Grand Challenge or Toy Story: Making complex systems useful

 

I have something here to show you. It's right here in my pocket. See. You do recognise it, don't you? It's the Tardis. You know, Dr Who's time-travelling phone box. It may be a bit difficult to see from the back of the room, because it is awfully small.

I got it from cyberspace. I downloaded it from some funky website called www.who.alpha.centauri. I used that new rtp tool. Haven't you used that yet? It's the reality transfer protocol, and its kinda neat. And here it is, the original Tardis, small, but perfectly formed. Complete, every point is there, and being points, they're not even more squashed up than when it was bigger.

It must be the real thing, not a model, because I used the internet, and everyone knows the Internet never trashes your files during transfer. After all, the net was designed by the folks who brought you the charms of Unix, the operating system for the Challenger space shuttle.

I thought we would do a little time-travel this morning, that we might become Time Lords for a day. Well, Time Squires, really. A Tardis this size is more a manor house than a castle out there in hyperspace. But if we are going to cut class like this, we'll need some sort of excuse. Naturally being scientists, this is not difficult - ad hoc, post hoc and even pre hoc rationalisation is our bag. That's what we do best. Maybe we got to be scientists in the first place by dreaming up ever more fantastic reasons why we missed the class or didn't do our homework: 'I'm really sorry, Miss, but I left my homework on Alpha Centauri'.

And here's our excuse - we need to do some time-travelling because we need to have a look at how we got into our current mess - why we're down here at the bottom of science's totem pole and all those single discipline traditionalists are up there on top of us. Or to put it another way - how can we get our snouts into life's trough. All we get to see are a row of fat bums and little corkscrew tails, while they've got their snouts in the trough and probably their front trotters too. The pigs. It's just not fair, it's just not right. Someone oughta do something. The government oughta do something. Where's the level playing field around here?

But before we time-travel, we need some context, so we won't be disoriented when we arrive. This is like getting shots for space-lag. It won't hurt, I promise. Or only a little bit, but it's for your own good.

And the context, of course is 'Grand Challenge or Toy Story'. Everyone knows the Grand Challenge stuff - heroic science, hard problems, big computers, new maths - but Toy Story, the new Disney movie - the world's first movie made entirely by computer - has that and more: it's got a plot - or as much plot as any Disney movie, but then it is made for children and Americans. Indeed the similarities between Grand Challenges and Toy Stories are not nearly as interesting as the differences, but for reasons that might be more different than you think.

Sure, they both use big boxes and cocktail maths to attack complex systems, and they both push the frontiers of visualisation and animation. Both have cowboys, but Woody - Disney's cowboy hero - is quite a nice guy.

But only one is useful in that it actually pushes the game forward. Only one results in that Eureka! for the rest of us: Why didn't I think of that! And that of course is Toy Story. Never mind that it makes money (I can't believe I'm saying this!), that it provides a revenue stream to convert into a liquid nitrogen stream to top up old Walt's vat. It actually uses lots of supercomputing, lots of science and lots of maths to do something useful. Before this, the only thing that ever got done in this way was global weather forecasting or modelling the big bang. Need I say more. The weather guys can't do any better than my great aunt with her arthritic hip, and the cosmologists can't find 90% of the mass of the universe. Maybe they should swap jobs: 'We can't find 90% of the weather, but it should be fine tomorrow'.

***

There's our context. That wasn't too bad. Now we can hit the road. Here's the Tardis. You may be wondering how we are all going to squeeze in, it being so little and all. Well, that's the easy part, we just need the services of a topologist. He can shoehorn us in, no problem. All he has to do is to declare that this room is homeomorphic with the inside of the Tardis, and then we're in. Trust me, I'm a scientist. I hang out with topologists, differential geometers even, they do this sort of stuff all the time. It's sort of mathematical bungee jumping. In fact, we don't even need a registered topologist, we can declare it ourselves.

What we do is just say all together: 'This room is homeomorphic with the inside of the Tardis'. Ready.

THIS ROOM IS HOMEOMORPHIC WITH THE INSIDE OF THE TARDIS

See, we're in. Point for point. Isn't mathematics wonderful! Isn't it useful!

Now we need to get this little baby moving, relatively speaking. As you can see, it's fully equipped, beautiful matching Planck and Hubble constants, although that Heisenberg drive looks a bit iffy. Here we go, easy now.

We're going back a half century to have a little look around. And here we are already. We won't get out, we'll just look. Looks pretty ancient all right. But wait a minute, everyone over there is asleep. Sleeping in front of . . . TVs! God, this is the fifties, not the forties, I'd recognise those genuine Danish-style TV recliner chairs any where, or at least any time.

I knew you couldn't be certain of that Heisenberg drive. We only travelled about forty years instead of fifty. That's OK, ecologists would just call this field measurement error. Actually, that's a lie, about the ecologists, I mean. They never really measure anything, they just make it all up. Their field trips are really just trips to the bar, where they sit down, order a beer, and copy the numbers from the bar tab into their field notebooks. That's why ecology papers are so awful to read, why ecological data always fits models perfectly, and why ecologists spend so much time in the field. It's also why, once they cotton on to this scam, students all want to do field work. But don't tell anyone though.

But look over there, those guys look asleep, but where are the TVs? Got it, it's a AAAS meeting. This looks awfully like the famous 1954 AAAS meeting in Berkeley. Now they were glory days for science. Korea was over, Vietnam was still a French war, no Sputnik yet to shake our confidence in the stately progress of Western science, students wore ties - not tie-dyed - on the campus, and everyone else was asleep in front of the TV. Berkeley was still just a campus, not a place to 'get your shit together'.

So what are they doing here? Let's listen in. Hey, these guys are reading their papers to each other and talking about forming a Society for the Advancement of General Systems Theory. As they said in Butch Cassidy and the Sundance Kid, 'Who are those guys?' I can just recognise a few. There's Ludwig von Bertalanffy, the allometrist, and Kenneth Boulding, the economist. There's Nick Rashevsky from the Committee on Mathematical Biology at Chicago, and Anatol Rapoport, the mathematician from Ann Arbor, and folks from Bell Labs and the London School of Economics and the Zoology Department at Cambridge. There are chemists and psychologists. This is quite a scene. There are even a couple of post-modern (in a strictly fifties sense) flaky Liberation Biology types hyping the general systems nature of the rights of man, for chrissake.

Have we lucked out or what? These guys are just like us. They could be our ancestors.

And here they are talking all kinds of neat stuff and forming a society and all. They've got papers on topology in biology, on closed and open systems, on entropy and information, on general systems as a new paradigm for science (but of course they don't use the word paradigm, Kuhn is still in intellectual nappies right now). They are also talking about models and modelling, adaptive and random systems. No kidding: Anatol Rapoport has a paper called Connectivity of random nets. Wow!

They are even talking about complexity. How about these section headers: The concept of complexity; Complexity in biological systems; Systems with the property of evolution; Does learning involve an increase in complexity; Complexity in time. This is all from the paper by Pringle, the guy from Cambridge. And he's a behaviourist who quotes Volterra the mathematical ecologist, Schrödinger the physicist, Shannon the mathematician, Fisher the geneticist, and of course von Bertalanffy who was to become the chair of this society. So this Pringle, whom I have never heard of, uses all the flash buzz words, cites widely across the disciplines and knows his grantsmanship. And this is all lifted from his paper. I bet these guys even gave the rest of science the shits with their smug, superior, nah-nah-we-know-something-you-don't-know airs, just like we do.

So maybe our meeting today should really be the 50th meeting of the Society for the Advancement of General Systems Theory, and not just the 3rd meeting of the Complex Systems Society or whatever. But it isn't. There are big questions here. Like: Did these guys stuff up? Did we? How long till lunch?

This clearly requires a serious historical analysis, and it's hard to find a serious historian when you need one. You'll have to put up with me, and I'm no Umberto Eco. That is, I really can't do that semiotic gig, and get down and get dirty symbol-wise. History is easy, actually. Even easier than ecology. You don't have to make up much stuff, you just read it off the dial of the old 20-20 hindsight meter. So unlike ecologists, historians are always right. And because they don't have to copy down all that data from their bar bill, they should have even more free time for drinking and making up really big lies, and not just teeny ones like scientists.

But history really needs comparative work if it is to be truly awful. You can get a bit tired and emotional from looking at only one set of fossils, and, as every paleontologist and anthropologist knows, you need two data points to fit a straight line - or, at least, a fifth order polynomial.

***

So it's back into the Tardis, and time to hit the old fast forward button. We'll try for half way between then and now. Whoa there, let's try for a smooth landing. Good. Now, where the hell are we? Take a dekko. Looks good, lots of flares and paisley. Must be the sixties. That's good enough for government work. Anyway, we all know that the sixties really went into the mid-seventies - from the Beatles' Love me do in 1962 to that day in 1975 when the last chopper lifted off from Saigon - the day the music really died. Now you can see where some ecologists learnt their measurement theory, and a lot else besides.

Now what have we here? Ooo boy, looks like another meeting, and looks like scientists again. You can tell - some geeks not only have pocket protectors but also HP calculators on their hips. These are fully badged theoreticians, I guess. But this place has class, style even. Can't be the States. And the people are talking animatedly - on drugs maybe.

I recognise some of them. There's old Wad Waddington, the geneticist and Rene Thom, the topologist. Rashevsky's not here, but there's Rashevsky's push, the two commie dicks, big Dick Levins and little Dick Lewontin. And isn't 'push' a great Aussie word, heavy and sinister. So much better than 'mob'. 'The Chicago mob' sounds to Aussie ears like a heap of sheep. And sheep these guys are not. They are on a mission from Marx. Worms in the apple of capitalism, the Rockefeller sponsored, n-star castle of Serbelloni at Bellagio on Lake Como in Italy. An early example of Gucci Marxists. For this is where we are, and this is one of the great meetings organised by Waddington for the International Union of Biological Sciences. They were called 'Towards a theoretical biology'.

The guest list reads like a Who's Who of science, because there were not only geneticists and evolutionists but also physicists, chemists and mathematicians. It was where Thom's catastrophe theory was first foisted on an unsuspecting English-speaking world. It was where Turing machines met epigenetic landscapes, where statistical mechanics was first used for biological systems, and where critters like 'self reproducing automata' first acquired evolutionary habits.

The only ones who weren' there were Robert Macarthur, the ecologist, and Bob May, the physicist. Macarthur was busy dying and his hagiographers were busy fighting for the right to interpret his work to the world, too busy it seems to attend the coolest meeting of the sixties. May was still pupating, physicists being larval ecologists, though most, like axolotls, exhibit permanent neoteny. So the meeting was bereft, thankfully, of chaos.

Great sixties stuff, really, when you look at it. Stream-of-consciousness writing. The odd poem in the proceedings. Lots of purple prose and 'we can change the world'. But these guys had the smarts to do it with their own French chef and silver service. The only way to bring on the revolution.

Forget the karma for a moment and focus on the gall. They had more front than Myers. You can just see Waddington going up to the IUBS and saying: 'I've got this great idea to help solve the world's problems. Give me a lot of money and I'll invite all my friends to Italy for the summer. We'll sit around and damn well solve those problems, no matter what it takes'. Sort of Haight-Ashbury by the lake, and beats the hell out of wading through rice paddies in Vietnam shooting at guys in black pyjamas - who shoot back. God, how I would have liked to go to those meetings! But they were only for the generation in charge, the generation who got to send my generation to the rice paddies, while they got plastered by the shores of Lake Como.

In fact this was such a good idea, or Waddington had so much gall, or he had the smarts to invite a Rockefeller or two, that they did Serbelloni the Sequel three times. Was this where the makers of Jaws and Rocky got the idea from? Anyway, they were so good at working through their problems, that they worked through them lots of times. In fact, that was what you did in the sixties, you just kept on working through stuff till you got it all together.

***

Now we've got our two points, we can do a Poincairé and build an elephant, or at least a bit of an elephant.

Look again at the General Systems push. What did they really have that the traditionalists didn't have? They had a few neat tools such as graph theory and information theory, but so did the traditionalists. They had some big, hard problems, or at least they said they were big and hard. But everyone says their problem is big and hard. That's what macho science is all about: saying mine is bigger and harder than yours. Most of all they had hubris. They said they were looking for the grand synthesis - the theory of everything - and they were damn well gonna find it. Not everyone does that. But they did have a penchant for stumbling on productive pairwise synergies between themselves while they were looking for the big gig. So, all in all, they failed pretty miserably, but that's no big deal. Science, after all, is just a history of failures, not successes, just like life is a history of extinctions.

They did some good though. They engaged with the broader society and tried to look at problems which were both real and complex, like ecosystems, economies and brains. Even if some of their engagement meant assisting at the birth of Liberation Biology, and not putting it in a sugar bag with a brick, and tossing it into the nearest creek. If we wanted intellectual ancestors, these guys are not too bad. They are by no means Neanderthals, even if they are not card carrying Cro-Magnons.

And what of the Serbelloni push, Waddington's boys? They had more tools, they even had sexy tools, like catastrophe theory, that not only made the earth move, but also promised to respect you in the morning. They were just so sixties: utterly promiscuous, utterly pleasurable and no long-term side effects.

But like the times, they tuned in, turned on and dropped out. They didn't really engage - complex problems yes, but real problems, no way, man. In harder times, like today, we might say they wimped out. No hubris, though. They were about their own heads, not everyone else's.

Can any of these guys be our ancestors? Have we built an elephant yet? We seem to be missing bits. Shouldn't our elephant have grand synthesis and cool tools, practical engagement and a seductive muse? Is this today's upfront, in-your-face science of complex systems while yesterday's is - well- like yesterday's news, consigned to the great recycling bin of history? Have we finally got the bison drawings right in our Lascaux caves, our labs, and divined their secrets thereby, while the Neanderthals down the valley just keep hunting them in the same old way? In your dreams!

***

It's time to tell you what we've really got, and what we should have, and how to get it.

So we come back to Grand Challenges and Toy Stories and how von Bertalanffy and Waddington map onto them. Von Bertalanffy and company were big on the former, while the Serbelloni guys were really on about the latter. Both failed, and so will we if we don't learn from history.

The first thing to learn is that hubris is OK, but it doesn't bring home the bacon. It needs to be focussed, engaged, to be useful. Theories of everything can waste a lot of time better spent on working on problems that are really challenging, and that have at least a chance of being cracked.

And we have such a problem right in front of us. It's called Australia. How do we sustainably develop this country? That is, how do we encourage economic development while maintaining environmental quality? For every Grand Challenge problem you can think of - evaluating the mass of the proton from quantum chromodynamics theory, predicting the tertiary structure of a protein from its primary structure, predicting global climate a year ahead - this one beats it: more complex in every way, more entities, more attributes, more scales, more interactions. And progress of the right kind has already been made.

Want to know how groundwater dynamics interact with vegetation dynamics, local economic forces and human social systems to predict the changes in salinity in irrigated croplands? Want to know that in real space at tens of metres resolution over hundreds of square kilometres? We are now closing in on how to do it.

Want to know how climate interacts with soil moisture through the vegetation cover? Want to predict soil moisture months ahead because that knowledge is the billion dollar advantage for Australia's agriculture? Want to be able to do that at the paddock scale for the whole continent? We think we will be able to do this in the next year or so.

Want to be able to model the dynamics of Australia's forest ecosystems taking into account their wood values, their conservation values and the values and desires of all the human stakeholders? Need to do that at the scale of hectares for the whole continent? We can do that.

Want to understand the dynamics of the economies of APEC, the Asia Pacific Economic Cooperation group, as they each move at different speeds towards a single market? Need to understand the feedbacks involved when the individual countries each model each other and change their behaviour thereby? Want to use neural nets and other adaptive techniques in their biggest application ever? We can see how to do that.

Sustainable development has come from nowhere to be the overarching political problem of the post-industrial, post-Cold War world. And better than that, the modelling and analytical paradigms have moved in a useful direction.

The first efforts - remember the shrieks and alarums of the Club of Rome - all converged on the Keynes Point: in the long run, we are all dead. But modern workers are making a vigorous effort to converge on the Bucks Point: that's not Buckminster Fuller, but dollars, the traditional conclusion of all science - more research (read money) is needed. Call it venal, call it crass, but it sure beats working for a living. And as a sign of engagement, dollars are society's ultimate blessing. Likewise, you can bet that a focus on the dollar is a sure sign of a science reaching maturity.

Let's look at the distribution of those dollars a little more closely. Let's do a Feynman approximation here. But before we do, we need to introduce some scaling factors for our overseas guests.

US visitors should multiply all numbers by 10 or 20, America being that big and that rich. NZ visitors should divide by the same amount, or replace all numbers by the NZ monetary constant, which is two small flat rocks.

Now where are the research dollars in Australia? Well, ARC, the Australian Research Council, has about $300 million highly contested dollars to spend each year. But it is the last redoubt of the traditional disciplinary scientist. And anyway only about $80 million goes to significant leading edge stuff. That is some crowded trough. And none of these pigs is about to move over to let a new little pig get his snout in.

But on the other side of this Animal Farm is another trough of about the same size, and not only is it much less crowded, but every year, there is some swill left over! I won't say they actually queue politely to get at this trough, like some piggy smorgasbord, but there is a more relaxed ambience here born of a belief that 'all animals are equal, but some animals are more equal than others', indeed a belief that some pigs have rights at the trough. This is the trough of the Research and Development Corporations, and, even though some of them have not yet discovered it, they are in the business of funding research on sustainable development. Indeed, some of the pigs busy accumulating frequent-trough points here haven't discovered it either. They are mainly csiroids, a peculiar breed of pig that believes profoundly in a full trough and priority access - a breed of pig which, though large, is not famous for the efficiency of its food conversion ratio.

We can take these guys, easy. This has got to be the softest target since an Iraqi hospital during the Gulf War - all those big red crosses saying 'X marks the spot'. And the csiroids are not exactly Saddam's Revolutionary Guards. A coordinated attack on the coffers of the R&D Corporations could release funds for complex systems research the like of which has not yet been seen in this country.

Which brings me to the second point. If we can learn the lesson of General Systems - to transform our hubris into something useful, to update the Grand Challenge theme - then we can learn from the Serbelloni crew to make our products sexy, to do the Toy Story gig. Whatever else it was, catastrophe theory was the cocktail maths topic for nearly two decades, it was that hot. Only chaos has come close since.

***

But look at the time! We must get back. We need to get back to our today with its cellular automata and genetic algorithms, its neural nets and artificial life, with it mixture of hubris and history. And we need to ponder if there is any more to this than more mixed metaphors than a drunken thesaurus.

We also need to do lunch. We need to get our snouts in a real, not a metaphorical trough. Alexis de Tocqueville must have been like me - in balding middle age - when he wondered what other pleasure but food could a man get reliably three times a day.

As theoreticians, we need to do some thought experiments, of which, as everyone knows, there are two types. There are gedunkin' experiments, which you think up over donuts and coffee, every computer nerd's staple diet. And there are gedanken experiments, which you think up over a bottle or two of an obscure variety of Traminer. And as jokes go, they don't come much gewurs than that.

And while we are meeting, I think we should consider two more things. The first is the notion of what makes good science better, and my suggestion that it is engagement: engagement with other parts of science, which we do, I argue, better than anyone. But it is also engagement with society and its problems, which, I argue, we don't yet do well enough. How we can get the best of Grand Challenge and Toy Story?

And the second I think we should consider deeply and seriously, especially when we are savouring the complexity of the wines from the best vines in our part of the galaxy. It is this. If the great mathematician Erdos thought that a mathematician was a machine for transforming caffeine into theorems, then what sort of machine did Plato have in mind when he said 'In vino veritas'? And would it pass the Turing Test?

***

Ladies and gentlemen, the Tardis has landed. Welcome to the here and now. You may leave by the forward door when we have come to a complete stop. Please be careful when you are downloading your intellectual baggage, it may have shifted during the flight.

Thank you and good morning.

| Roger Bradbury's publications | Tjurunga on complexity |