Friday, November 28, 2008
Death of an anachronism
Sad to report, George King, the mixed-metal, retail supplier of Tooting appears to have gone out of business. Stock looks to have been stripped out of his shop and the closed sign up. Where I am going to go for the odd bit of stainless steel now? There must be a demand for this kind of thing but the only other shop I know of that does this sort of thing is the rather grander Mackay's of East Road, Cambridge. Good but dear. Would not have matched King on price.
Rather better news though was the high standard of road manners in the rush hour between Tooting and Epsom. Impressed with the amount of giving way going on at junctions. Perhaps commuters who drive all the time know that manners works, perhaps in contrast to the school runners and Sunday drivers I am more usually in contact with.
And furthermore, I think I can promote myself corporal from lance-corporal in the army of the geek. After wrestling with a recalcitrant USB flavoured microphone for a while, suddenly thought to try plugging it into a differant USB socket on the same PC. And then it sprang into life, not all USB sockets being equal. A trick first seen from a helpful gent. in Exminster who pulled off the same stunt with a USB flavoured digital camera.
Back at the ranch been taking another look at Hacking on 'Rewriting the Soul'. Shocked again by the sloppy standard of reasoning reported among otherwise respectable doctors and scientists - in this case, in the area of child abuse and multiple personality. It is as if the players in a fashionable field leave their brains behind when they go out onto the playing field. Incidently, a propos of the claim that child abuse (in its current sexualised sense, rather than its Victorian sense) is social-class blind, he observes that this was a convenient story for public relations reasons, when the pioneers were trying to get the subject onto our collective radar, for which there is little evidence. To me, at any rate, the story is counter-intuitive: I would have thought it far more likely that child abuse, like many other evils, will be much more common in hard pressed poor families than in comfortably off ones.
More mundanely, I was struck by a headline in the DT the other day (11/11): "Harold Wilson had Alzheimer's while in office, study says". The story below the headline was rather more modest in its claims, but I was sufficiently struck, as a keen student of senior moments, to look up the article. I rapidly find my way to something called ScienceDirect, an offshoot of Elsevier, I think these last being a Dutch academic publisher. For the small consideration of $37.49, I can download a pdf of this about to be published article, by a researcher at the University of Southampton. Which I do, to find that the article is a sensible, if rather slight review peice, the sort of thing he can probably knock off, on the back of some graduate student he is supervising, on a dull Friday afternoon. So how is the thing priced? Do Elsevier arrange for the puff in the DT and whack the price up to catch the wave of orders, then drop it back to something more sensible afterwards? Does the researcher get any of the money? Does his (presumably) full-time employer?
To return to Alzeimers, it seems that you cannot know whether someone has it for sure unless you do a postmortem examination of the brain. And while the article was not explicit, it implied that in the case of Harold Wilson no such examination was done. And I had thought that there was always a postmortem when somebody important died - in the olden days to try and ascertain whether the son and heir (or anyone else) had been trying to speed up the course of nature. Anyway, the hypothesis of the paper was that Wilson did have Alzeimers and the idea was to examine his speech to see if there were any traces there. Maybe in the future, analysis of speech could be used to give early warning of Alzeimers onset. Not sure that this is particularly useful from a medical or social point of view, but it is interesting. And yes, on the basis of a fairly limited sample of speech and some complicated statistical calculations (performed by some package, rather than by the researcher, who only had to know how to spell the name of the statistical test), there was a change in speech pattern which might have had something to do with Alzeimers. It would be worth looking a bit harder. So a weak, if entirely respectable conclusion. So the headline - which may have sunk into the factlet store of plenty of people besides myself - is quite untrue. The study says nothing of the sort claimed.
The moral of the tale is that when you see a phrase like 'recent research suggests' or 'the latest study suggests' in a newspaper, you should be careful. The main business of journalists is to fill the page with fodder that sells newspapers. A good angle or catchy headline is much more important than the quality of the foundations.
This article was also interesting for the way in that the researcher was trying to replace expensive analysis of text by people with cheap analysis of text by computers. The catch being that computers do not yet understand the text that they are analysing, and so the analysis has to be fairly simple minded. Starting with simple things like word use analysis - although there are plenty of pitfalls even here. Now while I believe that word use analysis can be used to distinguish the work of one man from that of another and to date the works of one man, it seems rather a blunt instrument to detect a degenerative disease. It might - but it would not be a surprise if it did not. I think we will have to wait until computers have been trained - as they will - to be a bit cleverer.
Rather better news though was the high standard of road manners in the rush hour between Tooting and Epsom. Impressed with the amount of giving way going on at junctions. Perhaps commuters who drive all the time know that manners works, perhaps in contrast to the school runners and Sunday drivers I am more usually in contact with.
And furthermore, I think I can promote myself corporal from lance-corporal in the army of the geek. After wrestling with a recalcitrant USB flavoured microphone for a while, suddenly thought to try plugging it into a differant USB socket on the same PC. And then it sprang into life, not all USB sockets being equal. A trick first seen from a helpful gent. in Exminster who pulled off the same stunt with a USB flavoured digital camera.
Back at the ranch been taking another look at Hacking on 'Rewriting the Soul'. Shocked again by the sloppy standard of reasoning reported among otherwise respectable doctors and scientists - in this case, in the area of child abuse and multiple personality. It is as if the players in a fashionable field leave their brains behind when they go out onto the playing field. Incidently, a propos of the claim that child abuse (in its current sexualised sense, rather than its Victorian sense) is social-class blind, he observes that this was a convenient story for public relations reasons, when the pioneers were trying to get the subject onto our collective radar, for which there is little evidence. To me, at any rate, the story is counter-intuitive: I would have thought it far more likely that child abuse, like many other evils, will be much more common in hard pressed poor families than in comfortably off ones.
More mundanely, I was struck by a headline in the DT the other day (11/11): "Harold Wilson had Alzheimer's while in office, study says". The story below the headline was rather more modest in its claims, but I was sufficiently struck, as a keen student of senior moments, to look up the article. I rapidly find my way to something called ScienceDirect, an offshoot of Elsevier, I think these last being a Dutch academic publisher. For the small consideration of $37.49, I can download a pdf of this about to be published article, by a researcher at the University of Southampton. Which I do, to find that the article is a sensible, if rather slight review peice, the sort of thing he can probably knock off, on the back of some graduate student he is supervising, on a dull Friday afternoon. So how is the thing priced? Do Elsevier arrange for the puff in the DT and whack the price up to catch the wave of orders, then drop it back to something more sensible afterwards? Does the researcher get any of the money? Does his (presumably) full-time employer?
To return to Alzeimers, it seems that you cannot know whether someone has it for sure unless you do a postmortem examination of the brain. And while the article was not explicit, it implied that in the case of Harold Wilson no such examination was done. And I had thought that there was always a postmortem when somebody important died - in the olden days to try and ascertain whether the son and heir (or anyone else) had been trying to speed up the course of nature. Anyway, the hypothesis of the paper was that Wilson did have Alzeimers and the idea was to examine his speech to see if there were any traces there. Maybe in the future, analysis of speech could be used to give early warning of Alzeimers onset. Not sure that this is particularly useful from a medical or social point of view, but it is interesting. And yes, on the basis of a fairly limited sample of speech and some complicated statistical calculations (performed by some package, rather than by the researcher, who only had to know how to spell the name of the statistical test), there was a change in speech pattern which might have had something to do with Alzeimers. It would be worth looking a bit harder. So a weak, if entirely respectable conclusion. So the headline - which may have sunk into the factlet store of plenty of people besides myself - is quite untrue. The study says nothing of the sort claimed.
The moral of the tale is that when you see a phrase like 'recent research suggests' or 'the latest study suggests' in a newspaper, you should be careful. The main business of journalists is to fill the page with fodder that sells newspapers. A good angle or catchy headline is much more important than the quality of the foundations.
This article was also interesting for the way in that the researcher was trying to replace expensive analysis of text by people with cheap analysis of text by computers. The catch being that computers do not yet understand the text that they are analysing, and so the analysis has to be fairly simple minded. Starting with simple things like word use analysis - although there are plenty of pitfalls even here. Now while I believe that word use analysis can be used to distinguish the work of one man from that of another and to date the works of one man, it seems rather a blunt instrument to detect a degenerative disease. It might - but it would not be a surprise if it did not. I think we will have to wait until computers have been trained - as they will - to be a bit cleverer.
# posted by Jim @ 12:20 AM 
