A needle in the eye for science

On my commute home the other day I was listening to the Radio 4 PM show, which had an article about acupuncture. It was inspired by a news item about a recent trial (also reported by BBC news online) and was framed as an interview with two experts. One expert (I couldn't find out his name) was pro-acupuncture, the other (Ben Goldacre) was quite sceptical.

Now I have no axe to grind about acupuncture - it looks kinda sharp, but people seem to like it and as far as I can tell no-one's bled to death yet, but I found the whole article deeply depressing. Why? Because the scientific results were being horrifically misrepresented, and even the sceptic seemed oblivious to the perversion.

So here's the experiment and results in a nutshell:

Over 1100 patients take part in the study. One group is given conventional therapy, one is given acupuncture and the third is given a sham acupuncture treatment (that looks like acupuncture, but is essentially just randomly pricking people with needles).

The results:

1. 27% of the conventional group show an improvement
2. 47% of the acupuncture group show an improvement
3. 44% of the sham group show an improvement

So the question, science fans, is what does this tell us about acupuncture?

The pro-acupuncture guy stated (with no sense of shame as far as I could tell) that this proves that acupuncture works. Remember I was driving? I almost crashed.

Never have I seen such an incredible twisting of a set of results (actually that's not quite true, a few years back I saw an article in the Bristol Evening Post that said a car was stolen every 30 minutes in Bristol, and then a councillor was quoted as saying that this was ok, as a car was stolen every 3 min in the UK, and so Bristol was plainly better than most :-/ Maths, people. Maths! )

To me these results show two things (within the scope of the study):

1. Acupuncture is not significantly better than a random sham activity
2. Both Acupuncture and the sham are better than conventional treatment

So the most probable conclusion is:

1. Acupuncture benefits significantly from a Placebo effect, that is so strong that it even beats conventional medicine

This is not the same as "Acupuncture works", in fact its a damn site closer to "Acupuncture is all a load of bollards".

I'm sure that the German scientists behind the work came to this conclusion (the more balanced wording that is :-) but I was dismayed at the complete lack of ability of the general public, even a Radio 4 audience, to understand the results in an unbiased way.

The sceptic was trying to be nice, and was arguing that although the improvements were impressive, we should be spending our cash on less expensive Placebo effects. But I wanted him to just pick up the massive logical club lying conveniently next to him, and wallop that damn needle guy right in the family arguments.

Don't get me wrong, I haven't written acupuncture off completely. This is just one study, and it does clearly indicate that the Placebo effect is worth investing (and acupuncture is one way). But please don't get overexcited about your meridians.

And don't even get me started on homeopathy (where's my axe!). Dawkins says it best:

Hypertext Conference 2007

I've been on my travels for a while, culminating this week in Manchester, where I attended the ACM Hypertext Conference. Hypertext is a smaller gathering than it used to be and in recent years has been in danger of becoming a fringe event, however this year numbers were up and the conference felt very positive. Besides I think that everyone should hang out at the fringe occasionally, it's good for the soul.

In the interests of full disclosure I should add that I was part of the Program Committee and am also on the Hypertext Conference steering group - but to be honest my fondness for the Hypertext Conference comes more from the fact that it was the first conference series I went to as a PhD student than it does from my professional connections with the event.

For me there were a number of highlights this year (besides the fine wining and dining that always accompanies a Hypertext conference):

• Hypertext and Tragedy - my only official job this year was as part of a panel on Hypertext Tragedy organised by Nathan Matias. Now my ignorance of literature is dwarfed only by my ignorance of literary theory, so I was feeling a little bit of a fish out of water as I sat alongside Emily Short (Interactive Fiction author), Nick Lowe (Reader in Classics) and Kieron O'Hara (Philosopher and Epistemologist). However, I have been involved in number of projects concerned with building narrative systems (information systems that in some way use story structures) and I hope that I was able to convey a sense that engineers do have some appreciation of narrative theory, and get across our slight disappointment that we can't get more mileage out of the ideas of the Russian Formalists (let alone Aristotle).


• Un-hyping the Semantic Web with Ted Nelson - If Sir Tim is the Father of the Web Gods then surely Ted Nelson must be one of the Titans, probably Atlas, bearing the weight of all the Web's failings on his back (this is responsibility in the didn't-finish-Xanadu-first sense). He is also great at coining new words and names (including the word Hypertext) and this year I have added swarf (swoop+morph), flinks (floating links) and hyperorthogonal (orthogonal in n-dimensions) to my vocabulary. I've known Ted and his wife Marlene for a few years now (Ted was a visiting Professor at Southampton) and it was great to catch up with them and talk over one of Ted's bugbears - the Semantic Web. I can't help thinking that some of the Semantic Web's most vocal supporters are also it's worst enemies; the arguments I've heard against the web such as "no-one will ever agree the one-true ontology" and "RDF is no match for the expressive power of human language" are just staw men, made possible by people who oversell the Semantic Web vision. Much of my conversation with Ted and Marlene was about me trying to un-hype the Semantic Web and present it for what it is - a simple set of technologies that allow machines to exchange meaningful tokens - perhaps a better name would be the Semiologic Web. Intriguingly it occurs to me that Ted's ZigZag idea might make rather a nice storage mechanism for Semantic Web triples (or if you like Semantic Web standards might make a nice exchange format for zzStructures). I shall have to mull that over :-)
  
• Wendy's keynote - This year we had the conference dinner at Christie's Bistro, which is part of Whitworth Hall at The University of Manchester. Wendy Hall gave the after-dinner speech, choosing to reflect on the history of Hypertext and the divergence and potential re-integration of Web and Hypertext research. Wendy recently gave up her position of Head of School within ECS and it was great to see her get some of her bounce back now that the pressure has lifted. There were a lot of messages in her talk, but the one that I take away is the message that came through between the words - that research can be joyful :-)
  
• Discussing Iraq with Mark Bernstein - The final event associated with the conference was a fascinating talk by Mark Bernstein on the rebuilding of Iraq and the role or lack of role played by information technology in that process. In some ways technology has dramatically changed our perspectives on war - the blogs of the citizens of Baghdad or the online accounts of the soldiers serving there have formed an independent channel of information that has prevented the type of media-control that we saw in the first Iraq War, but I am also struck my how much has not changed. It seems that even technology cannot prevent wilful ignorance. After all the European imperialists managed in the 18th and 19th centuries (for good and ill) without a Wiki or Blog in sight.

No other conference I can think of has the shear diversity of Hypertext. I hope that Wendy is right, and that the conference continues to grow and attract renaissance men and women - it is an important part of my continuing education, and I'm looking forward to next year's event in Pittsburgh.