 My title was, What is Computational Thinking? And I had some questions. Who needs it? Why? And how can it be learned? And I've now added, can it be taught? First I'd like to know, are there any people here who are also members of CAS, Computing at School? One or two. Okay. I have the impression I'm a stranger in some terrain with which I'm totally unfamiliar and I probably don't belong, but that's a challenge for both of us. I'm going to try to find a way to bridge the gap. I don't teach in schools but I've been interacting with school teachers by email and meetings because there's been a revolution going on in the teaching in schools. It's sort of mostly been discussion for about four years in the last nine months or so after Michael Gove, the minister, and Eric Schmidt from Google and various other people made pronouncements and the Royal Society talked about how teaching in schools has to be changed so there's much more teaching of computation, computer science, much less emphasis on ICT, the use of technology. And my concern throughout has been that all the people engaged in that debate with very few exceptions, there are some exceptions, have emphasised what I think of as an inadequate reason, a short-sighted reason, for trying to teach kids to do computing, to think computationally. They've talked about the needs of the economy, they've talked about the needs of industry and they've talked about the needs of job. Whereas I want to talk about computational thinking, which is a phrase that comes from Janet Wing about whom I'll say a little more, because that's what we need to understand the universe and to understand ourselves. And the stuff about jobs and getting more students doing computer science in universities and technology and the economy, those are all of less importance in the long run. They may be important in the short run, but in the long run, we need to educate people to understand the world. What Janet Wing did about six or seven years ago was write a paper that was published in communications of the ACM in which she introduced this notion of computational thinking. And I'm going to read out a bit of it. She said, computational thinking confronts the riddle of machine intelligence. What can humans do better than computers? What can computers do better than humans? And she elaborated on all that by saying that if people learn how to design algorithms and data structures in the way that computer scientists and software engineers do, they'll be able to do a lot more things better. And she gave some examples. But the main thing is, as someone else commented, she's standing from the viewpoint of computer science, looking outward, saying, look, we're doing these things, creating algorithms, designing systems that can work. And the rest of you can do it, too, and you'll benefit. And I'm saying she's right, but I want to add motivation that I don't think she had, although some of her later presentations suggested that she's moving in that direction. So my motivation is, as I said earlier, that learning to do computational thinking is a requirement for understanding the universe. Universe has matter. Things move around. They collide. They change position. It has energy. There's the energy that drives your car. There's a potential energy that makes this thing fall when I let go, because it was gravitation. Matter can be arranged in many forms. And biological evolution has produced an enormous and wonderful variety of forms of plants, of animals of many kinds, and also many things that we can't see because they're too small, microbes. And these things not only have wonderful forms, they also have wonderful behaviors. Have any of you tried putting a squirrel-proof bird feeder in a garden with a squirrel? Well, there's a hand going up. Many of them get advertised, and then it's only a matter of time before the squirrels defeat those squirrel-proof bird feeders. I thought we had a squirrel-proof bird feeder, which was a little plastic dish with suckers that could go into the middle of a great big sheet of glass, which was a patio window, and there was a tray with nuts on and the birds could fly and happily eat their nuts. And it was squirrel-proof for a while until my wife reported to me that we have a clever squirrel who noticed this thing and worked out that he could, he or she, I don't know which, could climb up the side of a patio window, just think of patio window with a thin plastic frame at the edge. So there's this feeder, he has this frame and this squirrel doing this and working out in advance that it can do it, getting to a height just above the feeder and then leaping across. How do you leap across when what you're holding onto is a thin portion? It's not a tree trunk, there's no branch that you can hold onto. The squirrel was able to work it out, land on the tray. How does it do what it does? Well, it's got a nice body and it's got energy and it's matter and it can apply forces, but it had to do something else. It had to take in information, it had to analyse that information, interpret it and use it in order to make a plan. And then it had to carry out that plan. And it's almost certain that in making the plan it didn't have all the information it needed about exactly how to move its hands and feet any more than you would if you decided to climb a ladder to reach something, you wouldn't work out in advance exactly where you're going to put your feet in your hands. You know you can do it because you see that structure, you see the affordances in that structure and you work out what is possible and then you can make use of that in deciding what to do. Those are just simple examples of a general point which is that life is primarily information processing. There's matter and there's energy, but the production of that matter, the growth of the body of a plant or an animal is controlled by information, information in the DNA and all sorts of intermediate stages after cells start dividing. The behaviors that animals and plants produce involve use of energy and matter, but under the control of information a plant may send its roots where the moisture is increasing in quantity or some plants will detect in which way gravitational forces are operating and redirect their growth accordingly. Ancient microbes perhaps could detect when something was in contact that was food and then let it in and if it was something noxious, not let it in so it would change its molecular structures in such a way as to affect what happened using information but also in many cases the behaviors of organisms, in fact in most cases behaviors of organisms make use of energy that the organism has and it uses information to decide how to deploy that energy. There are some mixed cases, if you look at a windmill there are two things that are operated on by the wind. Here's an exercise for you. Where is the use of information? Well usually there is a part of a windmill that gets the information about the direction of the wind and then it makes it rotate. It might be a vein sticking out of the back or it might be a subsidiary propeller, but either way that allows the wind to rotate the main structure so that the big blades are facing directly into the wind and then they can collect energy from the wind but in order to face directly into the wind the windmill has to have the information about the direction the wind is coming from and that is a human designed piece of apparatus that uses information in order to work out how to get energy which is then transferred down some sort of axle to grind wheat or whatever. We have built many devices that use information in order to determine what to do. Thermostats that turn heaters on and off and of course many things that now fly aeroplanes better than humans can because they're constantly getting information from sensors and working out what to do. But the amount and variety and importance of information processing in organisms is much greater, much deeper and very little understood I would say. It's being understood more and more and one of the results of molecular biology and the work that flowed out of the discovery of the structure of DNA in the 1950s by Watson and Creek has meant that we have learned more and more about how information is transmitted, stored and so on and used in various places but I still think there's very little we know amongst the things we need to know. Now in education one of the things we need to know is what goes on when people learn. You can do research and we've had examples today and I'm sure there'll be many more in this conference which consists of trying out various things and then testing. And you may discover that you get results you like or you don't like or whatever but I am inclined to compare that with what the alchemists used to do. There are trial sorts of experiments and they learn lots of things. If you put this kind of stuff in with that kind of stuff and stir you'll get smoke or whatever or bang or something will solidify. They didn't know what was going on until there was a deeper theory which came out of physics and chemistry about the subatomic structure of matter and the way in which molecules could be combined according to various principles that could only be discovered or could only be explained later on and even now maybe not everything is in terms of quantum mechanics. But the main point is that when you do your experiments and your observations and try things there may be thousands of things you could vary. In the case of education the environment of the child the toys the child has played with before coming to school the kinds of things that go on in the family whether they have iPads or not nowadays the kinds of interests of the family the kinds of games they play those are things you can't control you can control a few things in the lab or in the classroom but you have no idea what the space of possibilities is how many teachers 30 years ago thought maybe we will have kids learning by holding little devices with moving pictures on them that they can poke with their fingers they couldn't possibly have thought about that and yet they were doing experiments and finding out what didn't and work. What I'm suggesting is we need a transition roughly analogous to the transition from alchemists doing lots of experiments and learning things to developing a deep theory of what's going on which is roughly what happened when he got the periodic table of the elements and law of constant proportions and theories about valency and so on explaining what happens during chemical reactions so we need theories about the information processing that goes on when people learn now you might think that that's simple we perceive things, we take in information we have it and then later we can use it has anyone here tried studying artificial intelligence or designing a robot to do something? yes, one or two, not many if you want to understand learning what better way can there be than trying to build a system that learns if you want to understand flying, Leonardo thought let's try to make a flying machine well it was a bit difficult at that time it was difficult to get the strength, weight, ratios right and the modes of propulsion right and so on and in fact the way birds fly is still totally unmatched by the way human made flying machines fly although there are some things in common there are similar principles of aerodynamics and there are trade-offs between stability and maneuverability and so on anyway that's a digression can we build machines that learn? well people have been trying and the simplest ways of course are to build databases into which information gets stored then you can try to write programs that interrogate that and that's happening more and more search engines work only because there's been a huge amount of pre-processing of the data that's been collected by software that reorganizes that data into new structures which can then be used to answer questions that people type in if people had to type in things that match the stored data they'd get only a tiny subset of the responses they now get from powerful search engines so maybe something like that has to be done by a child's mind the learner we may well take in lots and lots of information but there's no point just storing that information video cameras can do that but they can't use the information in the sort of way that you want a child to or a squirrel for that matter so if you want to understand what you're doing when you teach you really need to have a deep theory about what's going on in the learner's mind and at the moment I think our theories are not deep enough I'm going to give you a couple of examples my examples are things that I don't know the answers to and I don't think anybody else does either but they seem to me examples of the kinds of questions that we need to answer if you want to be able to understand what's happening during some of the important kinds of education we give I was going to give an example using the screen I'm going to try to do it without the screen and see what happens imagine a triangle it's got a line over there a line going up to a vertex and another line coming down I haven't drawn it but you can imagine it I could use a pen and use a whiteboard but let's try it without if I keep the bottom line fixed I can move the top vertex around in various ways so if the vertex is above the bottom line and I move it upwards what's going to happen to the area of the triangle is it obvious? Will it increase, decrease or does it depend on how much it goes upwards does anyone think the area increases when I move the vertex up? Everything else means it? Yeah, one person Well, the rest of you probably thought it would but thought it must be the wrong answer or I wouldn't have asked the question Well, the answer is it does, the question is can you come up with an explanation of why it must Well, the base remains unchanged there are two sides going up to this vertex and when I move it up those two sides go outside the area that was previously there in the triangle, is that obvious? so we have two lines like that, they go up to meet a point higher up, therefore the original areas included in the new area so the area must increase what happens if the point was somewhere out there so we have the base down here, that point's going over there it's not above the base, when I push it up one side is going to remain outside and go up, but the other one will go through the triangle, I'm going to leave it to you to think about that and to think about how you can change the argument to show that no matter where the top point is when it goes up, the area must increase now, why am I giving you that example? it's a tiny fragment of Euclidean geometry, what's Euclidean geometry? a body of knowledge that was built up a couple of thousand years ago and summarized in Euclidean elements a book that used to be taught in schools to lots and lots of kids, and when I was at school about 70 years ago, well later in that I started being taught Euclidean geometry and I was taught to prove theorems and so on has anyone in this room had any experience of being taught to prove a theorem in Euclidean geometry? well I'm glad to see at least about two or five percent or something, a tiny subset now, when we were taught there were the books and there were teachers what happened before there were teachers? where did it come from? the idea I want to leave with you is that as with squirrels, and if you look hard also nest building birds and many other animals and pre-verbal children before they go to school and even before they talk they're learning all the time, they're dealing with a world in which there are shapes and motion and things can change relationships and they learn to think about them and they learn to think about them in ways that sometimes gives them insights about what has to be the case it's not just that they've tried a thousand times and found it worked, or whatever, or the parents have told them they get an insight into why it must be, just as you saw why that area must increase and I'm saying that the ability to do that is a result of biological evolution which goes beyond just storing information, collecting data, looking for correlations and it's a deep product of biological evolution, if we don't understand how it works we won't understand what we're expecting our students to learn when they gain insights which are not just correlations and learned facts but the understanding of why things must be the way they are well I have a few vague ideas about that but that's another topic so computational thinking then is something we need to do if we want to do our education well and understand why it works and why it doesn't work and to know why new technology does help when it does help or why it's just something flashy that doesn't make much difference in some cases and I will put my slides on the internet if you look for Aaron Sloman my name is in the program presentations, you'll find a whole lot of slides these will go on tomorrow sometime and I welcome comments and criticisms, thank you for your patience