Thawing of permafrost causes positive feedback loop in greenhouse effect? This is claimed in my climate change class, which is admittedly pretty sparse in rigorous climate physics. The reason is that there are pretty large carbon deposits in glaciers which will further the greenhouse effect which will expedite thawing even more.
But as I understood it, most of the greenhouse effect is caused by the positive feedback due to water vapor (more water vapor -> higher temperature -> more water vapor...) and CO$_2$ emissions are bad mainly because they fuel this feedback loop.
If this is true, isn't it possible that thawing glaciers will stunt this effect rather than expedite it?
 A: Permafrost is, unfortunately, very complicated, even by the standards of climate (which is, itself, an extremely complicated system by the standards of the problems physicists usually deal with, if you want models which make useful predictions), so there are currently no very good predictive models of what happens as far as I know.
First of all, permafrost is different than glaciers or other ice sheets, with which you seem to confuse it.  Permafrost is not ice, it is frozen soil (soil which remains frozen for many years in particular): there is some frozen water in it but it's a really tiny amount compared to that in ice sheets, for instance.  The problem with permafrost is that it is extremely hard to observe: satellite observations tell you almost nothing useful about it because it is by definition, not visible from satellites.  And it is extremely spatially variable, and there is almost no mixing (unlike oceans say, or vegetation).  So just knowing how much there is is a serious problem which requires people to actually spend a huge amount of time in remote frozen wastelands with sticks: these are really difficult and slow experiments to do.
That said, we know that something like 9% of the Earth's surface is covered in permafrost, and estimates are that somewhere between 25% and 50% of the carbon in the soil (decayed plant matter) is in permafrost.  I have a note from a talk which says about 800Gt of carbon is in permafrost (not sure if this is CO2-equivalent or C).  Human CO2 emissions are something like 50Gt / year, so this is a lot of CO2 (but, if it comes out over 100 years, say, it's not that much).
There are two pathways for this stuff to be released: if permafrost melts slowly, then you get aerobic respiration -- it gets warm enough, bacteria get in and digest the plant matter -- and you get CO2 coming out.  If it melts quickly, you get anaerobic respiration, and methane comes out.
Both CO2 and methane are not good news, and methane is really quite bad news.  CO2 we understand really well, methane I think less well because it is not stable in the atmosphere (I imagine climate models deal with methane however, although I don't know the details).
There are apocalyptic scenarios where you get a lot of CO2 being released, causing permafrost to melt faster and then dump even more CO2 and (worse) methane as the melt rate gets high enough for the transition to anaerobic respiration to occur.  As far as I know people think these scenarios are unlikely, but currently the system has too many unknowns to model well.
One important point with any of these things: what we do know is that the system is rather far from equilibrium, and is not linear.  It is therefore not safe to make assumptions like 'x hasn't happened so far, therefore x will not happen in the future'.

Disclaimer: I work on climate modelling.
A: The logic of this proposed feedback is that the permafrost may melt and release methane which is a greenhouse gas, one that is more powerful than CO2 so a smaller increase of the concentration is enough to cause the same increase of the temperature.
Subsequently, the warming caused by this extra methane will cause an additional melting of the permafrost, and the release of an even larger amount of methane, and so on.
This positive feedback loop may sound good at the level of words but the reality is that the data indicate that there is no detectable change in the release of methane, let alone detectable evidence of any warming caused by changes of the methane, so the prerequisite for the proposed loop are clearly not fulfilled in Nature. It is almost certain than the changes of the methane concentration in recent 250 years have caused less than 0.2 °C change in the global mean temperature, a negligible amount.
Methane has the lifetime around 12 years in the atmosphere, before it's absorbed or disappears, much shorter than CO2. One unit of volume of methane has about 34 times greater impact than the same volume of CO2 (the global warming potential) if measured at the 100-year timescale.

This NOAA graph showed that the methane concentration in the atmosphere has stabilized below 1.8 ppm or so – more than 200 times lower than CO2's 400 ppm. In the 30 years on the graph, the increase was less than 10%. Because the concentration of methane is about 200 times below CO2 and because the global warming potential of methane is 34 times higher than CO2, the same percentage increase of methane creates about 1/6 of the warming caused by the same percentage increase of the CO2 concentration.
Moreover, the implicitly or explicitly claimed "tipping point" surely doesn't exist because different parts of the world's permafrost have different temperatures and if this permafrost thaws, it thaws gradually, anyway.
There's one more problem: the greenhouse effect is logarithmic ($ \Delta T = K\ln(c_{final}/ c_{initial}) $) so the rate of the increase decreases as the concentration of the greenhouse gas (methane in this case) goes up. So the rate of all the real-world processes, even if they started and became significant, will actually slow down rather than speed up.
The positive feedback loop I described – melting permafrost, released methane, methane greenhouse warming, even more melting permafrost, and so on – is independent from the positive feedback loop you mentioned, the increased evaporation of water in a climate warmed up by CO2 etc. But what is not different between these two loops is that there's zero credible scientific evidence that either of them is actually occurring.
Water molecules in the atmosphere may take the form of vapor but they also take the form of clouds and clouds mostly cool the system (although their effect on the temperature depends on the type and location of the clouds a bit). The net feedback caused by the water molecules in the atmosphere may have both signs and both signs seem to be equally likely according to the existing scientific research. What is excluded is the hypothesis that this feedback is huge, of either sign.
