Sci.env: IPCC understates the risks?

There is an interesting exchange today over at sci.env, which touches on some things I've been thinking about recently. See here for the thread & if you want to know who they are (well scroll up a bit); I'll extract some here and then comment. Both sides are people I respect (well they are on my blogroll) and yet they are in violent disagreement. I had the slightly unsettling experience of reading the first post, thinking "yes thats spot on" then reading the reply and thinking "hmmm, some valid points there too".

> I believe the IPCC genuinely constitutes a consensus, but I believe
> that the consensus severely understates the risks.

> There are several reasons for this. Notably:
>     - models are tuned for small signal accuracy and can't capture
> large nonlinearities

I don't really understand what you can mean by this point. The models
perform reasonably well across a wide range of conditions including the
6C cooling at the last glacial minimum, the ~12C annual temperature
cycle (more at higher latitudes), not to mention the basic spatial
patterns in the first place. A 3C temperature rise is not large
compared to the range they've already simulated, and there are good
reasons to expect the models to be largely correct in broad detail.

>     - carbon cycle exacerbating feedbacks are not sufficiently attended
> to, and are buried under the rug in the simulation scenarios

Carbon cycle feedbacks have been included in a number of models
(C4MIP), my understanding is that the effects are generally modest, and
even for the outliers it is not something that turns a mainstream
projection into a nightmare. One possible wildcard is a methane burp,
about which I know little but it does on the face of it seem worthy of
consideration.

>     - the IPCC seeks the most likely response of the system rather than
> the risk-weighted outcome, which essentially hides the worst cases

Um...no. That's simply not true. It describes the range of outcomes
(according to some rather vague probabilistic statements).

>     - most scientists are conservative in personality and don't like
> making a big fuss, so shy away from clear statements of the enormity of
> the risk we face

Well...this may be true but even if so is highly misleading. It's not
"most scientists" who we hear, either in the media or through
assessments such as the IPCC. It's those scientists who make their
opinions forcefully enough who are heard, and I absolutely disagree
that this subset are conservative in personality and do not like making
a fuss. 

The point of all this is, what-do-you-tell-the-public. There's no great problem with what-do-you-tell-fellow-scientists: thats easy: you publish your research and they read it, or not (thats not really true either when I think about it: its true within physical climate; but when you start to try to do impacts on ecology, then the Bio's need the phys stuff interpreted). But the public need it all interpreted: they are not going to read the original papers (even if they have access to them).

I have much sympathy for most likely response of the system rather than the risk-weighted outcome (its not technically true: the scenarios are not probability-weighted; but you know how it goes). When I give general-public talks (and you can find the stock one I do, by following a few levels of links from this, if you want to) I downplay the overenthusiasm you find in the media for disaster scenarios, but always with the ever-so-slightly guilty feeling that I may be wrong. I point that out too - I try to mention the uncertainty - but I have the impression that people have problems keeping up with everything and are going to miss the subtle side messages.