Maybe the greatest threat is computer modelers, instead.
In second place, perhaps, we have half-educated (at best) politicians who get wide-eyed over false-color computer graphics and physically meaningless statistical plots.
Philip Mulholland
September 18, 2020 1:52 pm
Even now I remember the arguments from years back about modeling (and not just about the spelling). Was it even a science or just a profoundly clever way of fooling yourself and others. I think we now have the definitive answer.
I can remember reading all the commented code in harry_read_me ten years ago. What an eye-opener – yet not now remarked even though some of the current modellers still use code as old as harry’s.
The so-called main stream media is right in there.
Old.George
September 18, 2020 2:29 pm
Having done computer modeling for a major pharma company in epidemiology, I made one for COVID-19.
Early on, when the data was sparse and all I had high numbers projected. As data changed, I adjusted. My latest projection is 250,000 +/- 10,000 total US deaths and herd immunity by Halloween +/- 2 weeks.. Among things ignored are mask-use, more effective treatment, and vaccine.
Doesn’t the wearing of masks, if it is as effective in minimizing the spread of COVID-19 as is claimed, necessarily fight against humans ever acquiring “herd immunity” against COVID-19?
Or does the reconciliation of these two issue require advanced tensor analysis that is only available to Dr. Fauci and a few other “experts” in the field?
“Or does the reconciliation of these two issue require advanced tensor analysis that is only available to Dr. Fauci and a few other “experts” in the field?”
The math is straightforward. Small changes in assumptions can lead to drastically different conclusions. Masks may have an effect on timing (lengthening the epidemic locally), but appear to have no effect on IFR. Where HCQ+ protocol was used the IFR appears lower (not enough data to factor in treatments.)
For example, if I were modeling influenza I would use a 1-day infectivity and masking or self-isolation of the sick.
We could simulate 1-day infectivity by doing $1 at-home tests which are sensitive enough to identify an infective person on day 2 and sometimes on day 1. Only those not testing negative today (or skipped testing) should wear a mask. It would shorten the time to herd immunity and to two weeks from implementation (see MedCram — 100% compliance not needed.) Lockdown could end immediately.
I’m both a computer scientist and a musician. Tuning a model is like tuning a guitar. The process is never finished and never completely satisfactory. Arlo Guthrie said that once you take LSD you realize there’s no such thing as “in tune”.
Face masks only slow the time to Herd Immunity. They do not reduce the area under the curve. The only thing that will reduce the area under the curve is a vaccine as a substitute for Herd Immunity via natural infection.
Hmmm . . . by and large, humanity has not been wearing face masks for control/protection against the spread and for limiting infection rates of the seasonal flu, which has been at epidemic/pandemic levels for over 100 years.
Why doesn’t humanity have herd immunity for the seasonal flu?
Why don’t we have immunity to the seasonal flu? Mutation.
Each year we vaccinate against the top 3 from last season. This vaccination provides herd immunity of the vaccine kind. Most die out, but there is an occasional mutation which takes a different antibody. The top 3 of these are vaccinated for the next year.
Compare with measles. Very little mutation. One vaccination to get your system to generate antibodies and done.
OK. Everything I’ve read says COVID-19 is a virus that is much more similar to the flu than measles.
Therefore, one might reasonably expect that a vaccine against COVID-19 and its probable mutations will never be equivalent to (aka a substitute for) “herd immunity”. And that herd immunity against COVID-19 is pretty unlikely anyway.
I have some questions. How do you represent the efficacy of mask use in a model? Is it anything other than a parameter with an assumed impact? Could such parameter not be labelled something else, say headphone use?
I have a hard time seeing, in light of the variety of masks and the lack of studies finding any clear benefit in infection control from even medical masks, how mask use could be accurately represented in an epidemiological model.
“How do you represent the efficacy of mask use in a model?”
As I mentioned I didn’t. It would be reflected in a lower R(t). A single parameter multiplying the R(t) without masks by, say, 0.98 could be put in. A more complex multiplier might try to include levels of PPE from full Level 4 (positive pressure suits), multiplier 0.00001, to bandana at 0.99999, and no mask at 1.0. Compliance data at the various levels are unavailable.
R(0) — the uncontrolled rate — becomes R(t) at time t. It must be < 1 for the virus to die out. See Rt.live.
boffin77
September 18, 2020 2:35 pm
Yeah . . . No.
I suspect there is no significant threat to Mankind / Humankind. Of more concern are the threats to the roots of our civilization, the values we treasure, and the ideals we aspire to.
CheshireRed
September 18, 2020 2:37 pm
Climate activist – ‘Make me a ‘climate change’ model projecting a catastrophic future’.
What kind of person do you think I am?
We have already established that. Now we’re just haggling about the price.
High Treason
September 18, 2020 2:50 pm
Lies, damned lies and statistics. Modelling is only as good as the inputs. If just one of input is out, the output should be in the bin. The error will be multiplied through the system until the output is useless.
With about 47 variables involved with climate ,with effects varying depending upon how they react with the others, things become so complicated that it is essentially impossible to model. Thus far, the models only come even remotely close when the data are manipulated to match the models (fraud.)
For a start, the weightings for each variable are guesses at best and the most important variable, solar output, is largely ignored.
As for the COVID modelling, it has already been shown to be badly flawed-out by a factor of a thousand, with death reports ascribed to be from COVID being plumped up with deaths with COVID. There are also cases of what are essentially bribes being paid to put COVID as the cause of death to further plump up the figures.
The cAGW and COVID scams come from the same fearmongering stable. Both of the scenarios are pretty hard to swallow if you look at them objectively. Both see the old conman trick of brushing off dissenting views and questioning. Both scams pretend to be real science-the appeal to authority. The science is just opinion for hire. Both see retribution and persecution of dissenting views. To quote Voltaire-“It is clear that an individual who persecutes a man, his brother because he is not of the same opinion is a monster.”
Here in Australia, Dictator Dan in Victoristan is proposing a Bill that calls for the arrest and indefinite detention for those suspected they might break curfew. This is to be conducted by police and public servants. This is as draconian as any despot in history has ever done. As it is, HCQ is banned, based upon one ridiculously flawed study using known to be dangerous doses. Those researchers obviously did not bother looking at the known LD50 data which has been known for decades. Doctors are essentially being forced to violate their Hippocratic Oath on the back of this study. Note that Dictator Dan was Victorian health minister for 3 years.
It is clear that Daniel Andrews is a tyrant. It will be interesting when the vote is put to the Parliament. Will the rest of the “leaders” follow the edict of the tyrant? I would like to know where the Prime Minister is.
The most disturbing thing is that there are a lot of sheeple who hail Dictator Dan as a strong leader. A true strong leader works within the rules to achieve an outcome. A dictator rides roughshod over the laws, rules, Constitution and human rights to achieve their outcome. The differences between a real strong leader and a fake strong leader(dictator) could not be any clearer.
Peter W
September 18, 2020 3:16 pm
My observation on the IPCC for some time has been as follows:
The United Nations is a political organization, not a scientific one. The IPCC was created by, and is controlled by, the United Nations. It provides the best science that money can buy – i.e., pay some scientists enough money, and they will find a way to tell you whatever it is you want to hear.
The first lpcc report from the scientists found on manmade climate change, the reports released to the public are written by the policy makers who then changed them to suit their goals.
Crispin in Waterloo
September 18, 2020 3:20 pm
The most clear and present danger is still nuclear war, even if it is over in one day.
Nuclear weapons have their own “Big Nuke” industrial complex. It remains more dangerous than ISIS and Iran put together.
Even a limited exchange of nuclear weapons makes climate change just spit in the ocean. The Greens should attack the UN for failing to end proliferation.
Without nuclear weapons, how many people would had died in World War III. Europe were fighting a major continent war every twenty years. You can bet without nuclear weapon is would have happen. Losses would been al least 200,000,000. Nuclear weapons may have been what saves us.
Philo
September 18, 2020 3:34 pm
If there were no computer modellers there would be no computer models.
Many talented programmers have shown that the weather is very unlikely to be predicted more than about 7-10 days at the outside. There are too many variables and too many unknowns as to how they interact.
Climate is the average weather.
Hence, no one can predict the climate more than a few days ahead.
Climate models and modellers are strictly POLITICS as usual, not science.
consistent weather is easy to predict. Where I live summer is very consistent.
The Fall & Spring turn, for a short period of time, are very chaotic & patterns not consistent. Winter time is generally easy to predict.
In the fall and spring turn a coin flip will get you the appropriate odds that the weather prediction is good two days out.
In the summer and winter a 10-day guess is correct more often than not.
Climate is generally defined by NASA, NOAA and other major climate study organizations as being weather over a specified geographical area averaged over a period of at least 30 years.
Climate is not just “average weather” independent of time.
So are you implying that under the major climate organization’s definition, in the absence of a specified time period, there is no such thing as climate? So the climate of the Mediterranean, for instance, doesn’t exist unless it refers to the weather that has occurred in that region over a specified period of three decades or more? Just describing it as warm dry summers and mild dry winters doesn’t cut the mustard?
More to the point, since the weather over a specified geographical area averaged over a period of 30 years is likely to differ from the weather over the same area averaged over 50 or 100 years, does that mean that today’s climate varies depending on the period over which one chooses to average past weather?
If so, would that make climate an abstract concept, a statistical one, or a purely imaginary one? Would it mean that climate or its derivative concept climate change were incapable of producing any physical effects and that any effects that are attributed to them are actually the effects of weather, which does consist of physical phenomena?
Tim, you posted: “So are you implying that under the major climate organization’s {sic} definition, in the absence of a specified time period, there is no such thing as climate?”
No, I am not. I am stating the generally-accepted method of distinguishing “climate” from “weather”.
Try to find a place on Earth that has not had at least 30 years of weather.
Logic 101.
You also posted: “More to the point, since the weather over a specified geographical area averaged over a period of 30 years is likely to differ from the weather over the same area averaged over 50 or 100 years, does that mean that today’s climate varies depending on the period over which one chooses to average past weather?”
You got it! That is EXACTLY the reason, for example, we can discuss the modern climate of the Mediterranean region and compare it to the climates that existed there during the Little Ice Age, The Medieval Warm Period, and even during the last glacial period.
Understanding 101.
“Climate” is indeed a relatively subjective term. “Climate” has never PRODUCED physical effects . . . it is the product of observations of physical effects over time. This misunderstanding is quite common to CAGW alarmists who want to fight climate change™.
Yes, in very stable climates it’s sometimes possible, but European weather is anything but stable especially in the west where the Atlantic likes to create chaos.
Météo-France talking about their own performance:
“The forecast check shows that the 24 hour weather forecast over France is correct about 90% of the time, and that the 24 hour temperature forecast at a given point has average accuracy of the order of 1 to 1.25 ° C. At 7 days, the accuracy is around 3 ° C.”
….they go on to say…
“Over the past thirty years, the quality of forecasts from Météo-France’s Arpège model has gained one day every ten years, today, four-day forecasts are as reliable as three-day forecasts at the beginning of the 2000s.”
I follow the forecast everyday and the four day forecasts are still very variable in there accuracy.
In real life computer models are useful for developing things like a better bumper or airframe before spending the money to design and build a prototype.
That prototype is tested in real life to see it works before going into production.
If it doesn’t, adjust the model using the new data and try again.
Of course, in ‘Climate Science”, the science was settled since Hansen’s hearing.
California bought it and now they have rolling blackouts.
Yooper
September 18, 2020 3:49 pm
Hang on. Didn’t that infallible religious body the IPCC say, unequivocally, that the climate is a chaotic system that cannot be modeled?
Doesn’t that settle the science?
Bh
September 18, 2020 3:54 pm
There are only two models worth diddly squat. The first is the 65 Shelby Mustang and the other model is often seen on a cat walk, and guess which one is the most reliable.
I disagree: 67-69 Camaro is a much better model than a 65 Shelby Mustang.
I will not comment on your other example because I am aware of the fashion industry and how fleeting that catwalk may be to anyone that does not conform to a designer’s idea of the perfect walking hanger that is typically 7 feet tall. 🙂
woz
September 18, 2020 3:56 pm
The most revealing statistical observation: GIGO! (Garbage in, garbage out, for the uninitiated)!
ScienceABC123
September 18, 2020 5:23 pm
The biggest issue with computer modelling continues to be getting it to agree with observation.
Such models live in a highly Darwinian environment. They work for a short while, until similar models are developed to compete with them whereupon they lose their advantage. Some of the new models exploit weaknesses to become the next short term kings of the market.
eo
September 18, 2020 6:09 pm
“Cross a great river and destroy a great empire”. It is part of human nature to get some advice on how the future will unravel before he takes a major or even minor decision. It used to be bone chips before computer chips, it used to be the crystal ball before the crystal (silicon) , etc. It was the oracle and gypsy ladies before the computer programmer. Anybody prediction on how the future of fortune telling will evolve—artificial intelligence based on the rubbish of the internet?
David Barber
September 18, 2020 6:22 pm
LOL
Ronald Bruce
September 18, 2020 7:43 pm
GIGO garbage in garbage out the first thing you learn in computer programming the second thing is that you can control the results so you get the output you want by manipulation of the coding. See the IPCC modeling and dictator dans super computer results mandating further lockdown. It doesn’t matter what computer you use if you write the code to get a predetermined outcome. None of the IPCC or dictator dans models have come true, the populace is just being duped. Using the term super computer is just an appeal to authority trick.
The models versus reality discrepancy leads to a simple hypothesis that is backed up by the data, and the hypothesis is that very little of late 20th century warming was the result of anthropogenic CO2.
Occam says we ought to be content with that, but that means ultimately that all the people working in climate science are a waste of grant money. And all the windmills, batteries and solar panels are, too.
No, I can’t see the IPCC admitting that, either. Can you?
Will they ever receive their comeuppance? Probably only under the despotic systems they wish to replace our liberty with. They might get eaten last, but they’ll always get eaten.
I work in IT, perhaps that is why I am cynical about models, but the problem stems from the publics view of computers as some kind of miracle machine. A sort of magic box, that is invincible, infallible, and god like.
Oh, and I worked on climateprediction.net by the way. Which was quite amusing, arguing with the rest about the impact of CO2. 🙂
Doug Huffman
September 19, 2020 4:35 am
Edward R. Tufte’s masterwork *The Graphical Display of Quantitative Information* is available for free download at Archive.org.
So true. But perhaps the villain should be computer modelling WITHOUT VALIDATION. Modelling is a valuable tool but only when it is treated as an hypothesis to be compared to what happens int he real world by experiment or observation.
Maybe the greatest threat is computer modelers, instead.
In second place, perhaps, we have half-educated (at best) politicians who get wide-eyed over false-color computer graphics and physically meaningless statistical plots.
Even now I remember the arguments from years back about modeling (and not just about the spelling). Was it even a science or just a profoundly clever way of fooling yourself and others. I think we now have the definitive answer.
I can remember reading all the commented code in harry_read_me ten years ago. What an eye-opener – yet not now remarked even though some of the current modellers still use code as old as harry’s.
Computer modeling?
The so-called main stream media is right in there.
Having done computer modeling for a major pharma company in epidemiology, I made one for COVID-19.
Early on, when the data was sparse and all I had high numbers projected. As data changed, I adjusted. My latest projection is 250,000 +/- 10,000 total US deaths and herd immunity by Halloween +/- 2 weeks.. Among things ignored are mask-use, more effective treatment, and vaccine.
Old.George,
Doesn’t the wearing of masks, if it is as effective in minimizing the spread of COVID-19 as is claimed, necessarily fight against humans ever acquiring “herd immunity” against COVID-19?
Or does the reconciliation of these two issue require advanced tensor analysis that is only available to Dr. Fauci and a few other “experts” in the field?
“Or does the reconciliation of these two issue require advanced tensor analysis that is only available to Dr. Fauci and a few other “experts” in the field?”
The math is straightforward. Small changes in assumptions can lead to drastically different conclusions. Masks may have an effect on timing (lengthening the epidemic locally), but appear to have no effect on IFR. Where HCQ+ protocol was used the IFR appears lower (not enough data to factor in treatments.)
For example, if I were modeling influenza I would use a 1-day infectivity and masking or self-isolation of the sick.
We could simulate 1-day infectivity by doing $1 at-home tests which are sensitive enough to identify an infective person on day 2 and sometimes on day 1. Only those not testing negative today (or skipped testing) should wear a mask. It would shorten the time to herd immunity and to two weeks from implementation (see MedCram — 100% compliance not needed.) Lockdown could end immediately.
I’m both a computer scientist and a musician. Tuning a model is like tuning a guitar. The process is never finished and never completely satisfactory. Arlo Guthrie said that once you take LSD you realize there’s no such thing as “in tune”.
Bloody hell. A well considered and cogent argument. I expect it at WUWT but still, it’s remarkable in the 21st century
Face masks only slow the time to Herd Immunity. They do not reduce the area under the curve. The only thing that will reduce the area under the curve is a vaccine as a substitute for Herd Immunity via natural infection.
Or a mutation of the virus.
Hmmm . . . by and large, humanity has not been wearing face masks for control/protection against the spread and for limiting infection rates of the seasonal flu, which has been at epidemic/pandemic levels for over 100 years.
Why doesn’t humanity have herd immunity for the seasonal flu?
Why don’t we have immunity to the seasonal flu? Mutation.
Each year we vaccinate against the top 3 from last season. This vaccination provides herd immunity of the vaccine kind. Most die out, but there is an occasional mutation which takes a different antibody. The top 3 of these are vaccinated for the next year.
Compare with measles. Very little mutation. One vaccination to get your system to generate antibodies and done.
OK. Everything I’ve read says COVID-19 is a virus that is much more similar to the flu than measles.
Therefore, one might reasonably expect that a vaccine against COVID-19 and its probable mutations will never be equivalent to (aka a substitute for) “herd immunity”. And that herd immunity against COVID-19 is pretty unlikely anyway.
Thanks.
I have some questions. How do you represent the efficacy of mask use in a model? Is it anything other than a parameter with an assumed impact? Could such parameter not be labelled something else, say headphone use?
I have a hard time seeing, in light of the variety of masks and the lack of studies finding any clear benefit in infection control from even medical masks, how mask use could be accurately represented in an epidemiological model.
“How do you represent the efficacy of mask use in a model?”
As I mentioned I didn’t. It would be reflected in a lower R(t). A single parameter multiplying the R(t) without masks by, say, 0.98 could be put in. A more complex multiplier might try to include levels of PPE from full Level 4 (positive pressure suits), multiplier 0.00001, to bandana at 0.99999, and no mask at 1.0. Compliance data at the various levels are unavailable.
R(0) — the uncontrolled rate — becomes R(t) at time t. It must be < 1 for the virus to die out. See Rt.live.
Yeah . . . No.
I suspect there is no significant threat to Mankind / Humankind. Of more concern are the threats to the roots of our civilization, the values we treasure, and the ideals we aspire to.
Climate activist – ‘Make me a ‘climate change’ model projecting a catastrophic future’.
Modeler – ‘I don’t come cheap’.
Climate activist – ‘I’ll pay your asking price’.
Modeler – ‘You got it’.
Exactly!
What kind of person do you think I am?
We have already established that. Now we’re just haggling about the price.
Lies, damned lies and statistics. Modelling is only as good as the inputs. If just one of input is out, the output should be in the bin. The error will be multiplied through the system until the output is useless.
With about 47 variables involved with climate ,with effects varying depending upon how they react with the others, things become so complicated that it is essentially impossible to model. Thus far, the models only come even remotely close when the data are manipulated to match the models (fraud.)
For a start, the weightings for each variable are guesses at best and the most important variable, solar output, is largely ignored.
As for the COVID modelling, it has already been shown to be badly flawed-out by a factor of a thousand, with death reports ascribed to be from COVID being plumped up with deaths with COVID. There are also cases of what are essentially bribes being paid to put COVID as the cause of death to further plump up the figures.
The cAGW and COVID scams come from the same fearmongering stable. Both of the scenarios are pretty hard to swallow if you look at them objectively. Both see the old conman trick of brushing off dissenting views and questioning. Both scams pretend to be real science-the appeal to authority. The science is just opinion for hire. Both see retribution and persecution of dissenting views. To quote Voltaire-“It is clear that an individual who persecutes a man, his brother because he is not of the same opinion is a monster.”
Here in Australia, Dictator Dan in Victoristan is proposing a Bill that calls for the arrest and indefinite detention for those suspected they might break curfew. This is to be conducted by police and public servants. This is as draconian as any despot in history has ever done. As it is, HCQ is banned, based upon one ridiculously flawed study using known to be dangerous doses. Those researchers obviously did not bother looking at the known LD50 data which has been known for decades. Doctors are essentially being forced to violate their Hippocratic Oath on the back of this study. Note that Dictator Dan was Victorian health minister for 3 years.
It is clear that Daniel Andrews is a tyrant. It will be interesting when the vote is put to the Parliament. Will the rest of the “leaders” follow the edict of the tyrant? I would like to know where the Prime Minister is.
The most disturbing thing is that there are a lot of sheeple who hail Dictator Dan as a strong leader. A true strong leader works within the rules to achieve an outcome. A dictator rides roughshod over the laws, rules, Constitution and human rights to achieve their outcome. The differences between a real strong leader and a fake strong leader(dictator) could not be any clearer.
My observation on the IPCC for some time has been as follows:
The United Nations is a political organization, not a scientific one. The IPCC was created by, and is controlled by, the United Nations. It provides the best science that money can buy – i.e., pay some scientists enough money, and they will find a way to tell you whatever it is you want to hear.
The first lpcc report from the scientists found on manmade climate change, the reports released to the public are written by the policy makers who then changed them to suit their goals.
The most clear and present danger is still nuclear war, even if it is over in one day.
Nuclear weapons have their own “Big Nuke” industrial complex. It remains more dangerous than ISIS and Iran put together.
Even a limited exchange of nuclear weapons makes climate change just spit in the ocean. The Greens should attack the UN for failing to end proliferation.
Without nuclear weapons, how many people would had died in World War III. Europe were fighting a major continent war every twenty years. You can bet without nuclear weapon is would have happen. Losses would been al least 200,000,000. Nuclear weapons may have been what saves us.
If there were no computer modellers there would be no computer models.
Many talented programmers have shown that the weather is very unlikely to be predicted more than about 7-10 days at the outside. There are too many variables and too many unknowns as to how they interact.
Climate is the average weather.
Hence, no one can predict the climate more than a few days ahead.
Climate models and modellers are strictly POLITICS as usual, not science.
consistent weather is easy to predict. Where I live summer is very consistent.
The Fall & Spring turn, for a short period of time, are very chaotic & patterns not consistent. Winter time is generally easy to predict.
In the fall and spring turn a coin flip will get you the appropriate odds that the weather prediction is good two days out.
In the summer and winter a 10-day guess is correct more often than not.
Climate is generally defined by NASA, NOAA and other major climate study organizations as being weather over a specified geographical area averaged over a period of at least 30 years.
Climate is not just “average weather” independent of time.
Very interesting, Gordon.
So are you implying that under the major climate organization’s definition, in the absence of a specified time period, there is no such thing as climate? So the climate of the Mediterranean, for instance, doesn’t exist unless it refers to the weather that has occurred in that region over a specified period of three decades or more? Just describing it as warm dry summers and mild dry winters doesn’t cut the mustard?
More to the point, since the weather over a specified geographical area averaged over a period of 30 years is likely to differ from the weather over the same area averaged over 50 or 100 years, does that mean that today’s climate varies depending on the period over which one chooses to average past weather?
If so, would that make climate an abstract concept, a statistical one, or a purely imaginary one? Would it mean that climate or its derivative concept climate change were incapable of producing any physical effects and that any effects that are attributed to them are actually the effects of weather, which does consist of physical phenomena?
Tim, you posted: “So are you implying that under the major climate organization’s {sic} definition, in the absence of a specified time period, there is no such thing as climate?”
No, I am not. I am stating the generally-accepted method of distinguishing “climate” from “weather”.
Try to find a place on Earth that has not had at least 30 years of weather.
Logic 101.
You also posted: “More to the point, since the weather over a specified geographical area averaged over a period of 30 years is likely to differ from the weather over the same area averaged over 50 or 100 years, does that mean that today’s climate varies depending on the period over which one chooses to average past weather?”
You got it! That is EXACTLY the reason, for example, we can discuss the modern climate of the Mediterranean region and compare it to the climates that existed there during the Little Ice Age, The Medieval Warm Period, and even during the last glacial period.
Understanding 101.
“Climate” is indeed a relatively subjective term. “Climate” has never PRODUCED physical effects . . . it is the product of observations of physical effects over time. This misunderstanding is quite common to CAGW alarmists who want to fight climate change™.
7-10 days is extremely generous. In some cases, 7-10 hours is generous.
Yes, in very stable climates it’s sometimes possible, but European weather is anything but stable especially in the west where the Atlantic likes to create chaos.
Météo-France talking about their own performance:
“The forecast check shows that the 24 hour weather forecast over France is correct about 90% of the time, and that the 24 hour temperature forecast at a given point has average accuracy of the order of 1 to 1.25 ° C. At 7 days, the accuracy is around 3 ° C.”
….they go on to say…
“Over the past thirty years, the quality of forecasts from Météo-France’s Arpège model has gained one day every ten years, today, four-day forecasts are as reliable as three-day forecasts at the beginning of the 2000s.”
I follow the forecast everyday and the four day forecasts are still very variable in there accuracy.
Well, FWIW, here is what is stated at NOAA’s SciJinks webpage ( https://scijinks.gov/forecast-reliability/#:~:text=The%20Short%20Answer%3A,90%20percent%20of%20the%20time.&text=Since%20we%20can't%20collect,assumptions%20to%20predict%20future%20weather. ) :
“A seven-day forecast can accurately predict the weather about 80 percent of the time and a five-day forecast can accurately predict the weather approximately 90 percent of the time. However, a 10-day—or longer—forecast is only right about half the time.”
The IPCC says in AR5 that the climate cannot be modelled because it’ a non linear, chaotic system.
In real life computer models are useful for developing things like a better bumper or airframe before spending the money to design and build a prototype.
That prototype is tested in real life to see it works before going into production.
If it doesn’t, adjust the model using the new data and try again.
Of course, in ‘Climate Science”, the science was settled since Hansen’s hearing.
California bought it and now they have rolling blackouts.
Hang on. Didn’t that infallible religious body the IPCC say, unequivocally, that the climate is a chaotic system that cannot be modeled?
Doesn’t that settle the science?
There are only two models worth diddly squat. The first is the 65 Shelby Mustang and the other model is often seen on a cat walk, and guess which one is the most reliable.
I disagree: 67-69 Camaro is a much better model than a 65 Shelby Mustang.
I will not comment on your other example because I am aware of the fashion industry and how fleeting that catwalk may be to anyone that does not conform to a designer’s idea of the perfect walking hanger that is typically 7 feet tall. 🙂
The most revealing statistical observation: GIGO! (Garbage in, garbage out, for the uninitiated)!
The biggest issue with computer modelling continues to be getting it to agree with observation.
Never let data get in the way of a good model.
I’m still waiting for that computer model that’ll make me millions on the stock market.
Such models live in a highly Darwinian environment. They work for a short while, until similar models are developed to compete with them whereupon they lose their advantage. Some of the new models exploit weaknesses to become the next short term kings of the market.
“Cross a great river and destroy a great empire”. It is part of human nature to get some advice on how the future will unravel before he takes a major or even minor decision. It used to be bone chips before computer chips, it used to be the crystal ball before the crystal (silicon) , etc. It was the oracle and gypsy ladies before the computer programmer. Anybody prediction on how the future of fortune telling will evolve—artificial intelligence based on the rubbish of the internet?
LOL
GIGO garbage in garbage out the first thing you learn in computer programming the second thing is that you can control the results so you get the output you want by manipulation of the coding. See the IPCC modeling and dictator dans super computer results mandating further lockdown. It doesn’t matter what computer you use if you write the code to get a predetermined outcome. None of the IPCC or dictator dans models have come true, the populace is just being duped. Using the term super computer is just an appeal to authority trick.
“Using the term super computer “
Just means they get the answer they want, quicker.
It makes no difference to the actual answer.
The models versus reality discrepancy leads to a simple hypothesis that is backed up by the data, and the hypothesis is that very little of late 20th century warming was the result of anthropogenic CO2.
Occam says we ought to be content with that, but that means ultimately that all the people working in climate science are a waste of grant money. And all the windmills, batteries and solar panels are, too.
No, I can’t see the IPCC admitting that, either. Can you?
Cui Bono?
Too few people of any importance, sadly.
Will they ever receive their comeuppance? Probably only under the despotic systems they wish to replace our liberty with. They might get eaten last, but they’ll always get eaten.
I work in IT, perhaps that is why I am cynical about models, but the problem stems from the publics view of computers as some kind of miracle machine. A sort of magic box, that is invincible, infallible, and god like.
Oh, and I worked on climateprediction.net by the way. Which was quite amusing, arguing with the rest about the impact of CO2. 🙂
Edward R. Tufte’s masterwork *The Graphical Display of Quantitative Information* is available for free download at Archive.org.
https://www.amazon.com/Visual-Display-Quantitative-Information/dp/1930824130
If all were aware of Tufte’s guides much angst would be avoided. He is a brilliant polymath.
So true. But perhaps the villain should be computer modelling WITHOUT VALIDATION. Modelling is a valuable tool but only when it is treated as an hypothesis to be compared to what happens int he real world by experiment or observation.