When Computer Models Are Treated Like Reality

On April 28, 2004, the SEC made a significant change in policy in the regulation of large investment banks.  On that day, they "decided to allow the five largest US investment banks to substitute advanced mathematical risk models for traditional capital requirements."  Al Gore has touted the "success" of such models as a reason to feel confident that computer models can accurately predict long-term climate trends.

But it turns out, as everyone is discovering this week, that computer models are not reality.  In fact, computer models are extraordinarily sensitive to their inputs, and small changes in their inputs, or the narrowing of models to ignore certain factors, can make them worse than useless.  Computer models are also very easy to force to a preferred conclusion.

  • RPJ

    Al Gore has touted the “success” of such models as a reason to feel confident that computer models can accurately predict long-term climate trends. – you made this claim before. I asked for your source and received no reply. In the absence of evidence to the contrary, I believe you have simply made this up.

  • Mike

    Well, I have a model that’s totally insensitive to small (or even large) changes in input. It simply reads whatever it’s given and prints “42.” Not that far, really, from what most climate models do, but much more efficient – it’s O(n) in the size of the input. I’m working to reduce it to O(1). 🙂

  • It’s OT so feel free to delete this.

    Climate audit found some big news regarding the latest hockey stick. The same data was used in a bunch of hockey stick papers.

    http://noconsensus.wordpress.com/2008/10/03/the-hockey-stick-data-hoax/

    I just thought you might want to know.

  • cfdman

    From a transcript of Al’s stupid movie, for RPJ.

    BTW RPJ, why don’t you just search the damn internet yourself instead of predending demanding proof is an argument.

    “It’s Natural!
    So the temperature increases are taking place all over the world, including in the
    ocean. This is the natural range of variability for temperature in the ocean. You know
    people say, “Aw, it just naturally go up and down, so don’t worry about it.” This is the
    range that would be expected over the last 60 years. But the scientists that specialize
    in global warming have computer models that long ago predicted this range of
    temperature increase.”

    In this quote, you will notice that Al is definitely touting the success of computer models to make predictions.

  • RPJ

    Yeah. Nothing about financial models in there. Clearly you didn’t understand what was being claimed.

  • Alan

    I’ve been programming computers for over 30 years now. Everything you get out of a computer model you put into it either as software or data. You may discover some correlations in the transformed data but it was always there to start with. The high CO2 sensitivity that the GW modellers claim to have discovered from the GCMs was put into it in the first place. These incomplete and grossly simplified models are useless.

  • Patrick

    Alan – you obviously have no idea, actually, about computer models. Your claims bear no resemblance whatsoever to the truth.

  • cfdman

    RJP,

    It is not necessary for Al to have actually verbally made the connection between weather models and financial models. Most if not all of the weather models are created by companies that do both (and did the financial models first BTW). For example, the quote below if from aer and is a quote from their weather forcasting model brochure:

    “Seasonal Weather Forecasts that Make a Difference.

    AER’s sCast is the most comprehensive seasonal forecasting service in the industry. sCast provides highly skilled regional HDD and CDD forecasts that are population weighted for electricity, gas and oil for energy and weather derivatives traders. This six month seasonal weather forecast product also provides extensive temperature and precipitation information in both tabular and map formats, making these seasonal forecasts easy to understand.

    Based on over a decade of extensive atmospheric research, AER’s seasonal forecast products have been used by the financial services industry for years. Unlike other seasonal weather forecasts, our forecast uses a unique initialization scheme that combines El Niño and warming trends, as well as, snow cover in remote parts of the world to predict overall global trends.

    Over thirty-four years of extensive seasonal hindcasting data has validated sCast’s suite of seasonal weather forecasts to accurately predict monthly mean conditions over 70% of the time.

    Proven science with proven results – sCast.”

    RJP ; Even if Al is completely unaware of this connection – touting the greatness of these models is in fact touting the greatness of models based on financial models that are used to calculate risk (the very type that has recently failed the world). And I agree that I did not completely make this connection with my first post – my bad. But again I maintain that asking for proof is NOT an argument. Neither is claiming that a reply is inadequate (although it was). You are wrong on this one.

  • John Galt

    The first thing I learned about computer programming was GIGO – Garbage In, Garbage Out. That applies oh so very well to our computer climate models.

    First, we don’t know all the factors that affect the climate. Next, we don’t know how all those factors interact and their relative weights. We don’t know how those interactions change dynamically as conditions change.

    What we do know is our computer climate models just use estimates for all those factors.

  • RPJ

    It is not necessary for Al to have actually verbally made the connection between weather models and financial models

    Actually, cdfman, it is necessary, if a sentence like Al Gore has touted the “success” of such models as a reason to feel confident that computer models can accurately predict long-term climate trends. is not to be a simple straightforward lie.

    And we are talking about climate models, not weather models.

    John Galt – how do you understand the climate, if not through models?

  • cfdman

    RJP said: “And we are talking about climate models, not weather models”

    According to dictionary.com:

    Weather: 1. the state of the atmosphere with respect to wind, temperature, cloudiness, moisture, pressure, etc.

    Climate: 1. the composite or generally prevailing weather conditions of a region, as temperature, air pressure, humidity, precipitation, sunshine, cloudiness, and winds, throughout the year, averaged over a series of years.

    So, you think they are different because one is a longer time frame?

  • RPJ

    Why do you think we have different words for them?

  • cfdman

    Maybe its that people who like more than one word for something are poorly educated? Like inflamable and flamable, two words that mean the same thing. I only ever hear ill-informed folks use inflamable.

    Using many large words makes many sensitive people feel good about themselves. I am ok with that I guess.

    Weather over a long period (whatever that is) is climate, climate over a short period (what ever that is) is weather. You will just have to get used to it.

    Thats the facts, man.

  • John Galt

    RJP Wrote:

    John Galt – how do you understand the climate, if not through models?

    Yes, computer models are useful in understanding the climate. However, the models are severely limited because our understanding of the climate system is limited and evolving. The accuracy of the models cannot exceed our understanding of the climate. The accuracy of the models is also limited by computer processing power.

    Science must be verifiable. Climate models must be verified against real-world climate measurements. Climate models which create scenarios which span decades cannot be verified for decades. Anything that is not verifiable is not science.

  • RPJ

    cdfman – not knowing that climate and weather are very distinct concepts is something most people would describe as ‘poorly educated’ and ‘ill-informed’.

    John Galt – are you saying that anything that can’t be verified right now is not science? Because if you are, you’re totally incorrect.

  • John Galt

    John Galt – are you saying that anything that can’t be verified right now is not science? Because if you are, you’re totally incorrect.

    Sure, it can be a valid hypothesis or theory and not be verifiable until a later date. However, unverified is still unverified, not matter how valid it may seem. That doesn’t contradict my previous statement.

    You seem to have a problem accepting the difference between a model and reality. Climate models are only valuable to the extent that they accurately portray reality.

  • OldManRivers

    Some of the (part) quotes I picked up here:-
    >>how do you understand the climate, if not through models?
    >>The first thing I learned about computer programming was GIGO – Garbage In, Garbage Out.
    >>computer models are useful in understanding the climate
    Relevant, possibly, but my favourite is:
    It simply reads whatever it’s given and prints “42”

    Yes, I’d argue about the quotation marks but not for too long. The late-lamented Douglas Adams told us that what mattered was the question- not the answer. Understanding! Understandable! But Meaningless when the answer diverges widely between the 8th and 9th decimal place.

    I sympathise with the GIGO standpoint and agree, with reservations, about its translation as ‘Garbage In, Garbage out’

    May I offer a new partial translation- Gore In ….

  • RPJ

    unverified is still unverified, not matter how valid it may seem. That doesn’t contradict my previous statement. – yes, it does contradict your previous statement. Unverified != unverifiable. Your final sentence is mere insult and offers nothing to the discussion.

  • cfdman

    RJP said:

    “cdfman – not knowing that climate and weather are very distinct concepts is something most people would describe as ‘poorly educated’ and ‘ill-informed’.”

    Are you arguing that that an ill-defined time scale difference between the two translates into “very distinct”? The dictionary uses the same words to define each, and even uses “weather” as the definition of climate. It seems that even providing quotes from the dictionary does not stop you from arguing a loosing point and throwing out distractions faster than they can be shot from the air.

    It seems you gave up actually reading the comments posted and object without filter. Weather and climate are the same thing, as I said they were. The only difference being the time frame, with climate being of some longer timeframe than weather. I provided the dictionary definitions and you resort to the “everyone knows” argument.

    You are a moron.

  • RPJ

    Weather and climate are the same thing. Ho ho! I do hope you do not typify the level of intellectual ability around here.

  • cfdman

    “Ho ho”

    Santa?

    I’ll chalk that up as a victory.

    thanks for playing.

  • RPJ

    You chalk it up as whatever gives you the most pleasure. I’ll chalk you up as utterly ignorant of basic terminology. Thanks for a good laugh.

  • John Galt

    Your final sentence is mere insult and offers nothing to the discussion.

    Yep. It was meant to be insulting. You’re argumentative just to be argumentative and insist upon taking the debate away from the topic.

  • RPJ

    What an unpleasant bunch of people there are around here. Clearly the idea of friendly discussion is quite alien to you.

    If you are upset about the discussion drifting, perhaps you could do what no-one else has yet been able to, and find any evidence that Al Gore ever said what the original post claimed he said.

  • John Galt

    What an unpleasant bunch of people there are around here. Clearly the idea of friendly discussion is quite alien to you.

    If you are upset about the discussion drifting, perhaps you could do what no-one else has yet been able to, and find any evidence that Al Gore ever said what the original post claimed he said.

    Point taken. Isn’t Al Gore’s book and award-winning infomercial based upon the projections of computer models? You’ll probably have to ask the blogger to explain his assertion as he’s the one who made it.

  • Michael

    RPJ-

    Let’s assume that you are correct, and that Mr. Gore never said any such thing. Let’s just strike the offending sentence from the post.

    Now, any thoughts on projections and computer models?

  • RPJ

    John Galt – I have asked him, twice, and there has been no response. As such, I don’t see a way to avoid calling his statement a lie.

    Michael – if you strike the false claim from the post, you are left with a statement that financial models have failed. That may be so but it tells us absolutely nothing whatsoever about the value of climate models, any more than it tells us anything about the value of models of the Earth’s interior used in geology, models of stars used in astronomy, models of circuits used in race simulators by Formula One teams, or any other kind of computer models.

  • cfdman

    not true, I provided a link between climate models and financial risk models. I also pointed out that Al did not have to even be aware of the link for it to be relevant.

    I knew you did not read my posts….

  • RPJ

    I think you’re a little hard of comprehension, cdfman. The original post claimed that Al Gore had ‘touted the “success” of [financial] models as a reason to feel confident that computer models can accurately predict long-term climate trends‘. Nothing you’ve said provides any backing for that claim.

  • cfdman

    I have already agreed that I was unable to find that Al had claimed to see the conncetion, only that the connection actually exists. It may be (as you have already noted) that Al has never touted the success of financial models. At least I have been unable to find anything that helps with this.

    My link simply shows that climate models are created by modelers who had previously written financial risk models. And that the companies themselves tout the success of their previously written financial models as an argument as to why a users should trust their climate models. I gave up being able to prove that Al was ever aware of this connection or ever said anything about it. I simply showed that it was true regardless of what Al had said.

    I felt this was a more useful avenue of discussion. But you have consistently failed to properly characterize the proof I provided. I pointed out more than a dozen posts ago that Al may have never said anything about fanancial models, but that the connection is still valid without Al’s help or knowledge.

  • RPJ

    You mention a link but you’ve given no links. You quoted some text from a weather forecasting company’s website, and it has not become any clearer with time why you thought this quote showed that there was any kind of link between climate models and financial risk models. The link you seem to think exist doesn’t exist.

  • cfdman

    “You mention a link but you’ve given no links”: I gave you the quote, the company name and the product name. I brought the information directly to this forum becuase you routinely spout crap at others who only provide links. The next time you’ll bitch because you were provided a link and you wanted the information directly. The fact that you never went and looked at “AER”‘s website is not my problem, I brought the relevant info here.

    “You quoted some text from a weather forecasting company’s website, and it has not become any clearer with time why you thought this quote showed that there was any kind of link between climate models and financial risk models” Its not clear becuase your stupid. The company that makes the climate modeling software touts their modeling expertise in the fancial risk modeling field as a reason why you should buy their climate modeling software. It said that right in the quote I provided, and in many other places on their website. If you cant see what this has to do with this post, you are a moron. This thread is based on the notion that Al gore engages in this behavior. But since I have shown that the climate modelers themselves do this, whether or not AL does it is irrelevant.

    You must just be pretending to be stupid and forget what you have been arguing all along.

  • Michael

    RPJ-

    Even without the remark about Al Gore, I think the post still implies that the failure of the financial models casts doubt on the value of all computer models. I’m not so sure that follows.
    And the failure of the financial models may have been more a function of how they were used than a problem with the models themselves.

  • RPJ

    cdfman – you showed no link between climate modellers and financial modellers. aer.com appear not to have ever been responsible for either financial or climate models. You are not making a lot of sense here.

    Michael – I’m not sure it follows either. I’m not sure there is a single discipline in modern science which does not at some stage make use of highly complex computer models, and I find the attempt to slur a science by means of spurious comparison to a non-science quite preposterous. It is like saying that climate models must be wrong because Windows crashes sometimes.

  • cfdman

    http://www.aer.com/products/scast.html

    RJP,

    Please find attached link, that was the first choice by searching Google for “climate model, AER, scast” (as I said it would be).

    I understand how difficult Google is to use, its perfectly understandable that you were unable to find this info.

    In the link, you will find a climate modeling company touting their success in writing fanancial risk models as a reason why you should trust their climate models. Due to the fact that this is the point of this entire thread, I am sure you will see how relevant it is.

    NOT

  • RPJ

    cdfman – as I told you before, this company does not even write climate models, they write weather models. There is a major difference. And they say nothing about having ever written financial risk models. You must be just entirely misunderstanding what they are saying. You should stop beating this dead horse, because you’re really making yourself like a complete idiot.

  • cfdman

    RJP,

    Your protests are like your intellect, vapid and moronic.