Friday, July 28, 2006


Well, I made it. No hijacking, no muggings (yet), but they lost my bags. When the carousel stopped and the official looked at me and shrugged I just had this feelinf of fatalistic acceptance. It could have been worse. It probably will be worse, in due course! They say I can have then back tomorrow (after another taxi trip to the airport and back, for which I'd be willing to bet no-one will offer to pay).

So I finally made it to the beach...dressed in heavy jeans and a distinctly anti-social tshirt after 27 hours in transit. There's a cosmic consiracy here to keep me pale!

Is Southern California Hell?

It’s certainly hot enough at the moment to qualify. And tree-hugging anti-consumerists would probably embrace the idea. But I’m thinking more pragmatically, not to say selfishly. After the week I’ve just had, I’m starting to believe it is.

I’ve been to California many times, and I generally like it. Palm trees, sun, sand, surfers…just like home. But this time was different. My first mistake was deciding to take the bus from Vancouver to San Diego. I’ve already covered that, so I won’t go into detail. However, after two full days in transit, I was ready for some time on the beach. Except that it was overcast and cloudy the whole three days I was there. Never mind; there are great bookshops and bars, and I managed to fill the time very pleasantly. Come Wednesday morning, and things started to go downhill again.

I was booked on the 10:00am bus to LA. I made it (just) and the trip was uneventful, if a bit tedious. I swear I’m going to take up writing mysteries…some of the junk I’ve read, in desperation, over the last few weeks must have been written as part of a third grade English assignment! Three hours later I’m in LA, facing the next decision. How to get to the airport? I ask around, and it seems the choice is a taxi at around $40, or two buses for $2.50. Call me cheap, but I can think of better uses for $40. Even in California, that’s a fair amount of beer! So I go with the bus option. Two hours, two buses but only one missed stop, and I’ve seen an awful lot of drab real estate and met some…interesting…people. Talking to an American friend later, he assured me that only vagrants take the bus in America. I know what he means. The bus driver was very sweet, though, even if she did forget to tell me which stop to get off at. It just meant I had an extra half hour to talk to her, as we went round the route again.

The metro bus drops you off miles from the terminal, but it only took me 20 minutes to figure out where the shuttles went from and lug my bags over there. Did I mention we’re having a record heat wave? From the airport another shuttle took me to the hotel, where, for the price of a beer, I dined on free bar snacks.

Fast forward to this morning. The hotel insisted that my room was not, as arranged, prepaid, and went ahead and charged it to my credit card. (Actually, I was amazed it went through!). Arguing about this meant that I arrived at the airport with very little time to spare, but fortunately the lines at the check in were moving pretty fast. At the check-in desk I encountered the next check. Apparently, in order to enter Brazil on a visa, you have to have proof of the fact that you’re leaving again. And my return ticket is an eTicket, and I hadn’t printed a copy of the itinerary. I’ve never been deported, and I’m usually pretty open to new experiences, but I still haven’t made it to the beach, and the North Sea is just not the same. The guy behind the desk was very nice about it, but firm – I can’t enter the country on a one way ticket. Somehow I have to show him my itinerary, which I have in my email inbox. I need the internet.

Picture, if you will, a crowded departure area in LA airport at 7:30 in the morning. Amidst the lines of travellers and piles of luggage, in front of desk #14, a middle aged scientist and a uniformed airlines clerk are kneeling on the floor, trying to persuade my laptop to hook up to the network, using the network cable he unplugged from the back of his computer. I think it’s safe to say we caused some interest. After 20 minutes we realized it wasn’t going to work, so he went off to confer with superiors. It was only after he’d got permission for me to go to the gate lounge and use the wireless there that I realized that I did, in fact, have a paper return ticket, at the bottom of my handbag. I have to say, he took it very well!

I’m now en route to Miami. The way today’s going I’m reasonably certain that either the plane will be hijacked or I’ll get mugged on the way to the hotel…

Tuesday, July 25, 2006

Obedience to Authority

I spent yesterday evening sitting in the Rock Bottom bar, sampling their microbrews (do they count as microbrews if it's a national chain?) and reading a reissue of Stanley Milgram's classic "Obedience to Authority". It's a totally scary book, and makes you look at human nature quite differently, as does much of Milgram's work[1]. I think he's very much under-appreciated outside the psychology community.

To recapituate quickly: Milgram set up experiments consisting of a 'teacher', a 'learner' and an 'experimenter'. The learner and the experimenter were in cahoots, and the real subject of the experiment was the teacher. The ostensible aim of the experiment was to investigate the effects of punishment upon the learning process, but the real aim was to examine under what circumstances people would obey authority, even when told to do things against their moral principles. The teacher had to read a list of word pairs to the learner, and then punish the learner with an electric shock when the learner got the pairs wrong. The strength of the electric shock increased every time. The learner, although not actually getting shocked, acted increasingly distressed, demanding to be let go (he was strapped in), screaming in pain, an so on.

They varied the experimental conditions in a number of ways, but the results were amazingly consistant. Around 60% of people would inflict pain up to a level clearly labelled 'danger' on the board they were using, if told to do so, even though most of them were clearly distressed by what they were doing. Milgram discusses possible reasons why people behave this way in considerable detail, but the bottom line is: people obey authority, even to the point of doing things they would normally find morally repugnant. And this was just a university-run experiment, for which they were paid $4.50! I should imagine the effect is even stronger for systems with real authority, such as prisons, the police and the armed forces.

It's a really depressing thought. The torturers of the Spanish Inquisition, the witchfinders, the guards in the Nazi concentration camps, the American soldiers in Abu Ghraib were probably all nice, normal people who loved their mums and went to church on Sunday...and found themselves in a situation where they behaved with cold-blooded cruelty to total strangers.

I haven't quite finished the book, so I don't know if he discusses it, but I've heard that many of Milgram's subjects were seriously traumatized by the experiment, and the realization of what they, personally, could be made to do. After this, the ethical rules were tightened, and this sort of research would no longer be approved. He does make it clear, though, that he had no intention of damaging his subjects; before he started he expected that most people would simply refuse to participate once the 'learner' started objecting. Fascinating, disturbing stuff.

[1] Milgram also did the famous 'six degrees of separation' research, which is a significant part of network theory, and so this post is *almost* relevant to my research!

Sunday, July 23, 2006

In Transit

Well, CEC is over. It turned out to be a huge amount of fun - great plenaries (on the whole!), quite a few interesting sessions, and of course the chance to network (it's professional networking, ok? Not just a chance to gossip over a beer!) with interesting people I only get to catch up with at conferences like this. The computational biology sessions, which took up all Thursday, were the highlight for me, of course. I gave two presentations, and actually got some questions, which is always a good sign. Your heart sinks when the session chair says "Thank you! Any questions?" and there's this wooden silence. After a minute or two the chair says "Well, I have one..." and you can see him or her desperately trying to remember what you were talking about. If all else fails "Where do you see this research going?" tends to be trotted out. I've been in that position as session chair myself. Fortunately, it didn't happen this time...lots of questions, to which I may or may not have given comprehensible answers. There were several really nice talks in those sessions, some of which will be useful for a review article I'm writing. I'll talk about that in more detail later.

I spent the last two days on a Greyhound bus. Left Vancouver at 8:30am on Saturday, and got to San Diego at 10:00pm on Sunday. I don't think I've ever been so glad to get into a bath! There's such a lot of gorgeous country down the west coast - one of the reasons why I chose to take the bus - but you get so tired sitting all cramped up for hour after hour that you really don't appreciate it.

I've figured out how to sleep sitting up on a bus, though. There's nothing to it, as long as you have a window seat. You fold yourself up, concertina style, and rest your head against a vibrating glass windowpane, down which a freezing draft is directed by the air conditioning system. Then you wait until all your muscles go numb and your thermoregulatory system shuts down, at which point you can doze lightly, disturbed only by the hugely fat guy beside you turning over and overflowing into your seat, the screams of the inevitable baby three rows up, and the woman directly behind you coughing her drug-resistant TB at the back of your neck. After eight hours of this the sight of a Burger King in the middle of nowhere is suddenly a vision of Paradise, rather than the plastic Hell you previously thought. There's nothing like queuing up for semi-congealed French fries and watery coffee at 6am to make you appreciate life in the salt mines...

I'll be in San Diego for the next few days, and I have two book chapters to write, so if I can tear myself away from the beach I might actually manage to do some work. Failing that, the blog may have to revert to a description of the sufers and bars of Pacific beach!

Friday, July 21, 2006

Self Driving Vehicles

We had the most entertaining plenary on Wednesday - "Winning the DARPA Grand Challenge", by Sebastian Thrun from Stanford. The challenge was for a totally autonomous vehicle to cross 140 miles of desert in under 10 hours. In 2004 no contestant made it further than about 10 miles, and most people (including me!) concluded that it was currently impossible. In 2005, five teams made it [1] including, of course, the Stanford team, who won $2 million for their efforts. Sebastian gives an excellent talk (he's clearly given it a million times, and knows exactly how his audience will react), and has a fascinating story to tell. Makes you want to get out there and into robotics!

Next year's challenge is to complete a 60 mile urban course, including other moving vehicles, in less than 6 hours [2]. The Stanford team is entering again, and Sebastian is very enthusiastic about the potential for self-driving cars. Listening to him, I was very sceptical...after all, these things are programmed by people like me! (Well, probably not just like me; no-one with any sense lets me anywhere near a Java compiler. I have to lock myself in my office and code when no-one's looking). My gut feeling is that cars driven by people are going to be safer than drive-by-wire. But Sebastian disagrees, and after all, he's the expert.

He also made a point which I hadn't considered, and which brings us, however tenuously, to aging. He discussed how devastating it is to an elderly person to lose their driving licence. It cuts them off from their social circle and, in today's car-oriented society, makes them dependant upon carers. But for the safety of others it often has to happen; we went through this whole business with my husband's grandfather, and I'll bet it happens to most people eventually. But self-driving cars would mean independance for people with Altzheimers, visual impairment, general frailty...all of a sudden it's a very appealing idea. So I'm cautiously on the side of automated cars. As long as they don't let people like me program them!


Tuesday, July 18, 2006

Human level intelligence

Day one of the Congress on Evolutionary Computation, and the theme was artificial intelligence. Talks, panels, a's a lovely idea, but I can't help feeling we're being just a tad optimistic here.

We're all engineers here (this conference is run by the Institute for Electrial and Electronics Engineers), so we all share the basic assumption that if a physical phenomenon exists, we can reverse engineer it, given enough time, technology, intelligence and (of course, because we're still academics) money. Intelligence is a phenomenon, arising in complex systems; if we can build a system complex enough, it will be intelligent. I have to admit, I have a lot of sympathy for this attitude. I see no reason why intelligence should be tied to wetware instead of hardware. Reasons may exist, of course, but we don't know them yet, so I think we're justified in exploring the possibilities. But surely it would help if we actually knew what we were looking for?

Intelligence is like pornography: you can't define it, but you know it when you see it. That's the working definition used by most of the computational intelligence community, and it does leave us in an awkward position. Alan Turing suggested the Turing Test as a measure of "human-level intelligence" in 1950(1). The idea is simple: a computer and a human communicate with an interrogator via a keyboard. The questioner can ask anything he likes. The computer passes the test if it can convince the guy on the other side of the keyboard that it is the human. Simple, understandable...and as far from being achieved now as it was when Turing proposed it. Consider industrial robots, 'helpful' computer programs, game there a skerrick of actual intelligence there?

Even a cut-down version, known as the Loebner Test, was never really convincing. Why has human-level machine intelligence been so hard a problem? I can see a number of possibilities:

1. In some way we don't yet understand, intelligence is a property of brains, and cannot be mimicked on other hardware;
2. Our hardware isn't yet sophisticated enough;
3. Our software isn't yet sophisticated enough;
4. Our working definition of intelligence is so vague / off-the-mark / completely wrong that we're working in absolutely the wrong areas, and missing something really obvious;
5. Some combination of the above.

Obviously, I go with option 5! The feeling around the conference is clearly positive: human-level machine intelligence is just around the corner. Indeed, one talk we had claimed that it's not just around the corner but over the doorstep, and making itself at home in an armchair by the fire, but I have some reservations, so I'm not going to relax just yet. On balance, I go with 2. and 4. We need *much* more complex hardware, and we need to know what we're doing with it. And it's not just around the corner, even if I do quite like the AI who's training me up in Eve Online. She has such a reassuring voice!

No ageing, but I think it's relevant. One of the greatest fears everyone has about the ageing process is its effects on intelligence and memory(2). These are issues worth contemplating by biologists and engineers alike.

(1) Good Turing Test page at
(2) Hey, I get to plug a book! Not mine, but by a friend of mine; if you're worried about your ageing memory (I've been worried about mine since I was 19) try The Memory Book:

Monday, July 17, 2006

Sleepless in Vancouver

I'm currently at the World Congress on Computational Intelligence in Vancouver, Canada. The conference starts today, so apart from two free glasses of red wine at the Opening Reception last night I really have nothing to report yet. Should be fun, though; a few old friends, and some interesting looking talks.

Getting here was interesting, to say the least. For some reason, I seem to be a bit disorganized this trip (a state which my friends and family will say is chronic, but they're mistaken). Anyway...I'm going to Fortaleza in Brazil after this, so I needed a visa. Because of delays in funding the flight (there's a world of pain here, I'm just going to gloss over it) I ended up applying for the visa the week before I left. That meant that I had to pick up my passport from the Brazilian Embassy between 3:00 and 4:00pm on the afternoon before I flew out of London. Someone's looking out for me, because it was ready, and I got to spend a very pleasant afternoon on a floating pub on the Thames, with my visa-adorned passport safely in my handbag. Flew out the next morning, and apart from the realisation that the bar at the airport in New York charges $10 for a beer (which I only found out half way through the second one. I was so upset I had to have another one to calm my nerves. Fortunately, there was a special offer - a shot of whatever you choose with each beer for a mere additional $2!) I made it to Vancouver more or less in one piece.

Because the university is not paying for all of this trip, I'm staying at a hostel in downton Vancouver. I arrived at midnight, in a less than lively state, and checked in. I'm in a dorm with three young women, and, quite frankly, it looks (and smells) like a teenager's bedroom. Well, the bedroom of three teenagers. Possibly a teenage boy's dream, but for me the whle setup lacks charm. I made my bed, went through my talk for the conference, and settled down for sleep. At this point it was 2am. On a Friday. By 3:30am the influx and efflux of drunken teenage girls changing their clothes (did I mention it was a teenage boy's dream?) had settled down, and I managed to doze. Until 5am, when some guy starting banging on the door and calling for Angie. Since we didn't seem to number an Angie amongst our ranks, he eventually went away. At 5:30 I decided there was no point trying to sleep, and went for a shower. I could get my own back by snoring, but as far as I'm aware you actually have to sleep to do that. Maybe tonight!

Turns out there's really not a lot to do in Vancouver at 7am on a Saturday, but the staff at the local Starbucks were very nice when, in my stupor, I managed to spill a Grande Americano all over their newly-cleaned premises (and my laptop bag, which now has a pleasantly evocative breakfasty scent that I suspect I'm going to have to live with for a while).

Conference starts today. Not a lot of ageing related stuff, but some interesting talks are lined up. I'm looking forward to it.

Sunday, July 02, 2006

Disposable Soma

I'm firmly convinced that for a biological theory to persist it has to have a catchy name, and the disposable soma theory certainly meets that qualification! Mind you, it's a good theory, too. At this point I should add a disclaimer - it was proposed by my boss(1) in 1977(2) (but that's not why I like it; I'm always happy to critique the boss!).

So, the basic question is "why age"? Why not live forever?

The first issue is that, in the wild, creatures almost never die of old age. You get eaten, or run over, or die of disease long before the grey hairs start to intrude. So, if you, as an organism, can take it as read that you're going to die, it makes evolutionary sense to maximise the spread of your genes. The ones you carry will die with you, but you want as many copies as possible out there. Given that you have a limited energy budget (as every dieter is well aware), the question becomes - how much do you expend on repair of the inevitable damage that's done to cells and chromosomes, and how much on reproduction? The disposable soma theory addresses exactly this question.

And the answer, of course, is different for different organisms with different lifestyles. You can live fast, die young and leave a beautiful corpse (and a lot of offspring), like the many sea creatures who contribute to the plankton layer; or put lots of time and energy into a small number of babies, as humans do. The best strategy depends on your lifestyle and evolutionary niche. Since the soma (body) is inescapably disposable, all the issues of game theory come into play - which is another story, for another time.

(1) OK, one of my bosses. As a minion in the university hierarchy I have an elegant sufficiency of bosses, and just as I think I have them all tabulated, I find there are more. One of the joys of academia, I guess...

(2) Kirkwood, T. B. L. 1977. 'Evolution of ageing', Nature vol. 270, p.301 - 304.