3 ms·

Er, no, he doesn't say that the main climate model is useless, or that "it's rubbish and doesn't work". He doesn't say anything even slightly like that. The par

by gjm11 17y ago

Er, no, he doesn't say that the main climate model is useless, or that "it's rubbish and doesn't work". He doesn't say anything even slightly like that. The particular thing he's saying is "meaningless" is a figure that has basically nothing to do with actual climate simulation. Clearly he's not impressed with the quality of the code. Fair enough. There's a lot of very poor code around doing important things. It looks to me as if after all Ian's cleanups, he ends up with code that produces essentially the same as the pre-cleaned-up code, which suggests that the code-quality issues he found don't invalidate the actual calculations. Another thing that's notable: here we have a lengthy, ill-tempered diary of someone's attempt to wrestle with this big hairy codebase and the big hairy pile of ill-formed data (supplied from the outside world; this isn't all CRU's fault) that it processes, and in all his notes of what he's doing and his interactions with colleagues we don't see anything like "Ran the code again. It's weird -- it doesn't predict any substantial warming at all" or "Joe told me to change X and Y to make the figures look scarier" or "Wait, this bit here is obviously fudged". This despite a clear willingness to moan about his colleagues' work. In other words, this seems like it would be exactly the sort of place where you'd see evidence of conspiracy and dishonesty if there were any, and there is no such evidence there.

1gor 17y ago
No, you are wrong. Read thee whole quote here http://www.cbsnews.com/blogs/2009/11/24/taking_liberties/entry5761180.shtml* http://www.cbsnews.com/blogs/2009/11/24/taking_liberties/ent... and observe that the programmer is convinced the whole database of reporting stations is bogus, and the model has been tweaked multiple times: "Each parameter has a tortuous history of manual and semi-automated interventions that I simply cannot just go back to early versions and run the update prog." also "One thing that's unsettling is that many of the assigned WMo codes for Canadian stations do not return any hits with a web search. Usually the country's met office, or at least the Weather Underground, show up – but for these stations, nothing at all. Makes me wonder if these are long-discontinued, or were even invented somewhere other than Canada!" So much for 'scientific model supported by hard data'.
- gjm11 17y ago
  I read not only the whole quote but the whole (very long) file from which the quote was cherrypicked. Of course the model has been tweaked several times. What the hell else do you think scientists do when they have a simulation of a large complicated system and there's a constant influx of data for seeing how well it works? The fact that they have to work with messy data is hardly their fault.
  - yummyfajitas 17y ago
    The fact that a sum of squares yields negative numbers is their fault. The fact that they can't reproduce their own numbers is also their fault.
mixmax 17y ago
My point was that it seems like the model might be flawed, or at least is poorly written and/or documented. As you say, this Ian guy is clearly not impressed with the code. This qoute describes it rather well: Ulp! I am seriously close to giving up, again. The history of this is so complex that I can't get far enough into it before by head hurts and I have to stop. Each parameter has a tortuous history of manual and semi-automated interventions that I simply cannot just go back to early versions and run the update prog. I could be throwing away all kinds of corrections - to lat/lons, to WMOs (yes!), and more. So what the hell can I do about all these duplicate stations?... The problem I see is not something as dire as a conspiracy, but the fact that this is the foundation of the western worlds policy on global warming and climate change. I think we all know how hard a model like that must be to write and maintain, and it seems like there's a good chance it doesn't give reliable results. I would say that it is a huge problem that multibillion dollar decisions are based on a model that may or may not be accurate, and that maybe we should start the discussion on climate change by putting more money into actually finding out where we stand by creating good peer-reviewed models with open data. As it stands now it seems like we might be flying blind. Sorry if I didn't make myself clear.
- gjm11 17y ago
  Whatever makes you think there is only one global climate model?
miked 17y ago
Er, no, he doesn't say that the main climate model is useless, or that "it's rubbish and doesn't work". He doesn't say anything even slightly like that. It doesn't make any difference what judgements he pronounces. What matters most is his factual observations. At one point he notes that a sum of squares routine sometimes produces a negative result(!). He also notes that they are not able to reproduce their earlier results using the model. Reproducibility is one of the foundations of science. Not only can others not reproduce their results, they can't even reproduce their own results. This raises a further question that I haven't seen mentioned: was this code and data under version control? And was that lack of version control considered a bug or a feature? We are talking about billions of dollars being spent based on models that apparently had no source control . If they did, they should be able to reproduce.

by gjm11 17y ago

1gor 17y ago

No, you are wrong. Read thee whole quote here http://www.cbsnews.com/blogs/2009/11/24/taking_liberties/entry5761180.shtml* http://www.cbsnews.com/blogs/2009/11/24/taking_liberties/ent... and observe that the programmer is convinced the whole database of reporting stations is bogus, and the model has been tweaked multiple times: "Each parameter has a tortuous history of manual and semi-automated interventions that I simply cannot just go back to early versions and run the update prog." also "One thing that's unsettling is that many of the assigned WMo codes for Canadian stations do not return any hits with a web search. Usually the country's met office, or at least the Weather Underground, show up – but for these stations, nothing at all. Makes me wonder if these are long-discontinued, or were even invented somewhere other than Canada!" So much for 'scientific model supported by hard data'.

gjm11 17y ago
I read not only the whole quote but the whole (very long) file from which the quote was cherrypicked. Of course the model has been tweaked several times. What the hell else do you think scientists do when they have a simulation of a large complicated system and there's a constant influx of data for seeing how well it works? The fact that they have to work with messy data is hardly their fault.
- yummyfajitas 17y ago
  The fact that a sum of squares yields negative numbers is their fault. The fact that they can't reproduce their own numbers is also their fault.

mixmax 17y ago

My point was that it seems like the model might be flawed, or at least is poorly written and/or documented. As you say, this Ian guy is clearly not impressed with the code. This qoute describes it rather well: Ulp! I am seriously close to giving up, again. The history of this is so complex that I can't get far enough into it before by head hurts and I have to stop. Each parameter has a tortuous history of manual and semi-automated interventions that I simply cannot just go back to early versions and run the update prog. I could be throwing away all kinds of corrections - to lat/lons, to WMOs (yes!), and more. So what the hell can I do about all these duplicate stations?... The problem I see is not something as dire as a conspiracy, but the fact that this is the foundation of the western worlds policy on global warming and climate change. I think we all know how hard a model like that must be to write and maintain, and it seems like there's a good chance it doesn't give reliable results. I would say that it is a huge problem that multibillion dollar decisions are based on a model that may or may not be accurate, and that maybe we should start the discussion on climate change by putting more money into actually finding out where we stand by creating good peer-reviewed models with open data. As it stands now it seems like we might be flying blind. Sorry if I didn't make myself clear.

gjm11 17y ago
Whatever makes you think there is only one global climate model?

miked 17y ago

Er, no, he doesn't say that the main climate model is useless, or that "it's rubbish and doesn't work". He doesn't say anything even slightly like that. It doesn't make any difference what judgements he pronounces. What matters most is his factual observations. At one point he notes that a sum of squares routine sometimes produces a negative result(!). He also notes that they are not able to reproduce their earlier results using the model. Reproducibility is one of the foundations of science. Not only can others not reproduce their results, they can't even reproduce their own results. This raises a further question that I haven't seen mentioned: was this code and data under version control? And was that lack of version control considered a bug or a feature? We are talking about billions of dollars being spent based on models that apparently had no source control . If they did, they should be able to reproduce.