Thing is, it's easy to build a model that "simulates" past data. That can be done with extreme accuracy... by polynomial curve fitting, for example. Heck, Excel can do that. The more challenging thing to do is create a model that predicts the unknown future.
Thing is, it will be some time before this model is proven wrong. And when it's proven wrong, somebody will announce an improved version.
-- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation
be
eck,
tn it's proven wrong, somebody will announce an improved version.
The Daily Telegraph is another rubbishy right-wing newspaper, and its scien ce reporters are British science reporters, who don't know anything about s cience.
Even so, if you read the article, rather than look at the headlines, you wi ll note that the new model is only achieving 62% accuracy on predicting wea ther that has already happened from the data available before it happened, which is pretty close to useless. There is a comment to the effect that the investigators hope to improve the accuracy to 80% with further work, which might be more useful, if it happens (which is unlikely).
In other words, it's the usual PR splash that has been taken more seriously than it deserves by a newspaper with a long history of getting science wro ng.
We know why John Larkin was silly enough to re-post it here - it's one of h is less-than-charming habits. What is puzzling is why he can't learn from e xperience that this is a habit he ought to break.
-- Bill Sloman, Sydney
On a sunny day (Tue, 01 Apr 2014 18:48:52 -0700) it happened John Larkin wrote in :
Its the usual quack. The latest British model, which simulates the climate on a more detailed scale, was shown to be 62 per cent accurate at giving a broadbrush prediction of winter conditions when it was tested on 20 years of retrospective data, the Times reports.
So 38 % of the time it will be 100 % wrong :-) 'broadbrush' for the rest of the time. Look up the 'butterfly effect'.
But I am nearly broadbrush 62 per cent certain some gov will use this to predict stunning glow ball worming in the next years, and use it to raise taxes. The taxes then pay for a bigger weather computah.
You know, if you run a simulation on the financial market that way for x years of old data, you can create a system that makes a profit (I did
You should be aware that the Telegraph science is written by people with almost no understanding of the subject. One of their previous science journalists couldn't even plagiarise text books reliably!
Only by a mathematical ignoramus like you who futzes with the numbers. Overfitting data is one of *the* most common mistakes and you have previously demonstrated that you don't understand why it goes haywire.
UK Met Office know what they are doing and their forecasting models are world leading. How else would you propose to train up a forecasting system other than by comparing its results against historical data (and then with new data as and when it is observed)?
Strictly the Excel charting polynomial fit in all versions apart from out-of-the-box XL2007 can create accurate polynomial fits. The version used in the spreadsheet is intrinsically numerically unstable and almost never gives a correct least squares fit even for a cubic. It isn't massively far out unless you have an X axis with data that starts well away from the origin (calendar dates for instance).
Microsofts official response was "business users do not care".
You don't get on with maths do you? Science is all about building and testing predictive models of the world to determine how it all works.
-- Regards, Martin Brown
n be
Heck,
at
en it's proven wrong, somebody will announce an improved version.
scale, was shown to be 62 per cent accurate at giving a broadbrush predic tion of winter conditions when it was tested on 20 years of retrospective d ata, the Times reports.
Don't. The butterfly effect applies to making detailed weather predictions for specific places at specific times.
Broad brush predictions - how much rain will fall over province over a mont h - are more tightly constrained by thermodynamics and are less susceptible to the butterfly effect. Farmers have been relying on this since we starte d get serious about agriculture - we don't know when it is going to rain bu t we can be fairly sure that there will be enough rain over the growing sea son to make the crops grow.
predict stunning glow ball worming in the next years, and use it to raise t axes.
Because you don't know what you are talking about.
-- Bill Sloman, Sydney
-- As opposed to the known future?
-- Perhaps he thinks that randomly winning a skirmish will expiate the errors made by conscious decisions leading to lost past battles. John Fields
How about Met Office, via the Times?
Done right, it doesn't go haywire. A good curve fit can beautifully predict the past.
How do you know that it's not effectively curve fitting? But to answer your question, publish its daily weather forcasts and see how it does.
Predictive models should predict! Thses sorts of things get publicity now, and nobody remembers years later.
-- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation
F=MA is usually accurate to parts per billion, for any sorts of materials. At relativistic speeds, it needs to be tweaked. Monday, the forcast here was for dry weather; Tuesday, we had pounding rain and hail.
-- John Larkin Highland Technology Inc www.highlandtechnology.com jlarkin at highlandtechnology dot com Precision electronic instrumentation
Obviously. Lots of things are predictable, knowable, and known.
Cheers, James Arthur
It's likely that we are witnessing a historical political perversion of the scientific community, the greatest ethical blunder in the history of science. I feel priviliged to witness this in my lifetime.
-- John Larkin Highland Technology, Inc jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Most scientists, too.
That's why it's the biggest ethical blunder in the history of science: because it will kill millions of people.
It's bizarre that it's fine to question relativity and quantum mechanics and the speed of light and the structure of the universe, but open-minded discussion of climate (of simulations of wildly chaotic systems with unknown parameters and unknown forcings and bad instrumentation) is severely supressed. People are arguing that questioning AGW theory is a criminal act. Sounds like Spain in 1600.
It will be fun to see this play out.
Maybe we have too many scientists, as we arguably have too many engineers. We have cranked up the quantity as we have cranked down the quality.
-- John Larkin Highland Technology, Inc jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
"Ethical blunder" is right up there with "terminological inexactitude."
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
Both sound like Bad Things to me.
-- John Larkin Highland Technology, Inc jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
The Times is a British Murdoch newspaper. It employs British science journa lists (who typically haven't studied any science at university) and - like very other Murdoch newspaper - presents the output of the denial propaganda machine as news.
can be done with extreme accuracy... by polynomial curve fitting, for examp le. Heck,
ct the past.
Obviously. The "going haywire" bit happens as you move outside the period y ou fitted - earlier or later..
ur
John Larkin hasn't noticed that this isn't a daily weather forecast - which is susceptible to the "butterfly effect". "Broadbrush" long-term forecasts are more constrained by the thermodynamics, and can be less susceptible to small changes in initial conditions.
that
when it's proven wrong, somebody will announce an improved version.
, and
Since you don't understand what it's claiming to predict in the first place , you'd be exactly the kind of nobody who wouldn't remember years later.
You didn't correctly understand what was being claimed when you re-posted t he article, so whatever you are going to remember is going to be reliably w rong.
-- Bill Sloman, Sydney
John Larkin thinks he can make fatuous statements like that because he's a gullible sucker for the output of the denialist propaganda machine.
The only "ethical blunders" around are being made by the likes of Lord Lawson of Blaby who have been paid to retail pernicious nonsense by the fossil carbon extraction industry, as a retail merchant of doubt.
A decade or so ago, he would have been telling us - for money - that smoking wasn't nearly as bad for your health as scientists claim. Now he is being bribed to tell us that CO2 isn't quite as bad for the planet as science suggests.
John Larkin is gullible enough to believe him and the other ignorant stooges that the fossil carbon extraction industry is paying to mislead the ignorant.
-- Bill Sloman, Sydney
Yup. In accordance with the national genius that Americans have for euphemisms. ;)
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 hobbs at electrooptical dot net http://electrooptical.net
As if John Larkin had a clue.
If the latest IPCC report is to be taken seriously, the failure to take ant hropogenic global warming seriously has the potential to kill billions of p eople - not mere millions. They talk about a human population crash from th e current seven billion to less than one billion. That would take a couple of generations, and would primarily reflect severely reduced fertility, but it would involve a lot of people starving to death, because agriculture wo uldn't work in a warmer world the way it does now.
John Pollyanna Larkin assure us that agriculture would work better, but he doesn't seem to know what he is talking about.
John Larkin still hasn't noticed that while weather is chaotic, climate isn 't. Farmers have been exploiting this for the past few thousand years, but John can't get his head around this simple distinction.
People aren't arguing that questioning AGW theory is a criminal act. It goe s on all the time. What is criminal is misrepresenting the science in a way that impairs people's understanding of what is actually going on.
People - like Lord Lawson of Blaby - who are paid to mislead the public sho uld be charged with fraud.
I'm not amused by what I've seen so far.
Unfortunately, the only way of assessing scientific or engineering quality is by assessing the quality of the work that they do.
We are pretty good at selecting students who are going to be good at passin g examinations. The SAT scores that do that well are less-than-reliable pre dictors of post-education performance, so if you want more good scientist o r good engineers, you have to put more students through university and sele ct the good ones by seeing how they perform after they have graduated.
There may be an element of diminishing returns here - for a start, students with lower SAT scores are more likely to drop out before they graduate - a nd some people will do what I did and find out that they should have studie d engineering rather than science, or what you did when you found out that you were better at marketing than engineering.
-- Bill Sloman, Sydney
It would be an interesting exercise to build a small but complex system and issue a challenge to climate modelers to model and predict the behavior o f the system... Then check the predictions with measurements. We need to think of a sufficiently complex system that can be built and measured. Reme mber the biosphere?
Mark
"
"Terminological inexactitude" is not an American euphemism.
Winston Churchill seems to have invented it, back in 1906
It was taken up by the UK house of Commons as way of saying an MP was lying without using the word "lie" which is prohibited as "unparliamentary langu age".
"Bad Things" isn't - in this case - a euphemism. but rather a symptom of an popular American habit of imprecision in language. John Larkin takes this to extremes by posting mendacious nonsense which he's been suckered into a ccepting by the none-too-convincing wiles of the denialist propaganda machi ne.
If he could do real scepticism, also known as critical thinking, he'd be le ss vulnerable to this kind of rubbish, but he wasn't paying attention durin g those of his undergraduate lectures which were supposed to get him starte d on this.
Too busy "doing engineering", from what he tells us.
-- Bill Sloman, Sydney
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.