Return to Website

Number Watch Web Forum

This forum is about wrong numbers in science, politics and the media. It respects good science and good English.

Number Watch Web Forum
Start a New Topic 
View Entire Thread
Re: Why politicians like computer models

I don't think any of the engineers on this board would disagree with what you described. Boeing either has just is about to test the wings on the 787. They used a model to build the wing. The model seems to have worked. I understand from some of the people who work there that there are a little more nervous than normal with this wing. Composites are still youthful and not fully characterized. Those engineers should be more confident than any GCM modeler though about the expected results. They have put strain gauges, thermocouples, and lasers and whatever other wonderful measuring devices onto the composites and measured the responses in as many situations as they could and fed that data back into their computer models.

I still bet every one in the control room will be watching with just a little bit of stress. Some of them may even resort to racing to the end of the building to find that shed where it is still allowable to smoke to relieve some of that stress.

The world is a crazy place right now.

Re: Why politicians like computer models

The problem with composites is to do with fatigue. Under repetitive strain, loss of bonding between reinforcement and matrix can occur and this can lead to rapid failure. Debonding is also exacerbated by exposure to aggressive environments and temperature change. Composites ain't too good in shear either.
Metals have the advantage of yielding slightly wihout loss of strength - all good stuff.

Re: Why politicians like computer models

I didn't really have engineering computer models in mind in my earlier post. Engineering computer models tend to be used to contribute to the design of some engineering product rather than in any decision making by politicians. I doubt that politicians even know there is such a thing as 'computer aided engineering' or 'computer aided design'.

As I remember it, engineering computer models did have a slightly dodgy image back in the 1980s, but not really in the same league as AGW. There was an entrepreneur called John Robinson who used to run a magazine called 'Finite Element News' and Robinson used to write to engineering firms hinting there could be some sort of disaster unless the companies sent people on his training courses. Then a UK quango called NAFEMS was set up which jumped on the bandwagon created by Robinson, initially claiming that the organisation was needed because of the arrival of cheap desktop computing allowing much smaller firms to get involved in computer modelling. But there have been remarkably very few incidents worldwide where engineering computer models have contributed to some sort of accident, and the number of incidents doesn't seem to be correlated with whether a given country has a NAFEMS-type organisation or not. What appears to have happened is that the worldwide engineering community turned out to be nowhere near as gung-ho in adopting computer modelling as Robinson and NAFEMS thought they were going to be. The gung-ho mentality is however much stronger in scientific computing, and disturbingly the distinction between real experiments and 'thought experiments' seems to have become blurred.

What I was doing in my earlier post is speculating as to where the enthusiasm of politicians for computer models is coming from. They don't particularly respect scientists as David Nutt and the drugs policy advisers can testify. They respect scientists with computer models. Politicians are not picking up the enthusiasm for computer models from the media or the general public. The enthusiasm seems to be originating from something within their own 'bubble', and I think it is tied up with the use of computer simulation of nuclear weapons, which politicians seem to have decided is going to work in advance of actually being able to do it.

Re: Why politicians like computer models

I suspect the enthusiasm of politicians for computer models comes from the experience of business. There was an apparent recognition in boardrooms that computers were the coming thing, but directors knew nothing about them. They therefore hired 23-year-old pony-tailed kiddies who had spent far too long playing games and needed to get out more as IT managers.
Politicians are now in the same bind and, as always, are desperately swivelling their eyes to see which way things are going so they can rush round to the front and claim to be leading. What better than to manufacture some bogeyman and select a small, neglected branch of science?
I strongly suspect the late Michael Crichton had it right with "State of Fear". If yoy haven't read it yet, don't bother with the story, which is just another American technothriller which could well do with a good editor, but do read the appendix and bibliography and look at the temperature graphs.

Re: Why politicians like computer models

There are two things that come together here, computer models and Politicians.
That is when we get a really serious problem.
I expect the reason the MP's expenses were so high is that they are suckers for double glazing, fitted kitchen and conservatory salesmen.
Confront them with a computer company and/or software and they seem to suspend all critical faculties.
They have short memories and are more vulnerable than most to the law of unintended consequences.
Inevitably almost every computer or system sold to the government costs far more than first thought, takes twice as long to deliver and only does half of what it was supposed to do and does that wrongly. yet still they keep buying.
The latest brainwave is to put all of our medical records on a central computer so some hospital or surgery in the Shetland Islands can access your data whenever they like.
Ah yes, I forgot, the other propensity of the government and its servants is to store as much data as they can on unsecure disks/laptops/memory sticks etc and scatter them round the country on buses, trains and in taxis for someone to find and publish in the News of the World.

It seems to me, when I visit my surgery, that my local GP has all the information he needs on his computer without having to link in to some mystical new government bought and paid for Pandora's box of a system.
AH, you say, but suppose you are taken ill in Bradford? Well, come to that, suppose I am taken ill in Riga or Los Angeles? This system surely won't be so extensive as to benefit me overseas. Given the number of Brits who go to Falaraki, get blind drunk and end up in some Greek Hospital, I would think this is a serious deficit.
The thing is, the only reason Cedar Sinai Hospital might need my medical records is if I am in Los Angeles. It seems to me that a far simpler and more useful solution is for each of us to be issued with an NHS memory stick or smart card. Or maybe an "app" for our I-phones.
Where I go my medical records go.
If anyone is going to leave my records on a train it will be me and it will be my records and not half the countries.
Each visit to the GP he logs me in with my swipe card/memory stick etc and updates it at the end of the session.
Job done.
best news, thee government has nothing to do with my information so they can't, like the DVLC, sell my data to the insurance companies or the pharmaceutical companies. In fact, they can't make money out of me whatever they do.
Built into an "app" on a mobile phone, or in memory stick with an "app" on my PC, it can issue me with medication alerts and keep track of my prescriptions.
It can even order me new prescriptions when I am online without my having to log into my surgery web site and do it myself.
In other words a different approach would be far less expensive and far more useful than yet another Government money-sink.

Re: Why politicians like computer models

Sacrilege, Anarchy --

What the heck are you talking about... Letting the patient have access to his own records. THOSE records belong to the doctor. They belong to the hospital. They belong to everyone but the patient.

Put that information directly in the hands of the patient and they might read their own file.

I just finished Phil Plait admonishing people like me for thinking that we might understand data or have an ability to analyze it. It is perfectly appropriate in his mind to prevent anyone but approved experts from having access to it.

Apparently, he didn't actually read any of the messages. Those comments were just short hand jargon that pack a lot of meaning into a small space.

Truly amazing

Re: Why politicians like computer models

Well, they can always make these records only readable by the GPs.
But why shouldn't we see our own records?
Doctors are supposed to discus things with us these days.
Plus if we know what is wrong with us we have the internet.

Re: Why politicians like computer models

I hope the sarcasm was apparent.

I think we should be able to read our own charts. Certain people though, should be warned heavily against reading their own charts.

I am sure everyone here fell into this discussion of statistics. The author attempts as has our bending author to explain the challenges faced by statistician, most especially those who use the computer tools but never get a grasp of the implications of the underlying math.

Which makes me say.

It is the people who interpret P < 0.05 to mean that there is a 95% probability of the study being correct who will be dangerous if they get hold of their medical files.

Re: Why politicians like computer models

The NHS system is only a database and an intranet, with an appointments system tacked on. A very big system, yes, but only the kind of thing that the business world has been doing for 20 years. They should just have found out what was available and bought that.

Re: Why politicians like computer models

The problem with off the shelf solutions is that they never match the customers needs unless the customer can figure out how to make excel accomplish the task.

People who can figure out how to use excel as a word processor, database, or any other task they need done, can figure out how to use an off the shelf system to accomplish their task. This may seem obvious, but it is something I run into constantly. People have visions of what they want a system to do which never matches what a canned product can do. you can usually point out how to get the system to mimick the "feature" they expect but it requires the user to adjust how they look at the problem. Ironically it is usually quite easy to get the user trained to do the task. It is near impossible to get the manager to get it.

Managers know better.

Re: Why politicians like computer models

I think they acquired an off the shelf ambulance system and it was tampering with that that cased some emergencies to be downplayed resulting in deaths.

The problem is that it was Government meddling as usual.

By the way, Brad, yes, I did detect the sarcasm. I should have shown that in my reply.

Re: Why politicians like computer models

The first mistake the bosses make is to ask dozens of 'user groups' (i.e. managers) what they would like from a system. This results in hundreds of incompatible demands and raises expectations that cannot be fulfilled. As Henry Ford remarked "If I'd asked people what they wanted, they would have said a faster horse". They then hand this over to consultants who win the work based on their size, cost (the greater the better, it proves they must be good) and ability to schmooze. The consultants have an interest in making the system as complex and expensive as possible in order that they may become essential and, of course, the last thing they want is for the project to be completed and their income stream stanched. The bosses then watch helplessly as a project that they do not understand turns into a money-eating monster.

The alternative is to find out what is available, with modifications that can reasonably be achieved and then offer the users a choice of three or four options. Moral: never let the best drive out the good.

Re: Why politicians like computer models

On the subject of misuse of models, here is a prime example:
BBC Link
Despite the fact that polar bears are apparently thriving, some guy has a “model”, which says they won’t. Cue the BBC in full doomsday mode and you have an impending disaster. A completely misleading piece by Matt Walker, who is well known for writing tripe like this!