I note once again a problem that the sciences are going to have to confront:
When someone's salary is entirely dependent upon not rocking the boat, you can no longer expect them to be objective in what they report--and this is as true of the sciences as it is any other profession.
Computer modeling is wonderful. Good skill to have. Requires hard work.
1. However, extrapolating vast conclusions based on computer modeling, without acknowledging some of the inherent limitations, is a bad idea.
2. Using these vast conclusions as a springboard for wide-reaching public policy proclamations that effect millions of workers and hundreds of industries, is a bad idea squared.
I'm done with global warming. I've concluded it's politicized quackery.
It reminds me of the old days of arguing Evolution on USEnet and having a programmer tell me that he had written a simulation that proves Evolution is a fact.
Models prove nothing. Period. Anyone who doesn't get this, doesn't have a clue.
Models can indicate lots of things that may then be proven or disproven by observation of the reality that is modeled.
My entire career path for the past decade has been about modeling; and the most valuable thing I've learned came from George Box: "Essentially, all models are wrong, but some are useful."
Doubtless those whose salaries continue to be 100% dependent on not-rocking-the-boat will say she's just old and addled, out of touch, "not up on the latest literature," yadda yadda yadda. Anything to avoid addressing her arguments head on.
Buildings are designed using the formulas of structural engineering, which represent a mathematical model of the real structure. Most of them stand up.
Airplanes are designed using mathematical models of structure and aerodynamics; however, they are never placed in commercial service until a physical airplane has been extensively flown by a test pilot.
Collateralized Debt Obligations were extensively analyzed by mathematical models which provided estimates of their safety...estimates which turned out to be very wrong.
Buildings are designed using the formulas of structural engineering, which represent a mathematical model of the real structure. Most of them stand up.
This has not always been true. I remember visiting the great cathedral in Florence. It has something like a 100 foot domed roof and was build ~1300. It took more than 100 years to finish. It was build almost entirely by seat of the pants engineering. They build cathedrals with domed roofs. Some of them fell down and some didn't. Over hundreds of years they learned what worked and what didn't work. The ones that are still standing today, are the ones that worked.
Do we want to do seat of the pants global climate change studies?
Anyone who thinks a computer model of a relatively simple system such as an architectural design is even in the same ballpark as the complexity required for a model of the entire Earth's climate should no longer be listened to on the subject of global warming.
The hardest computer programs to write deal with the physical world.
Buildings are designed using the formulas of structural engineering, which represent a mathematical model of the real structure. Most of them stand up.
The Brooklyn Bridge was built six times stronger than the models of the day required, because the designer knew there were things he didn't know. It is one of the few suspension bridges of it's time still standing. The Tacoma Narrows bridge on the other, aka Galloping Gertie, shook itself to pieces in a high wind.
Anyone who thinks a computer model of a relatively simple system such as an architectural design is even in the same ballpark as the complexity required for a model of the entire Earth's climate should no longer be listened to on the subject of global warming.
John, I think you misinterpreted Photon Courier. He listed three examples of applied modeling. One example -- models that have been developed over centuries of experience and experiment -- is reliable enough that we mostly treat the models as facts. One example -- models that are mostly a century old or less -- is applied every day, but we still don't trust it until we test the end results. And one example -- models that I'll bet are decades old at most, and probably less -- led to some really nasty fiscal collapses. I saw his comment as a demonstration that all models are wrong, but some models are useful.
Tom...I'm not sure the hardest models deal with the physical world. When you add human behavior, you are not only dealing with feedback loops, but with feedback loops that invent new feedback loops.
True. I should have said "The hardest programs I have had to write deal with the physical world." Or perhaps used the words "real world", instead of "physical world". Web pages and business data (not models, accounting, etc.) are easy, easy, easy - and look how badly we do them!
Actually, when it comes to predictions of global warming, they're incredibly robust. Incredibly.
When the global cooling scare of the 1970s wound down, it was immediately replaced by global warming. The National Academy of Sciences appointed a committee to report on the new threat of global warming. Based on the crude first-cut computer models of the day, the committee said that a doubling of CO2 would cause a warming of 1.5C to 4.5C. (If you're old enough, remember what computers were like in the late 1970s.)
Tremendous progress was made in our understanding of the atmosphere and in computer science. The hot summer of 1988 made global warming a public issue. The UN set up the Intergovernmental Panel on Climate Change (IPCC). The IPCC published its First Assessment Report (FAR). The FAR, based on the greatly improved computer models of the day, said that a doubling of CO2 would cause a warming of 1.5C to 4.5C.
All of those models assumed that little solar energy is absorbed in clouds. In 1995, the amount of solar energy absorbed in clouds was measured for the first time. It turned out to be 600% larger than the models had assumed. Science, the journal of the American Association for the Advancement of Science, noted that predictions of warming had not improved as much as one might have expected, but said that this new data would finally create "a brighter future for climate models". Later Assessment Reports were based on upgraded models with the correct value for sunlight absorbed by clouds. 1.5C to 4.5C.
The energy absorbed and emitted by atmospheric gases is calculated using data from the HITRAN database. The 1996 edition of HITRAN had a clerical error which was as large as doubling CO2, that is, the error was as large as the whole global warming problem. The climate modelers never noticed. The error was discovered after a couple of years by a contributor to HITRAN and corrected. The climate modelers never noticed. Subsequently, new measurements of water vapor caused the calculated absorption by H2O to change by about 95% of a doubling of CO2. 1.5C to 4.5C.
The IPCC published the Fourth Assessment Report last year. Based on the latest modeling, taking advantage of the progress funded by $4.5 billion per year for climate change research from the US government alone, the IPCC 4AR said that a doubling of CO2 would cause a warming of 2C to 4.5C, immediately followed by the statement, "Values less than 1.5C are highly unlikely."
You can't change the output of these models by changing the inputs. You can't even change the output by changing the models. "Frail"? Like a freakin' block of granite.
Hmm... this sounds exactly like what I've been saying for years...
Say, anybody notice that 2007 was .75 degrees colder than 2006, the largest temperature change in one year EVER?
By the way, NASA corrected their data recently (back in August) and it turns out 1998 is NOT the hottest year on record, and the top ten hottest years are NOT in the past ten years.
[gather data]
[analyze data]
[model a system]
[throw the results away]
printf("Doubling CO2 levels will cause temperatures to rise 1.5 C to 4 C\n");
... that does not constitute a robust model. That qualifies as "fixed output".
3.1.2008 4:18am
Commenting on Dean's World is a privilege, not a right. Dean is your host, you are his guest, and you should behave in that fashion. Dean is not your babysitter, nor is he your punching bag. Please remember this. In general, you are free to disagree with anyone on any subject you wish, but abusive behavior will not be tolerated.
Of course we all lose our tempers now and then. Dean freely admits to being imperfect in this regard, which is why regulars to this establishment will generally be cut more slack than people who we don't know very well.
Still: behave like an adult, or go find somewhere else to play. Thanks.
Shame on you, saying that on the internet!
When someone's salary is entirely dependent upon not rocking the boat, you can no longer expect them to be objective in what they report--and this is as true of the sciences as it is any other profession.
1. However, extrapolating vast conclusions based on computer modeling, without acknowledging some of the inherent limitations, is a bad idea.
2. Using these vast conclusions as a springboard for wide-reaching public policy proclamations that effect millions of workers and hundreds of industries, is a bad idea squared.
I'm done with global warming. I've concluded it's politicized quackery.
HankB
I pointed him to a simulation that proves that zombies are going to take over the earth.
Models can indicate lots of things that may then be proven or disproven by observation of the reality that is modeled.
My entire career path for the past decade has been about modeling; and the most valuable thing I've learned came from George Box: "Essentially, all models are wrong, but some are useful."
Never mistake the model for the reality.
Doubtless those whose salaries continue to be 100% dependent on not-rocking-the-boat will say she's just old and addled, out of touch, "not up on the latest literature," yadda yadda yadda. Anything to avoid addressing her arguments head on.
Airplanes are designed using mathematical models of structure and aerodynamics; however, they are never placed in commercial service until a physical airplane has been extensively flown by a test pilot.
Collateralized Debt Obligations were extensively analyzed by mathematical models which provided estimates of their safety...estimates which turned out to be very wrong.
A man who has learned much wisdom.
And like most wise men, probably the hard way...
This has not always been true. I remember visiting the great cathedral in Florence. It has something like a 100 foot domed roof and was build ~1300. It took more than 100 years to finish. It was build almost entirely by seat of the pants engineering. They build cathedrals with domed roofs. Some of them fell down and some didn't. Over hundreds of years they learned what worked and what didn't work. The ones that are still standing today, are the ones that worked.
Do we want to do seat of the pants global climate change studies?
They are ignorant of their own ignorances.
Buildings are designed using the formulas of structural engineering, which represent a mathematical model of the real structure. Most of them stand up.
The Brooklyn Bridge was built six times stronger than the models of the day required, because the designer knew there were things he didn't know. It is one of the few suspension bridges of it's time still standing. The Tacoma Narrows bridge on the other, aka Galloping Gertie, shook itself to pieces in a high wind.
Yours,
Wince
John, I think you misinterpreted Photon Courier. He listed three examples of applied modeling. One example -- models that have been developed over centuries of experience and experiment -- is reliable enough that we mostly treat the models as facts. One example -- models that are mostly a century old or less -- is applied every day, but we still don't trust it until we test the end results. And one example -- models that I'll bet are decades old at most, and probably less -- led to some really nasty fiscal collapses. I saw his comment as a demonstration that all models are wrong, but some models are useful.
Tom...I'm not sure the hardest models deal with the physical world. When you add human behavior, you are not only dealing with feedback loops, but with feedback loops that invent new feedback loops.
True. I should have said "The hardest programs I have had to write deal with the physical world." Or perhaps used the words "real world", instead of "physical world". Web pages and business data (not models, accounting, etc.) are easy, easy, easy - and look how badly we do them!
Yours,
Wince
Actually, when it comes to predictions of global warming, they're incredibly robust. Incredibly.
When the global cooling scare of the 1970s wound down, it was immediately replaced by global warming. The National Academy of Sciences appointed a committee to report on the new threat of global warming. Based on the crude first-cut computer models of the day, the committee said that a doubling of CO2 would cause a warming of 1.5C to 4.5C. (If you're old enough, remember what computers were like in the late 1970s.)
Tremendous progress was made in our understanding of the atmosphere and in computer science. The hot summer of 1988 made global warming a public issue. The UN set up the Intergovernmental Panel on Climate Change (IPCC). The IPCC published its First Assessment Report (FAR). The FAR, based on the greatly improved computer models of the day, said that a doubling of CO2 would cause a warming of 1.5C to 4.5C.
All of those models assumed that little solar energy is absorbed in clouds. In 1995, the amount of solar energy absorbed in clouds was measured for the first time. It turned out to be 600% larger than the models had assumed. Science, the journal of the American Association for the Advancement of Science, noted that predictions of warming had not improved as much as one might have expected, but said that this new data would finally create "a brighter future for climate models". Later Assessment Reports were based on upgraded models with the correct value for sunlight absorbed by clouds. 1.5C to 4.5C.
The energy absorbed and emitted by atmospheric gases is calculated using data from the HITRAN database. The 1996 edition of HITRAN had a clerical error which was as large as doubling CO2, that is, the error was as large as the whole global warming problem. The climate modelers never noticed. The error was discovered after a couple of years by a contributor to HITRAN and corrected. The climate modelers never noticed. Subsequently, new measurements of water vapor caused the calculated absorption by H2O to change by about 95% of a doubling of CO2. 1.5C to 4.5C.
The IPCC published the Fourth Assessment Report last year. Based on the latest modeling, taking advantage of the progress funded by $4.5 billion per year for climate change research from the US government alone, the IPCC 4AR said that a doubling of CO2 would cause a warming of 2C to 4.5C, immediately followed by the statement, "Values less than 1.5C are highly unlikely."
You can't change the output of these models by changing the inputs. You can't even change the output by changing the models. "Frail"? Like a freakin' block of granite.
Say, anybody notice that 2007 was .75 degrees colder than 2006, the largest temperature change in one year EVER?
By the way, NASA corrected their data recently (back in August) and it turns out 1998 is NOT the hottest year on record, and the top ten hottest years are NOT in the past ten years.
Funny how that doesn't get reported...
[analyze data]
[model a system]
[throw the results away]
printf("Doubling CO2 levels will cause temperatures to rise 1.5 C to 4 C\n");
... that does not constitute a robust model. That qualifies as "fixed output".
Of course we all lose our tempers now and then. Dean freely admits to being imperfect in this regard, which is why regulars to this establishment will generally be cut more slack than people who we don't know very well.
Still: behave like an adult, or go find somewhere else to play. Thanks.