

T3, the ice sheet response time, is the time constant of issue. I argue that T3 is of the order of centuries, not millennia, as commonly assumed. Growth of ice sheets requires millennia, as growth is a dry process limited by the snowfall rate. Ice sheet disintegration, on the other hand, is a wet process that can proceed more rapidly, as evidenced by the saw-toothed shape of glacial-interglacial temperature and sea level records. For example, I referred above to the 20-meter sea level rise that occurred in about 400 years during deglaciation 14,000 years ago.

The take home is that ice lost in centuries stays lost for millenia or more. Once the ice starts to go, it is gone. Hansen believes that T3 ~ T1 + T2 and is of the order of 1-3 centuries for reasons set forth in the Arxiv manuscript, the reference given above and elsewhere.

The likelihood that T3 is comparable to T1 + T2 has a staggering practical implication. T3 >> T1 + T2 would permit a relatively complacent “wait and see” attitude toward ice sheet health. If, in the happy situation T3 >> T1 + T2, we should confirm that human forcings were large enough to eventually alter the ice sheets, we would have plenty of time to reverse human forcings before the ice sheets responded.

Unfortunately, T3 ~ T1 + T2 implies that once ice sheet changes pass a critical point, it will be impossible to avoid substantial ice sheet disintegration. The reason for this is evident in the definition of the time constants. The comparability of these time constants, together with the planetary energy imbalance, make the ice sheets a ticking time bomb.

Is this correct? Wanna bet the planet friend? (OK, but we do need to concentrate on this issue until we better understand it, not to try and hand wave it away).