Ah, I see the difference Gerhard (and sorry for the miss-spelling as well). Thanks for taking the time to explain it to me. I do see that as a flaw in the IPCC report, and I should also be clear that I don't feel the IPCC is without bias, as Russell and others have argued very successfully. My only concern, in general, is that while there are many points that are yet to be hashed out, I would like to see the conversation limited to verifiable statements in clear language, with links and references that can be proven. If you can't remember where you heard it, please don't post it. In this case, you posted a reference and the language was perfectly clear and correct from your point of view, but I simply didn't understand it. I'm glad to hear you can see some agreement with my own position re: Erring on the side of caution. -- James Newton. -----Original Message----- From: piclist-bounces@mit.edu [mailto:piclist-bounces@mit.edu] On Behalf Of Gerhard Fiedler Sent: Monday, October 15, 2007 03:46 To: piclist@mit.edu Subject: Re: [EE]:: Global Warmimg - Al Gore James Newton wrote: > And the /current/ IPCC report, summary for policy makers, at > http://ipcc-wg1.ucar.edu/wg1/wg1-report.html specifically says that > "Most of the observed increase in globally averaged temperatures since > the mid-20th century is very likely due to the observed increase in > anthropogenic greenhouse gas concentrations." And also says: "In this > Summary for Policymakers, the following terms have been used to indicate > the assessed likelihood, using expert judgment, of an outcome or a > result: Virtually certain > 99% probability of occurrence, Extremely > likely > 95%, Very likely > 90%, Likely > 66%, More likely than not > > 50%, Unlikely < 33%, Very unlikely < 10%, Extremely unlikely < 5%. (See > Box TS.1.1 for more details)." > > Yet Gehard says that is not a statement of "confidence"? I wanted to stay out of here, because what I've said so far has been largely ignored, but since you're citing me (or questioning me) here, I'll repeat what I wrote earlier. First, there is a difference between a probability of a future event, calculated using a model and some input data, and the confidence in that model. What you were citing are the probabilities, not the confidence levels. As I wrote before, the IPCC's report is full of probability figures. (Well, not necessarily the actual figures, but expressions like the ones you cited, which can be translated into figures using their "Uncertainty Guidance Note".) But that same "Uncertainty Guidance Note" also gives expressions to be used for confidence levels, right next to the expressions to be used for probabilities; this is cited from an earlier post of mine: G> From their "Uncertainty Guidance Note for the Fourth Assessment G> Report" at G> : G> G> -------------------- G> Very High confidence: At least 9 out of 10 chance of being correct G> High confidence: About 8 out of 10 chance G> Medium confidence: About 5 out of 10 chance G> Low confidence: About 2 out of 10 chance G> Very low confidence: Less than 1 out of 10 chance G> G> [...] G> G> Virtually certain: > 99% probability of occurrence G> Very likely: > 90% probability G> Likely: > 66% probability G> About as likely as not: 33 to 66% probability G> Unlikely: < 33% probability G> Very unlikely: < 10% probability G> Exceptionally unlikely: < 1% probability G> -------------------- See, the second set of expressions talks about probabilities, and the first set talks about confidence levels. As you can easily verify, the highest confidence level they think they need a suggested wording for is 90% (that would be Russell's "P=0.1") -- they call it a "very high confidence". There's no suggested wording for 95% confidence (19 out of 20, in their way to put the figures, or "P=0.05" in Russell's way). So back to your question: No, talking about probabilities is /not/ the same as stating the confidence level of those probabilities. Try yourself what I suggested in my post on Oct 06 and search the report for any of the IPCC's suggested confidence level expressions (like I did), and then report whether your findings are different from mine. (To refresh the memory: I didn't find any of them. They just don't seem to mention them, even though they obviously found them important enough to add a set of suggested expressions to their "Uncertainty Guidance Note". I didn't search all parts of the report in all versions, though, so I may have missed some. But so far /nobody/ has cited /any/ confidence level from an IPCC report.) I asked then, and still wonder, "why don't they use the suggested confidence level expressions to go along with the suggested probability expressions"? A probability (as result of a calculation using a model and input data) without a confidence level (regarding the model and the input data) is not worth much. So whenever you talk about such probabilities, you need to state the confidence level to make any sense. This as in "Using my new model X, it is 'very likely' that it will rain where you are when you read this post. The confidence in the model is 'very low', though, considering past data." The first sentence doesn't really give any information to assess the quality of the model, just the result when applying it. It is the second sentence that talks about the quality of the model. Applying this to the issue at hand, I don't really doubt that they get to a probability of 97.5% of something happening when using model Y, but I'd like to know more about the quality of the model -- which would be expressed as confidence level. /That's/ the issue I was talking about (and if I understood correctly, Russell also). And nobody has brought forward any mention of any confidence data in the IPCC's report, nor any explanation why they don't state the confidence level together with the probabilities. (They themselves seem to think that this is something that should or could be done, or else they wouldn't have a set of suggested expressions for confidence intervals in their Uncertainty Guidance Notes.) [We're not used to this, from our work, because the confidence level of scientific models we're working with is much higher than of the data we're working with, so we just stay on the side of the data with our considerations. But once you accept confidence levels for your models below 99.9% (and according to their suggested wording, they mostly work below that level), you really need to start using confidence levels for the models, too.] > And Russell can not back up his claim that the IPCC have said in the past > that they were less confident. Please try to find (and cite here) a single statement of confidence level (NOT probability!) in their report (or, better yet, one that supports that a certain conclusion has a confidence level above 90% or 95%). > And in the mean time, is it such a bad idea to do what we can to reduce our > footprint on the earth? If we can error on the side of caution, why not? Now this is a completely different question, and I think you know I agree with you on this. But this doesn't put the confidence data into the IPCC's report. Gerhard -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist -- http://www.piclist.com PIC/SX FAQ & list archive View/change your membership options at http://mailman.mit.edu/mailman/listinfo/piclist