More on homeopathy

This homeopathy business is more complicated than it first seems and I confess I'm in two minds about it.

Firstly there's today's news that the House of Commons Science and Technology Committee is recommending the NHS should stop funding homeopathy on the grounds that there is no evidence that homeopathic remedies are more effective than a simple placebo.

But then I read that Franklin Miller of the US National Institutes of Health (in response to growing evidence of the power of the placebo effect) suggests that doctors and researchers should think in terms of contextual healing - that is healing produced, activated or enhanced by the context of the clinical encounter. Or as he puts it:

"Finding ethically appropriate ways to tap the use of placebo in clinical practice is where the field needs to be moving."

It's hard to avoid the conclusion that if you had to invent a practice designed to deliver the placebo effect you might well come up with homeopathy - or something very similar. Homeopaths believe in what they are doing, the remedies are harmless (as demonstrated by the recent mass-overdose demonstration) and there is good anecdotal evidence that people experience a benefit.

The problem however is that the scientific establishment just can't bring itself to countenance an explanatory framework that is so at odds with established standards. To give credence to such alchemical principles as the Law of Similars or the Law of Infinitesimals would be to open the door to all sorts of mumbo-jumbo: dowsing, crystals, pendulums and so on.

Every discipline needs a framework of some sort within which to operate. There are some who, by denying homeopathy any scientific credibility, hope to drive it to extinction. My own inclination is to regard homeopathy as an art and its laws as essentially poetic.

Is poetry (in this broader sense) something that nurtures and enriches our lives or is it merely recreational? Maybe it is time for a little humility and open-mindedness on both sides.

Homeopathy and the placebo effect

I've been thinking about homeopathy. This was prompted by a piece by Michael Brooks in the Aug 20, 2008 issue of New Scientist attesting to the efficacy of the placebo effect.

Then just recently I came across a second article by the same author in the Guardian Science Blog that discusses the placebo effect in the context of homeopathy.

For a simple sugar tablet to make you better, it helps if you believe it to be a real, potent medicine. Furthermore, it seems the effectiveness of any remedy is increased if patients are given time, listened to at length and fully consulted about their symptoms, feelings and treatments.

In a typical homeopathic consultation the practitioner will start by taking a detailed personal history covering physical, mental and emotional states. After identifying a remedy appropriate both to the patient and to his or her specific condition the remedy is potentised somewhat ritualistically by means of a series of dilutions and shakings (succussions).

All of which suggests to me that if the placebo effect is real and effective then homeopathy is perfectly configured to exploit it.

So instead of getting worked up about the fact that homeopathic remedies are so dilute as to contain no trace of their active agent, we should be giving serious thought to experiments designed to provide insights into the therapeutic process.

Here's the sort of experiment I'd like to see done:

Firstly, you recruit a statistically significant number of subjects all suffering from a condition considered (by homeopathic practitioners) to be treatable.

Half the subjects are told they will not participate further in the trial and that they will constitute a control group (group A). The other half are given individual treatment by a group of homeopathic practitioners, starting with a detailed case history and culminating in the identification and preparation of a remedy. The practitioners are told that the remedies will be delivered to all subjects once they have been correctly recorded and labelled and they have no further contact with their patients (at least until the study is concluded).

Of the subjects for whom remedies have been prepared, half are told (by the researchers) that the remedy will not be supplied to them (call this group B). The remaining subjects are all supplied with a course of pills but in half the cases (call it group C) the remedy prepared is replaced by a simple sugar pill. The only people to take the remedy identified and prescribed by the homeopath are those in group D (just 1 in 8 of the full set of participants).

At a series of regular intervals, all subjects are examined with respect to changes in their condition. It helps of course if the condition is such that it manifests itself in a measurable way (at least partially)

It is worth noting that groups C and D (considered in isolation) are engaged in a typical double-blind trial. And since such trials have so far failed to demonstrate the efficacy of homeopathy we might expect the same to be the case here. Or to put it another way: of the people who believe themselves to have undergone a full course of homeopathic treatment there might well be no evidence to suggest that the homeopathic remedy is any more or less effective than a simple placebo.

A similar uniformity of outcome might be anticipated amongst the people in groups A and B. None of these people were supplied with any pills and all believe themselves to be part of a control group.

The really interesting question is whether there would be any observable difference between the people who didn't receive a remedy (groups A and B) and those who did (groups C and D). If the second group were to demonstrate a significantly better outcome than the first then we would be onto something. Nevertheless, while homeopathy might take credit for the method of delivery, it would be the placebo effect that could justifiably lay claim to the attribution.

So while I am personally somewhat sceptical about homeopathy, I find myself having to allow for the possibility that, for many people at least, recovery from certain forms of physical and mental illness can entail a degree of willing self-deception.

I don't believe in offering homeopathic treatment on the NHS. Instead I believe we should be working towards a state of affairs where all doctors are trained to recognise the degree to which listening, attentiveness and respect can have a measurable effect on clinical outcomes. Who knows; homeopaths might have something to teach them in this area.

Omnivorist is now tweeting

And if you are on twitter too there is a tweet this button at the end of every post. If you're not, then please don't lose any sleep about it.

OCDM2

As if it isn't enough to suffer the distressing effects of horological monodigitism, I find myself plagued by a new problem - a mild form of obsessive compulsive disorder with musical manifestations (OCDM2)

I can think of no better explanation than to describe a typical episode.

I'll be enjoying the recollection of one of my favourite pieces of music - let's say Liszt's Fifth Hungarian Rhapsody - when I suddenly become aware that I have been whistling the same 4 or 5 bars under my breath for most of the morning. By the next day it has taken a firm hold and I find myself replaying the same loop in my imagination, more or less unconsciously. It's easy to put the tune aside once I become aware that I am playing it, but it has a sneaky tendency to start up again as soon as my back is turned. It's not unusual for a single tune to get lodged for 4 or 5 days, with occasional episodes lasting anything up to a month.

'Aha', I hear you clamouring to suggest: 'Why don't you just think of a different piece of music ?'.

If only it were that simple.

Certainly using one tune to drive out another is a sensible strategy. But consider this. There are only certain tunes capable of displacing one that has outstayed its welcome. The tunes you'd like to recall - the one's you're particularly fond of - they always turn out to be useless.

Just before Christmas, for example, I had a particularly persistent fixation with Happiness (by Goldfrapp). Now, on the whole, I consider this to be an intelligent and appealing piece of music but after two weeks of uninterrupted mental playback I was beginning to find it tiresome.

I tried a couple of alternatives. The Byrds: Eight Miles High - that classic from the very zenith of Californian hippy culture - never really got much of a hold. Much more promising was Friday Night and Saturday Morning (the Nouvelle Vague version featuring Daniella D'Ambrosio). I've worked with it in the past and know from experience that, while very effective as a musical purgative, it can be a devil to get rid of once it's got it's feet under the table. But - mercifully perhaps - it didn't take on this occasion.

So I was stuck with the Goldfrapp for a few days more before I did what I knew I'd be forced to do all along.

You see there's a fiendishly subtle twist to this particular neurosis that makes one suspect it to be the work of some malevolent intelligence. It's this. While there is no particular difficulty in identifying a tune to do the business - it will invariably be one that is both more banal and persistent than the tune it displaces.

So, out of desperation, I forced out the Goldfrapp with Jesus Wants me for Sunbeam which I endured for an afternoon before resorting to Puff the Magic Dragon. Beyond this point the choice suddenly becomes quite limited as there are only a handful of tunes that are sufficiently fatuous to deal with Puff the Magic Dragon. Of course, there's always The Chicken Song - but that could be regarded as overkill. No, there's really only one candidate as far as I'm concerned and that's Lily the Pink (by The Scaffold).

Some might regard the fact that this song topped the UK singles chart for 4 whole weeks in 1968 as no more than a minor cultural footnote. For my part, I consider it as clear evidence of a significant public health risk.

Crocodile tears

Having picked up from somewhere that Alastair Campbell had come 'close to tears', on the Andrew Marr show, over the matter of the sexed-up Iraq dossier, I eagerly tuned in to the BBC website only to find that the tears were of the crocodillic variety and in any case didn't amount to very much. All the same, he did seem a bit upset and it seems that it's all to do with the way everyone seems to want to go on and on about the lead-up to the war.

What he doesn't seem to realise is that, while he may be tired of the whole matter of Iraq, the rest of us are just warming to the task of understanding exactly how things turned out as they did.

Computer Chess

I have just read a wonderful article by Gary Kasparov called The Chess Master and the Computer. In it he describes how the earliest efforts to construct chess-playing machines were motivated in part from a desire to understand human learning processes, but how this ambitious goal quickly gave way to what's known as the brute force approach in which millions of possible board positions are systematically evaluated for the best possible move. The result is that there are now chess programs capable of playing at grandmaster level on a powerful PC and the question of the superiority of the human player or computer has been resolved in favour of the latter.

But Kasparov does not leave the matter there and this is where the article is so interesting. Though he was defeated in a chess match by IBM's Deep Blue in 1997, there is no resentment in his critique. While bemoaning the contemporary pragmatist approach in which 'the dreams of creating an artificial intelligence that would engage in an ancient game symbolic of human thought have been abandoned', Kasparov remains true to the conviction that research into game playing machines is as much about understanding human thought processes as it is about demonstrating the calculating power of computers. And he describes a new form of chess competition - advanced chess or freestyle chess - played between teams of human players using computers. Some of the insights gained from such competitions, in which the only constraint is the time available to each team, are fascinating - but for that you should read the article for yourself.

Reading the Kasparov article brought to mind a computer chess project of my own - mobility chess.

The way chess programs work involves the computer identifying all the possible moves available to it and, for each of those moves, all the possible responses available to the other player and so on, with the resulting board positions proliferating like the leaves on a tree. The number of moves that the computer can look ahead depends on the computer power available. In practice, the search is refined by considering only sensible moves on each side. Each of the board positions examined is given a score according to an evaluation function that assesses the value of the pieces and their positions. The computer selects the move that will lead to the board position with the highest score despite all the best efforts of its opponent.

The unique feature of mobility chess is the extreme simplicity of its evaluation function. While most evaluation functions attempt to assess the inherent strength and security of individual pieces or patterns of pieces, in mobility chess the only thing that counts is how many distinct moves there are available to choose from. For a given board position, the program simply counts the number of alternative moves available to the player whose turn it is and subtracts the moves available to the other player.

I wrote the program and played it a few times. It played an interesting game considering its sole goal was to keep as many options open as possible. It's principal weakness was an annoying tendency to develop its rooks too early but then that's something I recall doing myself when I first played chess as a child. Since the evaluation function in mobility chess is so simple the program has time to examine more positions - that is, to look further ahead. Perhaps looking a little further ahead would be sufficient to demonstrate the folly of early rook development - who knows? It would be an interesting exercise to develop a common framework within which different evaluation functions could be played off against one another. Would the winner be the function that took a very sophisticated (if computationally costly) view of a position or one - as in mobility chess - that is almost trivially simple?

There's no doubt which outcome I would favour; I'm drawn to that science (or is it art?) that concerns itself with ways in which complex behaviour emerges from very simple rules. But, of course, complexity is not guaranteed. Mobility chess might equally well turn out to be extremely boring. I haven't played it enough to guess at how things might turn out.