Ivor Catt. 1995 transcription of talk with Rae West

28 Feb 95: I audio tape recorded 90 mins of chat, mostly on the possible unravelling of 'physics', the peculiarities of academics & Trinity College, Cambridge, and history of computers and how a company he worked for spun things out for a year until the guarantee ran out; why in his opinion array processing would be tremendous; also material on Iann Barron and other 'geniuses'. I told him my motive was to listen to me, as much as to him – he said that was a perfect reply. His explanatory style is bad – I should say worse than Harold Hillman's. Anyway, here it is. NB I found (if my script program is right) 3/4 of the talk was his, 1/4 mine.
====
IC:   Come on! Hawking, Paul Davies, Gribben–
RW:   Penrose?
IC:   Penrose! That's right, yeh. And and now at High Table not that that's a good spot the guy next to me the physicist said Penrose is a charlatan.
RW:   Sorry where was this?
IC:   Well that was in Trinity College, Cambridge, you know, with the fellows. And one of them was a physicist.
RW:   Really?
IC:   But that doesn't mean that much.
RW:   And you were at high table or something?
IC:   Well, if you've graduated from Trinity College Cambridge you're allowed to go back and have a free dinner–
RW:   A free lunch!
IC:   –with the nobs four times a year. But once a year you're allowed to take a guest. So Sue [Warman] and I go up once a year.
RW:   Oh that's nice
IC:   And and the last time, oh it was so transparent. What happened was erm I'll take you through this one first. Erm you can see the whole structure of the thing. It's a bit like the–
RW:   Shall we move into there?
IC:   What, are you recording now? I suggest you don't have the wire across the er
RW:   It's just out of interest. Otherwise I'll forget what you said | You haven't got any biscuits, have you?
IC:   Yes, yes! Here start the words of wisdom of the guru Ivor Catt on Tuesday the..
RW:   No, no, no, I'm more interested in my own contribution, actually.
IC:   presented to his disciple Rae West, who is taking the appropriate submissive posture
RW:   Bollocks. Bollocks. I want to polish up my vocal style a bit
IC:   Ha ha. Hey that's the perfect reply. That you're actually recording yourself
RW:   I've tried this with um. I think one of the reasons Hillman is so underrated is because he's too, um by the way, I've brought the [video] tape [with copies of Hillman's films]
IC:   By the way, I'm, embarrassed about that [Shakespeare audio of Burford on de Vere] tape and suggest you bring.. I know it's there.. {I think he lost it]
RW:   It doesn't matter. How much did he charge you for these things? [I see and comment on a pile of Hiram Caton's booklets on AIDS on a chair - brown-backed thin volume 'The AIDS Mirage']
IC:   Look the only thing I know about cost is one of the journal articles said it was 5 dollars. And that I learnt from somebody that an Australian dollar is a bit less than 50 p. Now you can imagine that he wouldn't be concentrating on price at the moment. I mean he's got other things to think about. That gives you a feel.. I'll send money to him later. But um I've asked him for thirty so far. I've received one, well, he gave me one. Then I received five, which are gone. Five of those are bespoke to Joy Bell. Four Sue is gonna lay out in various ways. So I'm down to one anyway. You don't want any more, do you?
RW:   No, not really. I gave a photocopy to Hillman, a bit unethical in a way, copyright, but I suspect in practice he wouldn't mind
IC:   Yeh, but you have to jump the gun because it's a matter of time
RW:   In my experience anyway Hillman doesn't actually read stuff you draw to his attention, you know, you give it to him, he looks interested, but he doesn't read it.
IC:   Yeh. Lemme take you through. So we go to Trinity. And there are all these fellows. You know, Trinity is the tops, isn't it. You know. From the point of view of snob value. If you're a fellow of Trinity. So we're among these guys and erm then erm if one of them goes up, upstairs afterwards you can all follow and have madeira and port–
RW:   Oh, walnuts!
IC:   –and the whole fantastic business. So we went up there and it so happened that er some kids had gone up there, you see
RW:   Kids? You mean undergraduates?
IC:   No, no, no! These are the proper people, but they're younger. Now what you found was, let's concentrate on one of them. He'd gone to Oxford, got a first in maths. So then he can go and do his PhD, you see, which I think he did in Cambridge. So he does this PhD in one of these theories – you know, like string theory or something. Then he does so well in that,
RW:   Superstrings?
IC:   No! No! Oh yeh! Then he's got his PhD and he probably produced something pretty good. And I put very good in inverted commas. So then he became a fellow of Trinity. No! He got a research grant, or something like that and he gets 13,000 a year and he doesn't have to do anything. Now, it would be nice if he published the result of his | brilliant | thoughts
RW:   Lucubrations
IC:   and those views, but look at that guy. He gets his first. Look! Look! First of all he goes to school and he hits the jackpot. He can do all the maths. Then he goes to Oxford – which is wrong, obviously. He should have been at Cambridge, but it's close enough. He gets his first. So then he gets the PhD and he doesn't earn too well on that but he's dedicated to his own brilliance, and the myth that he's the salt of the earth, isn't he, by then. So he will work for not much. Then he gets his PhD then he's immediately turned into a fellow, a research fellow, on 13,000 a year. He lives in this beautiful place and he doesn't have to do anything. Look. What what does he actually know and where has he been? That's a rhetorical question. He then produced this concept, he's across the other side of the table, of, you know they've got this idea that if there's uncertainty then another universe develops? You know this idea? You don't know this stuff?
RW:   Yeh. The reason I was shaking my head was you know that people believe this stuff kind of thing
IC:   Yeh. But here here here we're at the high point of that religion, aren't we, because he's saying that. You know, he's just come off high table and he's up, you know, he has the power to let us have madeira, port, and all these other things, and stilton. And you know the special Trinity madeira
RW:   I thought it was port and walnuts..?
IC:   Yeh. That'll be another college, probably. And THEN he presented this stuff about erm you know they have something like Maxwell's demon and you're not sure which way it goes so these different universes develop, and they all coalesce, one lot say you just have a look, and it'll coalesce into one of them.
RW:   In a sense they all coexist at the same time. So time doesn't exist.
IC:   Yeh. So Sue listened to this stuff and of course I said to her afterwards, what about the principle of conservation of energy? You see, because if you start developing all these universes, there's gotta be more and more energy. But of course that's PHYSICS, they're mathematicians, you see.
RW:   Mm
IC:   And when I went, I, I, you know sometimes I'm visiting guru on Wednesday afternoon, and I went to Imperial College and I did my guru thing, and that was in the physics department. And the physicists were really angry that Salaam and the mathematicians were getting the Nobel prizes, and the physicists were not, you see.
RW:   Mm
IC:   So there's that thing going on. But the important PRINCIPLE, I mean it's very interesting to propose that within that universe of discourse, you know with all these multiple universes, the concept 'conservation of energy' does not exist, you see. Now erm the guy came from UMIST who comes down to see me occasionally, and he found that I was based on conservation of energy. Which I thought was general, round the world, you know, that that was a primitive. He said, Oh no. Joule worked, you know the mechanical equivalent of heat
RW:   Yer
IC:   He was in Manchester and he said, you're Manchester and Cambridge, you see. That's where conservation of energy is up front.
RW:   So Oxford is Hegel or something? Berkeley?
IC:   Well, the point is | course they don't care, because they're just bread and circuses! They don't need to agree with each other. All they need is to put in some waffle to get the next funding, and get the next Nobel prize and stuff, don't they.
RW:   Hm
IC:   You see. But, but, do you see the insight into the idea that there's deep fragmentation among the establishment
RW:   I'm surprised he came out explicitly with it
IC:   What? Above the high table?
RW:   Yeh. About Cambridge and Manchester
IC:   That's Joe Marsh. He's at UMIST. He's history of science.
RW:   Oh right
IC:   And he was down to our, he's an archivist, and I've got some letters from Heaviside and stuff like that.
RW:   Joe is J–o–e is it? J–o–e?
IC:   Yes. Why, have you come across him?
RW:   No. I wasn't sure whether he was male or female
IC:   He's a man. But er anyway the point is that when you look at the the the four, I mean we thought, I mean we said years ago that, and I said this in a lecture in Edinburgh, that Paul Davies represents everything that we regard as bad in um science today.
RW:   Without exception he's–?
IC:   Yeh, because, it's all very well saying in modern physics, and the whole thing's betrayed. But if you say, another way of understanding what we're saying is, well, read and listen to Paul Davies. And you can see that that | And then of course long before Hawking was invented I had the genius, you know the scientific genius, it's in my book Computer Worship. You know, I described it.
RW:   Hm
IC:   I didn't go as extreme as Hawking. I mean the idea that, if you can't even feed yourself you're more of a genius than I ever dreamt of. Um and then the voice comes out of a computer. I mean it's straight out of Wizard of Oz, isn't it. It's the same as Wizard of Oz.
RW:   Hm
IC:   And then you pull aside the curtain, and you see this poor little chap. But I mean even THEN they don't rumble it. You know it's just. But I've said I can't compete with that kind of thing. I feed myself, I speak in proper English, my parents weren't murdered in Auschwitz [pron Auscherwitz], and I don't come from eastern Europe. I'm not Jewish.
RW:   You don't have an east European accent, do you?
IC:   It's quite obvious I'm not even average scientist 'cos I don't have everything..
RW:   You don't have a thick accent.. I don't know if this has occurred to you, that it's related to the British class system. That um see you've got a whole lot of working class people but rather than educate them up, which is a dangerous thing, you educate up Jews, who are keen to work
IC:   Ah! Ah!
RW:   Do you see what I mean? Has that occurred to you?
IC:   No. No.
RW:   It might be worth looking into that. You could try this experiment. You sometimes see 1930s films where you have an eminent medical specialist, you know, he looks a bit like Freud and he's got steel glasses. He says Ah! Zis is vot ve call ze syndrome! Something like that. Imagine, you replay it with subtitles, no, voice dubbed on in a Yorkshire accent, Cockney accent; it's completely unacceptable, isn't it? It would look ridiculous if somebody says, By eck, by 'eck, this is a syndrome! Up north!
IC:   But you see Paul Davies
RW:   So you've got to have the voice!
IC:   Take one of these guys, Paul Davies. Twenty years ago I wrote an article, it's in one of my books, there's two. One is called negative time, and the other is called the sign of time. Basically, the argument is that um Minkowsky [pron. –cow–] you know, you've read this stuff, haven't you?
RW:   Yeh, I know what you're talking about
IC:   So anyway, I sent it to Nature. Now Nature said, that's already known. Um Paul Davies this and this, you see. So I wrote to Paul Davies and said could you show me the page in, you know where is it in your books? So he said, read such and such a book. So I read that book, didn't find it. So I wrote to him again, said where is it, he said read such and such a book, you see. So I read that book, and wrote back and said, can't find it. Cos Paul Davies like they do talked about time running backwards, you know, and how the universe might be going. Which is nothing to do with what I'm talking about. In the sign of time. Quite obviously. Yes?
RW:   Um. Um? Yeh
IC:   Cos I can justify that if you want me to.
RW:   Yeh. Carry on.
IC:   It's nothing to do with the sign of time.
RW:   Carry on
IC:   All my point is that when you walk across town to the station, you cover three miles. You LOSE one hour. You don't gain one hour, you see.
RW:   Hm
IC:   So the sign of speed has been written wrongly. Cos speed is positive distance over negative time and they've ended up positive, which doesn't matter until you start trying to get clever over space–time continuum and stuff like that. THEN you have to be rigorous. You know. People like Newton didn't have to be rigorous; it was all obvious. OK? So that's what I was talking about. Anyway so Gribben said it's in my other book so I read the other book so I wrote to him and he said I have nothing to add to what I wrote I those books. This is twenty years ago. So er Gribben was an advisor to Nature
RW:   Sorry. Don't you mean Davies?
IC:   Sorry. Paul Davies. And and um then I said well, what was Gribben? Well, what Gribben was, he finally ends up as a lecturer in South Bank Poly or something like that you know like I was, well I was lower grade, further ed, and then it's turned into a university. Now he took his course notes because he teaches a noddy course of relativity you see, which is pretty noddy, but that's.. now of course he's a lecturer in a university, you see, and he doesn't know, he never WILL know that there actually is a subject underlining his career. See he thinks it's just good fun. See. So then, now he's a university lecturer. He's already got his noddy book published. And now he writes a book published by the University of Surrey or somewhere like that, you see. So then he's on his next step. So now he applies for his job. Now what are they gonna do in the University of Newcastle? Look at the various applicants. He's published books, he's a lecturer at University in Surrey, so he got the Professorship in Newcastle!
RW:   In physics?
IC:   Yeh. So NOW he he knows that he must be expert in the subject, because everyone's treating him as an expert.
RW:   Hm
IC:   But he knows that, huh, he'll never know that he's an impostor, you see! Then finally presumably the chickens came home to roost, it got out of hand with all this, you know, the kind of thing you do is, if a universe is passing you at the speed of light, then it's got no width so maybe you don't notice it, I mean that kind of stuff. You know. You use the Lorentz contraction and that. He plays these games and gets half a page in the Guardian. But after a few years it starts to become ridiculous like they were becoming ridiculous the last few days [NB: This refers to a then–recent Channel 4 TV programme series on a Sunday with a comic actor interviewing Hawking et al] so he scarpered to Australia. Now his COVER will probably be that there was some scandalous unprincipled behaviour over cutting staff – you know, he'll throw up a smokescreen. He'll take off. Now, what about the Australians? They know he's big. He's one of the big men in England
RW:   Hmm
IC:   You see. He's published all the time, books, and in Nature. This is what people like me are up against. Another one, the other one was um Bell. You know, became Professor of Physics at University of Hull. We interviewed him and he didn't know any electromagnetic theory. He said he didn't! He just said. NO! Sorry. He said he wasn't interested in electromagnetic theory. But but he was the one answering me, representing the establishment! You see and he, we saw his career path, which I could take you through. Um what happened was I found he was answering me in in Wireless world, you see. So um and writing some pretty silly stuff which now he refuses to allow me to republish. He was Professor – he had been Reader in Electromagnetism [this is a University position below Professor] at Birmingham University, and then became Professor of Physics at Hull University, you see. So I said right, he's accredited! You know, I need to talk to him, when he shot me down, or appeared to, in Wireless World. He was the only establishment response. So I happened to be in Hull, I was again a mini–guru that time, you know, giving a lecture. And erm I and my wife and Bell went to lunch together. And my wife started interviewing him, you see. And he was boasting about how marvellous it was that he got freebies to Australia with his wife.
RW:   Yes
IC:   And um after a while my wife started talking about electromagnetic theory. And he and he said well, it's not really interesting. And she said, but you were Reader in Electromagnetism in Birmingham University. And he said, well, only because the post came up, and I applied! You see! So
RW:   Did you do anything about it?
IC:   Well, no, I missed, if you're taping me, that was wrong. The point is he wasn't interested in electromagnetic theory. He's a lecturer in Birmingham University, a post comes up, he applies, he gets it, you know, then then he's got to make noises, publish or perish, so presumably he published some stuff. Then his next step is he becomes Professor at Hull. NOW presumably when Catt's giving deviant trouble over em theory the pressure's on the establishment who's gonna answer him. And like a fool he goes and answers
RW:   If anyone
IC:   Yeh, well, they shouldn't. Normally they don't answer at all. But he made the political mistake of answering, which didn't do him any good at all. Um but ah but then it caused me to find out, well not me, I mean I wouldn't have interviewed him, my wife, my then wife, she said well, you were Reader in Electromagnetic theory. Only because the post came up!
RW:   Extraordinary..
IC:   Now, the point is, electromagnetism is a safe area, because it was all signed and sealed a hundred years ago. So if you're just a career man, you know, you don't want anything new
RW:   It's like a Reader in biscuit technology. That's not a very good analogy. Reader in Darwinism.
IC:   Yeh. And you can just copy the old textbooks and they're your textbooks. There's no problem. Nobody'll rock the boat. Everybody knew that, that you can read the textbooks. It's all signed and sealed a hundred years ago, you see. So er erm
RW:   That's why it's called a 'reader' perhaps! Ha ha
IC:   But you see when you look at Paul Davies' first book he's got these pretty pictures of someone standing on a rotating turntable and then – do you know the kind of thing you get? Well you know you have the two explosions at the two ends of a train and stuff like that.
RW:   Yes. Echoes coming back..
IC:   But they're not interested in it, they don't know anything about it. They're not interested in it, you see. They don't need to know anything about it.
RW:   OK. Now. I've got a new idea for you, actually, as a result of my stuff with Hillman. I've got a whole lot of stuff about Hillman here. I finally understood what he's talking about. Basically what he's saying, which you perhaps haven't grasped, is that if you look at cells under the electron microscope everything you see is an artefact. It's all bullshit. There's whole careers built up on the idea of things like the 'endoplasmic reticulum' and 'synapses' in the brain, like on the – did you see Greenfield's, Susan Greenfield, Royal Institution lecture at Christmas. He wrote to her actually, I've got a copy of Hillman's letter to her, I photocopied her reply. It's quite funny, about 'viewing' – she was so pleased to have the opportunity to 'view' his paper. Good word, view a paper. But you know regrettably she couldn't read them, she was off on a trip. Anyway, it occurred to me you see that what he says basically is that you've got these great hefty expensive pieces of equipment, right, electron microscopes, so it's very difficult for anybody else to duplicate these experiments. You can't very well have one in your back room, sort of thing. And that it's at arms length – and you have all these technicians who do all the preparation, cutting up and slicing and vacuum deposition and stuff, so you get these rather gullible people who look at the pictures and they interpret them as ordinary photos. And you look at the thing and think it's this, that, or the other. And they've built up a whole theory around it. And he reckons that all of this is crap. Something like say 60% or so is his figure of so–called fundamental research into cells is just bullshit. They're just wasting their time. NOW I don't know if you find it convincing; I'll show you his stuff – oh, including the [video] tapes, by the way, I've got them
IC:   Do we throw them up on here?
RW:   If you want to see, Yeh. They have this weakness, you know there's this joke, about if you're gonna do a lecture, first of all you tell 'em what you're gonna tell 'em, then you tell 'em, you know. He doesn't do that, you see
IC:   Oh, I don't like that
RW:   You don't like that? Well he's got the opposite idea. He starts, his very first words are something like the cell was discovered by Robert Hooke in 1650 or something, and he goes right through and he doesn't ever – for example, he talks about the endoplasmic reticulum, which should be, from his point of view, in quotes, or should be the 'so–called endoplasmic reticulum', but he treats it perfectly seriously until at the end, you know, you suddenly realise that he's been speaking ironically. So he does have rather, you know, his delivery and script is not very–
IC:   Shall we watch it right now?
RW:   Go on then. Let me just get to the point I was making. If he's right, then all the stuff is sort of bullshit, caused by having very expensive equipment. It did occur to me, you know, watching this stuff both at the Royal Institution the year before, the physics bloke, and also watching this comic actor chap at CERN, [=Conseil Européen pour la Recherche Nucléaire] you know, they've got this bloody great expensive thing, that that might be artefacts too. They've got this huge great
IC:   But that is nonsense! I'm sure CERN is nonsense. I've told you that before
RW:   Do you think they are artefacts? For example–
IC:   It's just total nonsense
RW:   It might be an exact analogy with what Hillman is saying. These things like – you have a computer which is programmed in a way which you're not told about which generates
IC:   AH! Now, I can answer that one.
RW:   Yeh. Can I just make another point? You've got, typically what they want to see is something like a bubble chamber, isn't it, you've got a little explosion and all these things shoot off
IC:   That's what they do all the time. That's all they ever do!
RW:   Yeh. Wait, wait! The point is that's not a – you know that the uncertainty principle. Says that, in the proper sense, you can't actually do that. You can't tell if you detect a thing you can't tell quite where it is
IC:   Yeh
RW:   Now they've got these bloody great detector things which, as far as I can see, they never actually specify how they work. OK. So presumably in fact you've got some great big metal plates and stuff and occasionally you get little tiny charges on them. I mean absurdly insignificant. And what the computer's job is to generate, you know, is to draw a picture. Make a curve of best fit
IC:   Oh, the computer draws it! And they don't photograph them any more!
RW:   That's right, yes. It's all done–
IC:   Ha ha ha!
RW:   They have a whole wall of computers. They showed you this room
IC:   I thought they were still photographing them!
RW:   No. Imagine this. You've got a thing the size of a cathedral
IC:   Oh. Bigger than that
RW:   Yeh. And there's a little tube where the business is done sort of thing. And presumably evacuated and all the rest of it. Lots of magnets and a great big detector – you know those space capsules, a sort of triangular shape. Only much bigger. There's also a whole wall of computers. And the bloke says, gosh. Are they out of date? Is that why they're so big? And they say no, they're state of the art stuff. So the computer draws the picture. It's exactly like an electron microscope in the sense you've got a, you know, the thing is carefully constructed, and how it's done is not revealed. Know what I mean? The people–
IC:   Yeh!
RW:   So it might be, the whole thing might be an artefact. Um, anyway, do you see the point I make? All I'm saying, I'm not talking about the theory, all the waffle of relativity. I'm talking about the actual CERN thing. It might be that all this stuff, all these little things, might be complete bullshit. As with electron microscopy, they might be wasting their time. I'm not saying it IS, because in order to establish that you'd need to know all about, you know, like how the computers are supposed to be programmed, how they actually are programmed, how the detectors work; every one of these is a very difficult thing to get into, isn't it, and also of course you've got no way of – you see with electron microscopy you have published books on technique, you know, because it's fairly established–
IC:   Yeh, I know. You can't find – I still have no information whatsoever about what they get up to in CERN
RW:   Yeh that's right
IC:   And I've been lying in wait for decades now
RW:   It's a shame you didn't watch – they also have these – there was a very hairy bloke who was their chief theoretician
IC:   Wait a minute, maybe I, I'll tell you how much I do know. Go ahead.
RW:   Well I mean I was just saying it's interesting here again. This comic actor with a purple hat on interviews this chap with long hair, with a pullover, you know the kind of thing, he's the guru figure, and he's supposed to be the chief theoretician, and he's at the level of drawing a couple of slits and showing a couple of things interfering and not being sure whether photons are in one place or not. I mean, you know, like 1930s stuff. And this chap did try to ask some sensible questions, didn't he, this comic actor
IC:   I think–
RW:   I don't think he was up to it though
IC:   I think they're playing with fire letting an actor, people like that, come in. I mean they're really going pop now, all of them, Hawking and Penrose and – I mean they're too excited about being on the media. And I said a couple of years ago they were they were gonna self–destruct
RW:   Yeh
IC:   That group of four because um
RW:   Do you take my point about the possibilities of this artefact thing which you haven't grasped yet. I'll try to explain exactly what I've grasped about Hillman. And then you might try to apply it to physics as well.
IC:   This is this classic business of I already knew that, you know, and I'm gonna put you down by saying I knew it
RW:   Well, that's good, I hope so, because if you can put some, supply some
IC:   Right, well Dave Walton, my co–author, he got the top first in Newcastle University, he then became a lecturer in Trinity College Dublin
RW:   In physics?
IC:   Yes. And he said you're not allowed to do theory and practical in, and then in particle physics. They're completely different groups of people. So one group did the experiments, and then deliver the results to the other group, who develop the..
RW:   As photographs or whatever
IC:   The point is, normally you can't get information from these guys. They're all smokescreens
RW:   Yeh
IC:   Buried in smokescreens. But Dave tells me because he is my co–author you see and he was there and he got the top result and he got his PhD and then he lectured in Trinity College Dublin, and he said they're completely different communities the theoreticians and the practical. NOW the other one of course is the physicists in Imperial were moaning about the mathematicians getting all the accolades, remember. But this one – oh no. Actually, I dunno whether the theoretical guys in this particle stuff are mathematicians or physicists.
RW:   Well they may get the accolades but surely the physicist types get the money, don't they? Look at CERN?
IC:   I don't know. No, CERN might be full of mathematicians for all I know. I don't know. But you know the last tunnel they were gonna build was 50 kilometre diameter, in Texas. And it was ten billion dollars. Then it was cancelled. They've gone over the top
RW:   Ten billion. Christ
IC:   Ten billion dollars. Ten thousand million dollars.
RW:   Yeh
IC:   It was fifty kilometres. Because they keep building bigger ones. But er but er what I like is they all do the same thing. I mean I had a, I was long stop for our computer, we used to sell computers that didn't work. You know I worked for a company called urm er–
RW:   It wasn't CTL was it?
IC:   Modula. Yeh. We built these computers that didn't work. And then the customer used to give trouble, you know because he thought they should work. Anyway, I was erm
RW:   They did work occasionally, didn't they? Half an hour, or a day or something?
IC:   Mm. So I was on a political down, so I became long stop for the customers who were about to sue, you see. You know, they'd have the, I've written this up–
RW:   Yeh
IC:   –We'd tell 'em it was their software team, and then we'd tell 'em it was their hardware team, they should have hired our software team, we went through all this. You had to hold out for as year, because the guarantee was a year. And you blame it, and they wanted to blame their own software men, because their software men were uppity, you know. So you'd last for four months. And then you'd send in your, our, software team, and then after a few months – oh and they'd fight their software team, their management would know, it was most unfortunate – but then we'd blame their hardware team who should have had a maintenance contract from us, right. And that would go on for a few months. And our maintenance engineers would go in and change modules, you see. When I went in they'd changed every module so it wasn't the original computer. Twice. Then um they'd send Catt in because this was the way to get rid of Catt, because nobody could do that job, right. So what I used to do was [chuckle] I'd go in to this customer and he'd BLAST at me whoof! and I'd step aside and he'd just fire all this stuff and I'd encourage him to blast my company! Heh heh heh. He'd just fire away into the distance! And he ended up really liking me! [Laughs]
RW:   Hm
IC:   We could get through the one year guarantee period. Then we'd say [waves] bye bye!
RW:   CTL, was that, was that the one, the bloke–
IC:   Iann Barron?
RW:   With two rs
IC:   You see Iann Barron was a Hawking. But he used to walk on | he used to propel himself and feed himself. He did speak in a squeaky voice.
RW:   Wasn't he, er, what's it called, the transputer. I did once see that in operation. I couldn't see what the fuss was about–
IC:   I've had a lot to do with Barron over the years. I could tell you all about about Barron.
RW:   Was the transputer, was it as good as, you know, the Pentium or something?
IC:   Well, you see, Iann Barron worked for um Elliott. There were two computer companies. There was the one I was in, Ferranti in Manchester, which was an offshoot from Camb er Manchester University, and then Elliotts was an offshoot of Cambridge University. And um so Barron commuted between Cambridge University and – actually he wasn't a genius. You know, he probably didn't get a first, which is a problem. But you know you can cover your – in the steeplechase you don't have to clear every fence. I mean he wasn't a proper genius. He was a genius, but he hadn't done completely the proper way
RW:   When you say genius, you're using that in the ironic sense, yeh? Or you're not sure?
IC:   Of course! Of course!
RW:   Perhaps you don't know
IC:   No! Course! And um then so he er was, now I worked in Ferranti and he unbeknown to me was working down in Elliots and we used to compete for the market, our Sirius and their Elliot 803. And we used to say, you know every time a new peripheral appeared, like magnetic tape or something, you used to have to go through this horrendous design procedure to interface it into the computer, and we all said we must have a standard interface. Right?
RW:   You mean like an RS232 or something?
IC:   No no no. A socket where you can plug either a printer or a–
RW:   Yeh. Like an RS232 port
IC:   No. That's only one wire
RW:   All right. A thing with lots of, OK
IC:   More like an I triple E. But of course management wouldn't let you do it, because that was long term, and every time a new peripheral, like a card reader, came along, you cut a hole in the side of the computer, grabbed some wires, and lots of new software and so on. And and Elliott and Ferranti and Manchester University and Cambridge University WOULDN'T go for a standard ?interface. And this really irritated Barron. So he collected up ?Shorter, whom he collided with at the coffee machine, you know in the corridor of Elliott, and three other pawns, what he called my pawns later on, and they set up a company, and they got Maxwell money heh heh heh heh a hundred thousand from Maxwell, and um it started making a machine. And he said he only set up a company in order to make a standard interface. So Modula 1 was set up because of the frustration that you weren't allowed to standardise the interface. See
RW:   Mm
IC:   And er then er but then he became a genius and the whole thing was hyped up to the sky. You know, like Sinclair and all the rest of them
RW:   Tell me about the transputer
IC:   And then he ended up tripping over his own geniosity, the way they all do
RW:   He set up CTL, didn't he?
IC:   Computer Technology Limited
RW:   I remember working on those machines. It was a sort of mini.
IC:   Yeh. Some of my design's in that. Yeh
RW:   Really?
IC:   Yeh. Mm. And and–
RW:   It kept stopping; whenever I tried to compile–
IC:   But I know the level of Barron's competence cos I've worked with him technically and um I know the games he plays. You know. I've had a lot to do with him
RW:   Is CTL still in existence?
IC:   No. No. It soaked up five, ten million pounds of investment and then was sold for 300,000.
RW:   What happened to the transputer?
IC:   Oh, it's still bubbling along. But the other thing – he wanted a standard interface but he also had the idea that computers should be attached to telephones, you see. That was the other thing.
RW:   Like faxes, you mean?
IC:   Faxes? This is decades before faxes.
RW:   OK. But you mean down phone lines?
IC:   Yeh. And of course that you couldn't do, either. Partly because British Telecom [sic] wouldn't let you, but companies wouldn't let you attach computers to the telephone system because that that's too technical. You know, you know, management is technology free and what's a connection between a computer and a telephone? And um so then, but that's why we had a telephone interface on Modula 1. Nobody else had it. But THEN that became the Transputer
RW:   Are you saying –
IC:   The transputer is a computer attached to four telephone wires.
RW:   When you say a telephone, you mean | a vocal thing? Like one of these cup things, you plug it in?
IC:   Doesn't matter. I mean the the point is he said–
RW:   Modem
IC:   –the computer in one town is able to send data to a computer in another town. Which you weren't allowed to do
RW:   That's why it's called 'trans', is it? Because it's a computer that–
IC:   Yup! And he was fixated on that, you see. So that is the transputer. A transputer is a computer with four telephone links to another computer. It's just, he knew that miniaturisation was coming. I mean all these things are at a very shallow level
RW:   Yer. Um.
IC:   But then he'd hype so hard. He was the genius. He'd got the transputer. Nobody knew that it was actually a Modula 1 with four telephone interfaces. I mean I analysed what what was going on in his head, you see. And they were serial, because telephones are serial, which is actually correct. But not four. I mean the number four. I dunno where that came from
RW:   His favourite number or something. He had an idea about squares or something
IC:   But then you see um you weren't allowed to build array processors. You know you weren't allowed to have more than one computer in a machine. More than one. That was–
RW:   That was sort of cheating, was it?
IC:   Well, that was dogma. Because von Neumann was a genius. And if that was a good idea, he would have put two in. In 1940. Right? So you weren't allowed to have two. And if you put two in that created problems for all the programmers and every single processor was surrounded by thousands of programmers. You know, who knew there was only one processor. And they went to time–sharing and I mean I've published that the larger part of software is to try and funnel it all through one processor. So you couldn't just deliver them two, or ten, or a thousand, d'you see. And then what happened was they used the von Neumann machine and then when you, they said why can't the other processor do alternate instructions?
RW:   Literally?
IC:   Yeh. That's right. So then they said what about the branch, or the jump?
RW:   Or the loop–
IC:   So then you you have look ahead, if you jump, branching, then you have more look ahead processors to do both routes and er it was an enormous industry this, you see. Now if you come, but, so that's what 'multiprocessor' meant. You take a standard programme and you tried to cheat and make it a sequential, what you might call a list processor, you see
RW:   Hm
IC:   And how do you try and speed that up? And THAT was another genius. Amdahl. You know, who went around. He was a world genius
RW:   Oh right, yes
IC:   And he er went round the world saying if you put in four or five processors you hardly speed up the machine.
RW:   What was that, he designed his own machines didn't he, the Amdahl, was it
IC:   Well he was in IBM and there were things you couldn't do in IBM and he lost patience and he went out and got Japanese money
RW:   Did he get anywhere with that? Amdahl computers?
IC:   Yeh! It took, he said, he said: you can take 10 or 20% of the IBM business, because the anti–trust laws were strong enough to stop IBM from sabotaging you, and um IBM were on terrific markup, and he got 20% of the big end of the computer market. But the thing is that he er was part of the uniprocessor tradition, which still rules. You're still not allowed to have more than one processor
RW:   Isn't it difficult to connect them together? That's the problem with several processors, or hundreds of them, is that they're all doing – co–ordinating them? At least that's the standard model, the standard picture. Like if you've got your little PC–
IC:   Yeh, but I mean, you try and put four engines in a car, you'll end up in a hell of a mess. I mean
RW:   Yeh, well isn't that part of the–
IC:   But it's nonsense! It's nonsense!
RW:   Why is it nonsense?
IC:   Well, the reason von Neumann – von Neumann, at that time. Even when I started, our one arithmetic unit cost um £15,000. It was the equivalent of quarter of a million pounds today
RW:   You mean a thing that did adding, subtracting, multiplying and dividing?
IC:   Yeh, that's all it did. It cost a quarter of a million pounds. Now you start talking about building a machine with a million of them or a thousand of them it's nonsense. That was where your cost was, and so you funnelled everything through it. And now, when a processor costs you um you know a fraction of a dollar we still worship von Neumann. It's the same thing as Aristotle.
RW:   But I still don't see. How could you get around that? I mean the problem–
IC:   How do you NOT get around it! The way the way you get round it is by not doing the whole array of tasks which a computer is supposed to do. You don't do um traffic management, you don't do weather forecasting, you don't do global warming simulation, you just don't do the full– all you do is is the subset of tasks which digital electronics can do, which is the subset that can be done with just one processor. That's why you end up word processing, because the person sitting in front of it only does one thing at a time
RW:   So are you saying the computer should be completely redesigned to do things like–
IC:   I redesigned it 300 years ago! I was talking about this 25 years ago!
RW:   So how does your system work? Suppose you wanted to simulate–
IC:   You can't build it! It's a multi billion dollar industry and you can't get started!
RW:   No, I'm just asking you theoretically how you would set about for example–
IC:   I thought you knew all this. This is my computer architecture stuff
RW:   Well, I've read it. I thought as far as I could gather all it was was just elaborated chips and nothing particularly – I've missed the point?
IC:   No. Look. Look. In the in the Australian copper mines [Note: I recall my heart sinking at this point at the prospect of yet another interminable diversion from the point] you dig out the copper and then you have to build supports. And they put in what they call slurry. OK?
RW:   Hm
IC:   You put in this stuff and you add the minimum amount of cement to er stabilise it. There are all these stresses in this, and this is thirty foot wide and three hundred foot high this pillar, or something. I only found out because I was in the Youth Hostel in Skye and the other guy couldn't get away. You know you can't find out from professionals what they're doing unless they're trapped with you. Yes?
RW:   Mm
IC:   Because they hide behind jargon, and insults, and various mechanisms. Anyway, he he'd been there, you see. And to work out, and it's gradually flowing this pillar, because you put in the minimum cement, mortar, cement
RW:   Concrete or whatever it is
IC:   Cement. You know
RW:   Dust. I know what you mean. Calcium oxide
IC:   Portland.
RW:   Whatever it is.
IC:   Yes. So then they have to do a computation on this pillar and they run it through the Cray machine for you know for a month. Just a simple calculation of how the stresses, and then of course you've got a problem with weather forecasting. If you're forecasting 25 hours ahead, 24 hours ahead, how can you do it, it takes you 25 hours to do it. That was the problem for decades so you couldn't have weather forecasting.
RW:   Yeh, I know all that. How do you get around it?
IC:   Cos it took too long
RW:   The problem is if you've got lots and lots of data, can you handle it all simultaneously..
IC:   Well. ALL because ALL of the problems I've ever thought were very important involve and I've published this many many times a multi–dimensional array of | values. This three dimensional array and at every point. The obvious one is weather. Five hundred feet above Swindon there's a certain temperature, pressure and humidity and so on
RW:   So by multidimensional you mean several readings? Is that what you mean? Umpteen readings? So the wind direction could be, you need three vectors for wind direction, you need temperature, moisture, there are about ten readings–
IC:   Wait a minute. If you're trying to do global warming simulations or weather forecasting or if you're trying to find oil, you know, you fire explosives, and they go down under the ground and reflections back?
RW:   Yeh
IC:   Or if you're trying to do ANYTHING what you end up, virtually everything except working out somebody's payroll or something, or word processing a letter, which is what we've ended up doing–
RW:   You end up with lots of little–
IC:   What you've ALWAYS got is, and it's usually three dimensional, and we know how many points there are, it's a thousand by a thousand by a thousand, and at every point there's um various things, you know.
RW:   Yeh, maybe ten to the power–
IC:   And what then happens, in the next microsecond or second or month or hour, that affects the adjacent points
RW:   Yeh
IC:   And there are eight or whatever adjacent points and then that affects the next one and it flows out. Now while that's happening, the adjacent one's are happening, you see. Now what they call that is computational fluid dynamics. CFD. Or finite element analysis. Now if you, I've been to the Cray. I've talked to Cray men. Cray machines with a single processor in the middle are all doing this task. With one processor. And what you do is if you've got a thousand by a thousand by a thousand you lose a factor of a thousand million in speed. And that proves you got to have as faster processor!
RW:   But how do you–
IC:   To make the processor faster and faster and THEN they have look ahead and try to do two calculations at once!
RW:   But how do you get round that? The question is how do you co–ordinate the things together?
IC:   It's easy!
RW:   Is it?
IC:   Yeh! Because because you see you see
RW:   How do you store your values in your system? Do you have a big sort of RAM?
IC:   Every node, I've designed the machine and you can have it for a quarter of a million pounds and the project is a forty million pound project and it's a multi billion dollar market! Now. What you do is I have a two–dimensional array which is self–organising, self repair and at every node there's a processor and there's a RAM and there's a ROM.
RW:   Mm
IC:   And and I can deliver the program from outside globally at the rate of a thousand megabits. A hundred megabits, sorry. OR each processor locally can take off and obey its own um instructions which are in its own RAM.
RW:   So if you want to simulate weather forecasting, each little processor has the task of looking at all its neighbours
IC:   No. Delivering to its neighbour. Its neighbour says hey it's hotter there so I'm gonna up my temperature.
RW:   OK. So if one of them receives from its neighbour–
IC:   But they all do it in parallel
RW:   Yeh. It must also give to its neighbour, mustn't it
IC:   Yeh. Uh.
RW:   And it's programmed–
IC:   It's very simple
RW:   And it's programmed specifically just for that, like in the case of temperature–
IC:   No, no. You know. You feed in the program before you start. And what you end up having is, now. I emulate the third dimension. The conventional machine is zero dimensional, you know it's got one processor, and every time you pretend you've got a further dimension you lose a factor 1,000 in speed. Now mine, theirs, is one dimensional. Mine is two dimensional and it loses a factor in speed by pretending it has the third dimension
RW:   It's still a thousand times as fast?
IC:   Well, it's a million times faster than the Cray machine
RW:   Why did you pick on a thousand as a factor?
IC:   Because that–
RW:   If you want to simulate the entire global atmosphere a thousand points is not enough
IC:   In all the, now if you want to send TV pictures from Britain to the USA you've got to compress on the fly. Because satellite costs five million. And the upfront compressor which has to do it real time is a quarter of a million, my machine, and then the decompressor the other one is a quarter of a million, right? Now TV is a little less than a thousand by a thousand, right? And if, generally they're a thousand by a thousand..
[END OF SIDE A OF TAPE]
.. I published it and everything. The point is, the point is, I was talking to someone called Jill. I was at a birthday party yesterday. And I said, this is not my problem. You know. Um. In order to survive, a country can steal technology from abroad, like we could steal the German rocket technology or the Americans could. But it cannot allow the creation of wealth based on its own high tech innovation, because of the political implications. You see. If you can STEAL it from the inventor, you know, maybe you can show that he's a fool and he gets angry, and you fence him off, you see
RW:   Yeh
IC:   But OUR society can't even steal it from local people. It can steal it from foreigners, you see, and thereby you get rid of the political implications. You know. The RUSSIANS could STEAL um the American atomic bomb capability, but they couldn't afford to allow their own scientists to develop atomic bombs, because their own scientists would then become too powerful
RW:   So what did Jill say to you?
IC:   Well, I said, look, it's not my problem. You know, I've designed the machine. I've published it. You know. It's a multi–billion dollar industry. But it's not gonna happen and it's not my problem. And she said what about the sort of satisfaction of having something of yours built? I said well, I've had that once. My previous one came to market in the USA. It's not my problem
RW:   What was that one?
IC:   Well, that was wafer–scale integration and memory
RW:   You mean the thing, the Anamartic thing?
IC:   Yeh. But you see, all the time in Anamartic I really pushed to have one logic gate added to every RAM, because then you could have array processing. But you weren't allowed to have any processing in the array of memories. Because then management would lose control.
RW:   Is that the reason?
IC:   Cos management understands RAMs. That's difficult enough for them. EVEN in a company set up to exploit Ivor Catt's inventions, they can't go beyond the first step, which is to have cheaper RAMs.
RW:   What would you have, if you've got logic gates added to RAM, what's the effect of that?
IC:   Then you can have – look. What happened originally [again my inward sighing] was right at the start I was in Motorola in um 1960 | 4, and when you laid out a chip with the tape, if you made one mistake it cost you fifty thousand or something, I'm exaggerating. So ?Hazelet came in, he was the bright young thing, and said let's have a computer lay out the chip. You see. And top management in Motorola were sold on this and ?Raislin and ?Hazelet were in on this. What they did was they said we'll get to the top of the company by being in different departments but by supporting each other.
RW:   Mm
IC:   You know, and we'll knock everybody else out, because they won't know what's hit them. If they're hit from two different places. So there was ?Raislin and ?Hazelet. This is unusual. Usually there isn't conspiracy in industry
RW:   They were different departments – one was in chip design, and the other was in–
IC:   Yeh, but they knew they could reinforce each other. I BELIEVE that was the plan, you know. I mean they indicated to me that was it. Well anyway, ?Hazelet er er got to top management and he said we can't afford the errors as the chips get more complicated. At that time, the most sophisticated chip had 16 bits of memory in it
RW:   Yeh, yeh, I've seen the kind of thing
IC:   Yeh. That's the Honeywell Transitron memory. But the errors, and so management said oh yeh, the modern thing's computers. And I got to his group and I said, you got to have content addressable memory which is you got to have a modicum of search or processing ability distributed throughout your memory. A year later – oh well, I said go and talk to ?Sealbach, who's deputy head of integrated circuit R & D, who's pushing for content addressable memory. That is, you can address by content, rather than by physical location, which is the RAM.
RW:   I don't understand what addressing by content means
IC:   Instead of saying I want, which is still the only thing you're allowed to do–
RW:   Number 25,000 or something
IC:   Yeh. You can say, um are there are any words in that memory whose um where this particular field is 1010110, you see. And it comes back and says yes or no. It's trivially..
RW:   Hang on a sec. Suppose you've got a thousand of these..
IC:   Yeh. It does it in parallel
RW:   Suppose it's got a thousand of these or something. You wouldn't get a thousand addresses back, would you?
IC:   I SAID, you didn't hear me. Are there any any words in memory where that field is this number?
RW:   So it's looking for–
IC:   Any. Any words in memory where that field is this number
RW:   So you're looking for just one
IC:   And you get the answer yes or no
RW:   You're looking for just one of them? At least one of them?
IC:   No. You're not listening to me!
RW:   Yes I am! You said–
IC:   Well once it hears there's one it knows it's got a problem, because there may be ten thousand, so it has to do something else
RW:   I still don't understand why you call it content addressable. You seem to be saying you've got a whole lot of RAM, OK, straightforward, and it's looking for a particular configuration of bits. If it finds them–
IC:   That's all we could do at the time
RW:   If it finds just one of them, it says there is one
IC:   Well if it finds one or more it says there is one
RW:   Is it? What's it say in addition to that, or is that all it says?
IC:   It says that. Because that's true and easy to engineer.
RW:   So what's the point of that? Why should it want to know that? That's just a useful thing, you think that might be useful? In other words you can address it..
IC:   Look. Look. Look. Von Neumann [again; sinking feeling] It was so expensive to make an arithmetic unit, it was so expensive to make memory, that they were separated. This is in the 1940s.
RW:   Sure.
IC:   Now. And core memory, you've still got to keep them separate. Because magnetic core memory, you can't search a load of doughnuts. Because to read a row of doughnuts, you've got to him em with two amps, and they go crazy
RW:   Yes, yes, OK
IC:   But RAM, semiconductor memory came in, and we made semiconductor memory or were going to but we couldn't compete with core memory
RW:   OK
IC:   So I said, don't compete head on with core memory. Give an added capability.
RW:   OK. What is–
IC:   But you could not give an added capability, because in the computer industry, when you have a new technology, you're only allowed to exactly ape the previous technology and include all the inadequacies of the previous technology.
RW:   Yes, it's not surprising that happens, is it. But just explain to me why content addressable memory–
IC:   Because that was the first step towards array processing. Right. Now the next thing, so the first thing I did, if you read my 1972 book, is saying we've got to have content addressable memory
RW:   But all that means, content addressable memory, is that you can say, you interrogate a whole batch of memory, a whole lot of it, for one particular thing
IC:   In a microsecond
RW:   –and know it's there or not
IC:   In a microsecond, instead of in in ten milliseconds
RW:   Thank you. That's all it does?
IC:   That's the first step
RW:   Yeh. OK
IC:   NOW. You see. You engineer that. BUT. What I realised was once you're into RAM – we were into er CCD at the time. And that's bucket brigade. And all our memory was rotating, going from capacitor to capacitor to capacitor–
RW:   God, like these things, with mercury, the thing goes along it
IC:   Yeh. I first published pushing content addressable memory, also called associative memory, in '69. In an article in Nature, in New Scientist. But in '74 um bucket brigade, you know which is in cameras now, was running neck and neck with semiconductor, with RAM, right?
RW:   Mm
IC:   Now, properly, the serial thing should have held its own, but it couldn't, because of the commitment to magnetic core memory [small magnetic rings threaded on rectangles made of wire] er er ambience, you see. The idea that it should sit still was too strong. Cos that's what the previous technology did
RW:   Yep
IC:   Right. Now, once you've got the stuff rotating, THEN you don't search a specific field, you you go for a particular field at a particular time
RW:   You just wait for it
IC:   Right. Now. If you then add, the next thing you add is, IF this field has a certain value, then add 27 to this other field, they all come to you, down, all in parallel. Right? So that's the next step. Now, what is the cost? If you have a 32 bit, they were bigger then, a 32 bit word, it costs you so much in semiconductor memory. You find that a processor is one exclusive–or and one bistable. A serial processor. It costs you 3% of what's already there. Right? So you could have um the thing where you go in with a, it's called a mask. And anywhere with a tag of a certain value, you could modify another field in that word all in parallel
RW:   Yeh I see what you mean
IC:   Now now. For virtually no cost. On this rotating memory. It's more difficult with RAM right but you still should do it with RAM. But BECAUSE the concept of manipulating memory in situ was illegal, the technology moved away from it.
RW:   Um. See what you mean
IC:   Right? But it was standing there waiting for us. And the feeling against bucket brigade memory was so strong that when I worked in er say Nimrod, you know, and we were getting returns back from what were called the targets and we had to record them all this serial stuff was better because it's coming in serially right and they were equal price with RAM, you had to buy RAM. Although it's more awkward, because everybody knew that the proper thing to have was RAM. Because that had the tradition, that was core memory. You see. And everybody panicked at them moving, not realisng that if it's moving, then the data just presents itself to you. In general, you locate stuff in memory by physical location, and by time.
RW:   Mm
IC:   But the time dimension of of of accessing memory has gone, because it all pretends to stand still. That's a whole dimension to memory accessing which has been thrown out. The whole thing's very very doctrinaire
RW:   Yes, I understand what you mean.
IC:   Right. Now if you go into, now in my case I came out years later, and the bucket brigade, the rotating, has been junked. So then I've got RAMs. And I said, right, we'll make these, and I've got a cheaper way of making RAMs. It came to market, right. But I said, if you just add one gate to each RAM, which costs nothing then you can start to process the stuff in parallel and then you start upping your speed of your machine by a factor of 1,000 a factor of 10,000, but you couldn't do it. You weren't allowed to add this one gate. We were adding, because the configuration logic, you know to connect the good ones and bypass the bad, was in; that was my idea, you know. That was 200 gates. But that was for avoiding the bad ones. I wasn't allowed to add one more gate in order to increase the speed of the machine by a factor of 100, because then the management would lose control. Catt would be in control. You see?
RW:   You think that's the reason? It's not just conservatism?
IC:   It doesn't, it doesn't matter. I mean, nobody thought through this. They knew that we weren't gonna do that. You're not, if you're making, if you're using a technology, a new approach to a technology, in the computer industry, you mimic a thing that already exists
RW:   Yeh, isn't that–
IC:   And you don't offer one way er unidirectional compatibility. You're not allowed to say, I can do exactly that, but if you're interested, I can also do this
RW:   Hm
IC:   You're not allowed to have that further facility even if you keep it secret. That's the political reality in the computer industry.
RW:   Couldn't you have smuggled it in some other way, by pretending your gates are specially designed to– [I was thinking of the stratagem by which e.g. an early desktop computer was called a 'calculator with visual display' to smuggle it past the official computer people; or applying for two separate items – processor, display – each of which is below a figure at which interest is taken. He could have caled it an 'external memory check line' perhaps]
IC:   Yeh, but it got so bad that they wouldn't even let me on site. I mean, one time, take Motorola. At that time, the semi–conductor people who had nothing to do with the computer industry, said we must, we can now make integrated circuits. You know, you have the two transistors on a chip, you know, but we can get more–
RW:   Mm
IC:   –so we'll make logic, you know. Well obviously that's the next step. So they read the logic books. And in the logic books was the AND, the OR, and the inverter, you see, and the NAND and the NOR. And so I went after them, I said, no, there's the exclusive OR, you see. But they weren't gonna build the exclusive OR, because that wasn't in the old technology. And they refused to. Then in that case, Ron ?Treadway, cos ?Nerude said that if you make anything but ECL I'll resign, the head of R & D
RW:   What's ECL?
IC:   Emitter coupled logic. We didn't make TTL, because ?Nerude was married to ECL. So Motorola only made ECL. And he said, if you make anything else, I'll resign. So the years went by and we lost the market, so in the end they said we must make TTL and so ?Nerude resigned and so TI was the biggest TTL manufacturer so we stole ?Treadway and he came along, see, from TI to Motorola. So I got to him, and I said, look, there's the exclusive OR, and I convinced that one man. It's a basic logic unit.
RW:   It's very important isn't it.
IC:   And er convinced him and he slipped it in. But normally, if it hadn't been for that – I published the neglected exclusive OR, published an article, and I got hold of the man who was introducing TTL into Motorola, convinced him, so that's how the exclusive OR came in as a basic logic function. But in the case of my own company, developing my own – and that's why it's a 286! You know, the NAND is a 74 hundred, and the NOR is a 74 ten and things like that, but the exclusive OR is a great long number, cos that came years later, and had to be hidden away as a peculiar job. You see –
RW:   Tell me again why–
IC:   It's very very conservative
RW:   I don't find that extraordinary. Why is it that, that er, that if you add extra gates to RAM that that allows array processing? I still don't understand
IC:   | Because | what they're doing now is so slow, so incredibly slow, you know –
RW:   If you want to find out what's in one RAM area you have to, send it, you set the address, don't you, you set 16 lines say, the pattern of them –
IC:   You get the word out, then you analyse it, then you send it – that's called the von Neumann bottleneck
RW:   Right, and that takes two or three–
IC:   Fifteen years after I published on it this guy got a prize in the IEE for coining the phrase the 'von Neumann bottleneck'. That is, you have to have one line between processing, between memory and processing, and everything has to go to and fro–
RW:   So how does having your gate on each RAM–
IC:   Well, lemme tell you the the, if you forget von Neumann and history,
RW:   OK I'm quite prepared to do that
IC:   If you go into the technology now
RW:   Yeh
IC:   And say, what should I be doing? The first thing is you've got the physical reality. You know, like the weather, like temperatures and stuff, which is the real world. And then you've got the real world of the technology, which is a two–dimensional array of bits and pieces.
RW:   Hm
IC:   Transistors and resistors–
RW:   When you say two dimensional, you mean?
IC:   We've NEVER done three dimensional circuitry. It is two dimensional.
RW:   OK. Hm
IC:   And so we say well, that's the difference, between that and that. You know, this is 2–d and that's, probably, 3–d. You know, cars driving around the town is only 2–d, probably. Or driving around the country. But generally that's 3–d. So then you THEN you say map that on that and it maps directly. But not if you've got a load of programmers in between, you see
RW:   Hm
IC:   What, and what I called it, and this is 1972, digital analogue of reality. The curiosity that you digitise everything, for rather peculiar reasons, but it's a straight analogue, right?
RW:   Umm
IC:   Now, there are two kinds of reality. There's the reality where, like roads, where you get a peculiar array of traffic junctions. Or there's reality like weather or those pillars, where it's uniformly distributed, you know, in a three dimensional array. Now, um the one I'm best at is the one like weather and so on, you know, or air traffic control, where I take the whole of Europe, you know, and I map my integrated circuitry onto Europe. You see. And all I have is I have a processor in RAM for every square mile of Europe, right, and I hand the plane on from one to the other. You know, because you've got a plane, which is in one particular square mile
RW:   What's a square mile? Do you mean a sort of prism extending indefinitely upwards?
IC:   No! No, that's the point. No no. It's only two dimensional. But what I have is at every node I have a processor and I have RAM in pages, that I then commit the first page to ground level, the next page to 500 feet, the next – and what you do is you time share, you know, this instant you're dealing with that level, then you move up, you're dealing with 500 feet, then you move up, then you move up, so you lose a factor of 20 in speed, not a factor of a thousand, cos normally, if you have a three dimensional array, one dimension is less than the full thousand, you see
RW:   So you go up, like, sequentially–
IC:   So the whole thing is ordered. The other thing you do is, somewhere like London, you have to have a tighter, you have to have more, so at one level of pages you use the whole array for an area of only thirty miles square, right, and then you have a processor for every you know thirtieth of a mile square, right, and then what you do there's an ordered procedure where you move in and out. But they're all moving. And this is very easy to program. Right?
RW:   I still don't understand why RAMs having logic gates RAM attached to them
IC:   Yeh, but nowadays, you see, if you take the RAM, the RAM let's say the megabit or the four megabit, if you take – oh. What you then say is, how much area should be for memory, compared with processing? And then I say, if I went 50 50 I won't be far wrong. I can only be wrong by a factor of two. You know! The best ratio of processing to memory at each node is between 100 to 0 and 0 to 100, isn't it. Right?
RW:   Um
IC:   So if I go 50 50, I'm home and dry. So I do that. And I say a RAM, alongside a RAM, you can have, with the same area, um you can have your 64 processors, you see. So I divide the RAM up into 64 segments and each segment has a processor. And up to recently when you read a 1 megabit RAM you actually took a whole column off, you read a whole column, then you chose one. And so I took that away, and I had a row of processors and 8 bits would go into the first processor, and eight into the second, it worked beautifully, you see. And I had the man who'd already, you see, already in Anamartic, manufacturers of the current state of the art RAM, you know, the best, were willing to change the step and repeat, to leave a space where we could put the Catt configuration logic for fault–tolerance, and self–organising self–repair. And they were also willing to have them slightly widely spaced so we could put in our processors. You see.
RW:   Um
IC:   So it was all worked out, right. The trouble is, we had to get in at the 1 megabit or the 4 megabit stage, where you have the two–d array of memory lots, and when you read something you took a whole column off. You see. The trouble is, when they went to 16 megabit, they went another way, and they weren't taking off – you know – up to 1 megabit, when you read one bit, say you had n bits in memory, you actually took off square root of n bits, so we could process them all, you see. So the 1 megabit and 4 megabit were beautifully married to what I was talking about. But then, nothing happened for a decade, you know, except, right, and the point is, what I have to try to do is to find out what the latest thing is, you know what in fact IS the 64 megabit, you see. Now if the 64 megabit is not really much cheaper than the 4 megabit, then it doesn't matter. We can go back to old technology, you see. But I have to have that information. And this is where I just get bored. If I know, if I conclude that nothing's gonna happen anyway, you know that if you come back in 50 years' time, you'll find er that global warming is going to be simulated on one processor, then I say, so be it. It's not my problem.
RW:   Um
IC:   You know. You can have a whole world with temperature varying at every spot and at every height, and they're gonna feed it through one processor! Well! Fine, fine. I'll go away. And I er to say, how do you do it otherwise, do you see that's absolutely bizarre! How do you do it! Well, with difficulty! And so they say we'll make our processor faster and faster this this er you gotta build an array of processors which maps the shape of reality. There there's nothing – oh, and you know another thing that happened. There's a rule that says if you spend so much on a computer, you then spend a hundred times as much on software. So I said, software's so expensive that I'm gonna change the architecture of the computer to reduce the software. But I'm gonna spend money. [Giggle] See so they say yeh, well, it's gonna be more software. You see, if you buy a million processors from me, you pay more than if you buy one processor from Cray. If you pay ten times as much, it doesn't reduce the software, you have to pay ten times as much. Cos of the rule! That whatever the hardware price, the software costs a hundred times as much! But I say, but I've DESIGNED the hardware BECAUSE of the cost of software! Ah, yeh, but you don't realise, everybody finds in practice that software costs are always higher than they expected! [Strikes fist in palm loudly] See! You can't! Ha ha ha! So the concept of modifying the hardware in order to reduce the admitted enormous software costs is an illegal concept! Ha ha ha! If a man from the moon came in, and he said, you got a machine that I paid £500 for and I've been spending three years programming it, you see, well wouldn't you think of making the machine easier to program? Oh no, you don't understand these things. We've been there for thirty years. That's enough of that stuff, isn't it? How much of a, how much, I shouldn't get enthusiastic any more, should I?
RW:   No, you shouldn't. That's why I tape record it.
IC:   OH. Let me tell you one thing. The archetypal memory, what it should've been, was. It would have been nice if it was moving, but there are problems with that which I could have described to you, which I missed. BUT what you have is that at every location in memory you have a very long word – they're long words by the way – and what you do is you go in and you check with one field and on the basis of the content of that one you modify another field
RW:   Um
IC:   That's what you do. That's the the–
RW:   It's content addressable. Right you're saying if you find the first bit–
IC:   Well, no, that's the next stage from, that's array processing, and you do that globally. But there's another thing, where each node talks to its adjacent neighbours, all in parallel.
RW:   I see what you mean. So you're saying that if the first part of it says sort of 1000 then the rest gets changed–
IC:   Add 17 to the third part. That's the sort of thing, that's what it is, you know. Now the point is I know that's what they're doing in weather forecasting, but they won't tell me, they'll only insult me, you see. But but what happens is that I was with Sinclair, I was with Future Products
RW:   I don't know what Future Products is
IC:   No, because we're in Britain, and British companies don't have any future products. Right? Tut. Um
[RW Stands up, looks at framed monochrome picture of sports team in Cambridge about 1950s, with Catt; cricket? It has a computer–printed bit of paper in Gothic script tucked into the frame]
What happened was there was a battle in one of the choirs where I sing, you see, about would we go along with the conductor, you see. Then what's interesting is the conductor's father is there, and I'm there, you see. So I made out this a joke, I xeroxed it and gave it to–
RW:   Oh, there's two Davises. Trinity College..
IC:   I need to go to the loo. Are you going to switch off the recorder? Switch off the recorder | Certain historical events have got in the way of the natural evolution of the thing. And I'm telling you what is the relatively natural way that it develops. [Sits down with pad of blank paper; starts to sketch] Er and I'll go back, now, what happens is Birmingham, let's say 1961, right, and then we have Sirius, £25,000 and we have the analogue machine, right. And you gotta listen to this,
RW:   OK I'm listening
IC:   and not expect quick results. It will be quick provided you don't get impatient. So. What now happened is if you have a motor car, say Ford have a motor car, and the want to know about the way the the the er dampers and the springs work, right?
RW:   Uhum. Suspension
IC:   Suspension. They build a model, a mechanical model. And we're just coming out of that. And then the concept of building an electrical model arose, so you could have a system where, you know a spring is represented by an inductor and so on
RW:   Yeh
IC:   You know any analogy?
RW:   Not the detail
IC:   The basic one is a spring, a mass, and a dashpot, you see. A mass, a spring – this is standard school stuff now – and a dashpot
RW:   [Looking at sketch] A dashpot is a pressurised cylinder?
IC:   It's a thing that you lose – no it's not, actually. It it you burn up power as it moves in and out. It's just a friction thing.
RW:   OK
IC:   It's not what you think it is. There's a hole here. Now you take the mass and you let go and it moves around and the spring compresses and decompresses and this burns up. And THAT you can find is maybe the inductor, the capacitor, and the resistor.
RW:   Hm
IC:   If you want to know what a thing like this will happen, instead of pulling the mass out and letting it go, you put a voltage at a certain point on a circuit like this.
RW:   Right
IC:   You know, you'll suddenly switch a voltage, and you'll watch what happens. And the pattern of voltage and current will be the same as the mechanical movement here.
RW:   Should be. That's what the model says, that's what the maths says. Yeh.
IC:   So that was one tradition
RW:   Are you saying that's identical, or just similar? I mean you must appreciate if you've got a spring–
IC:   Yeh, that's a perfect spring and a perfect mass–
RW:   Well, exactly. A perfect spring's not something you get in real life, is it!
IC:   Yeh! That's what you start with! If you're designing a suspension bridge, you first start with perfect components then you build in the imperfections. But you have to understand the perfect one first. You know, you know. And
RW:   I'm sorry. I'm not sure whether, how serious you're being. Obviously
IC:   Yeh, but the imperfections, if that has some losses, you then put a resistor across the inductor. The point is you then build in the imperfections. Anyway, that was one tradition, which was going strong. Now another tradition was um that somebody realised that um there's an op–amp, which is drawn like that, [triangle with apex to right] and if you have a signal, an electrical signal | the problem when you're dealing with things like electricity is the signal gets smaller and smaller.
RW:   Um
IC:   You got to amplify it. Now if you have 100K there and 1K there, that amplifies by a factor of 100, so anything you're doing, you can amplify. Analogue. This is all analogue, you see. But then it was realised that if you do that, that's an integrator. What happens in real life is, if you accelerate for a certain time, you end up having travelled at a certain velocity, which is the integral of your acceleration. So you put in a voltage here which represents the acceleration, and out comes a voltage representing your velocity, and then you put it through another one which represents – so that was ANOTHER tradition.
RW:   Using the amp
IC:   Using the op–amp with feedback. Right? Which isn't the same as that. But there were two – there was ordinary modelling, where you build a ship a tenth the size and see what happens, there was the electrical equivalent of
that, but there ALSO was another type of model, right? Now THAT was the analogue computer. And it was selling for £25,000
RW:   Um
IC:   Now. The reason that was in trouble was you couldn't go above 1% accuracy. Or 0.1%. And we wanted more. Because as you go through various stages the inaccuracies build up.
RW:   Yeh
IC:   And it turned out that in the digital, you could simulate the analogue. Because the digital computer's so versatile, you could pretend it was an analogue computer
RW:   Um
IC:   And it turned out that OUR computer could simulate everything that could do, AND have better accuracy, and better speed. So the analogue computer was dead.
RW:   Um
IC:   But what wasn't realised was we were only SIMULATING the analogue computer. It was still there inside that machine.
RW:   Um
IC:   And to the extent that you transform, you see the data in the memory WAS all this stuff, or all that stuff.
RW:   Um
IC:   Because it had to be. Cos that's what you were dealing with
RW:   You were dealing with maths
IC:   Yeh. So, but what wasn't understood was it was thought that within computers, every time you transform the data, you're becoming more efficient, more efficient, you know like high level languages and stuff like that
RW:   Hm
IC:   See buried inside your computer down the various stages of transforming, was that motor–car. And I said, what you've gotta do is climb back out of those transforms, in order to increase the efficiency. And I said right at the back end of the computer, is not lots of little doughnuts any more, is actually a piece of material which has a lot of the characteristics of the thing you first started with. In the case of weather and stuff – not in the case of the motor car
RW:   Yeh. I still don't see the point you're getting at. I thought, the way I summarised it was a whole network with a whole lot of values stored, connected up by wires which have some way of, which actually have to be engineered so you make the components–
IC:   I have a thing like that, you see. Now what exists at a point in – Oh. The next thing is, because of this inaccuracy, we went digital. And whenever you do anything now, in the real world, you gotta first digitise. Do it all digitally, and then transform back
RW:   Well, that's usually true, It's not true for photography, for instance. Photographs aren't
IC:   Not talking about photography! I'm talking about MY world of computation
RW:   OK you're thinking anything digital and digitised–
IC:   No. Cos weather is not digital. But if you wanna get anywhere in forecasting the weather, you digitise your input, do all your work, and then go analogue again. You know. Right?
RW:   Yeh
IC:   That's because of this inaccuracy. And the reason for this inaccuracy–
RW:   I don't think – it's because with digital computers you can just do a whole lot more. You don't have to build up millions and millions of components like this, with little amplifiers and stuff
IC:   I can explain you WHY we don't do analogue, right, why we go digital. And I don't wanna go down that route, right, cos it's not
RW:   I don't think it's to do with accuracy so much as–
IC:   Now. Can I go on? I wanna finish this?
RW:   Why are you taking me up this blind alley of accuracy? I think the reason that they use digital–
IC:   Right, you've ignored that, you've said to hell with that, we didn't go digital because of accuracy. That's right. I don't wanna talk about that one. Now. What I then said was, what is the reality five hundred feet above Swindon or at one pixel of a TV frame, the reality is there are certain features to it. You know like temperature
RW:   The ones you're choosing
IC:   Yeh. Now take that weather, or take that pixel, image enhancement, that relates to the previous value at that position on the screen in the previous frame. And it also relates to adjacent values.
RW:   Yeh, yeh
IC:   You know, if you go down and you got a blue one in the middle of lots of yellow you say that's a bit odd
RW:   Yeh, yeh, right, right
IC:   and you do a bit of computation. Locally, you never, if there's an odd bit of blue in one pixel does it have any relevance to what's going on up there? And that is true, of ALL your problems. You know, a traffic jam in in in Harrow has absolutely nothing to do with the traffic situation–
RW:   Yeh, it's true up to a point.
IC:   –on the South Bank
RW:   I think you can always er for example if you've got a thunderstorm, you're trying to model that–
IC:   It's still true
RW:   Well, lightning is to all intents and purposes instantaneous–
IC:   And the problem is, the way computers have gone, when they get to the array processor, they have a thing called hypercube, where the access of every processor to every other one is the same, there's no um um improvement of access between two adjacent processors, and that's what they're aiming at, that any processor can act, acts with any other processor, exactly the opposite of what you want, and exactly opposite of what the real world is like
RW:   Hm
IC:   Because the effect you know in Wimbledon, of a car pile–up in in um in in Harrow is gonna take an awful long time to come through
RW:   I don't think it's always true that um sometimes–
IC:   No it's always true. You cite a case where two cars crash in Harrow and ten seconds later there's a problem in–
RW:   Well, you're deliberately selecting an example where it doesn't apply
IC:   No I'm not
RW:   Well, all right, take weather, and take a thunderstorm
IC:   Yeh. I'll tell you why it's true
RW:   If you model lightning–
IC:   Shall I tell you WHY it's true? Because we all adhere to the idea that there's no instantaneous action at a distance
RW:   Well, lightning, you now, if you're computing is for practical purposes instantaneous, isn't it? Or to take another example. Suppose you–
IC:   That's just silly. Can I leave that one?
RW:   It's not silly. If you're modelling the weather thunderstorms are likely to happen. And you can have–
IC:   Yeh so you have a lighting flash above Swindon and that has an instantaneous effect on the weather in Perth in Australia? Yeh. You're just talking nonsense
RW:   You're saying, you're talking of long distance. But over a short interval–
IC:   That's what I said. That's what I said. You must have the machine that relates to real problems
RW:   Yeh
IC:   Has better access to adjacent processors than to ones at the other end of the machine
RW:   You didn't say better. You said er, um, exclusive. It's only to connect up with neighbours. But in the example of a thunderstorm–
IC:   No. No. In my procedure – I'll tell you what my procedure is. I have my array. And you can, you can go in, globally. It's easier to talk to it than listen to it, right?
RW:   When you say talking, you mean in a purely ordinary computer sense of having a lot of wires?
IC:   I can send stuff, data instructions, to every processor
RW:   In the usual sort of way?
IC:   Excuse me. At the rate of 100 megabits.
RW:   And this is in the usual sort of sense?
IC:   In parallel
RW:   This is in the usual sense of having addressing lines and data lines?
IC:   No! No!
RW:   No? OK. So how do you do it?
IC:   | Look. If you have a processor
RW:   Just explain. Don't draw a picture
IC:   If you have a processor | it obeys a series of instructions
RW:   The processor does, yes. It carries out a series of instructions.
IC:   CARRIES OUT a series of instructions. Those instructions either come from its own RAM, or they come from me
RW:   Or from its ROM?
IC:   And when they come from me, I can send those instructions to every, all the one million processors in the machine, at the same time, at the rate of 100 megabits.
RW:   Using the ordinary addressing modes system, or something like it? They don't get in there by magic? If you have wires connected to them
IC:   I'll get the pictures, shall I?
RW:   No, don't bother. I'm, no, no, honestly your explanations are worse than fucking Hillman's I tell you
IC:   So one access is 100 megabit, coming in.
RW:   But that's done by conventional means, of wires, or something like that?
IC:   Sure. It's all conventional
RW:   Bits, yeh?
IC:   There's nothing unconventional. It's one wire and it's serial
RW:   I think I see what you're saying. All I'm saying is under some circumstances like a lightning flash that in effect would affect umpteen of these. Not a huge amount. Not, but, I mean if you're simulating, suppose you're trying to simulate just one cumulus cloud–
IC:   Shall I tell you one other thing? This–
RW:   [Loud] Suppose you're trying to simulate one cumulus cloud, right? That's a reasonable thing to do. In that case, IF you get a lightning flash, then this thing won't work because
IC:   Yes it will work
RW:   Because it just won't have–
IC:   Because I wouldn't do it in real time
RW:   Well, I mean, I'm just pointing out, I'm being deliberately awkward and picking an example–
IC:   Yeh, you are
RW:   Where, similarly if you're trying to model something where, suppose you've got a bit of wire and it re–emerges somewhere else, like a–
IC:   Well, it doesn't
RW:   –pylon or something. Well, if it did happen, then it would bugger up your system. So I can see why people MIGHT–
IC:   So that's why, I went to Cray, and I spent years researching what the big powerful computers were actually working at. And I found that 70% of their time they were working at problems as described by me ALREADY although the machine is inappropriate to that
RW:   You mean; weather forecasting and
IC:   Finite element analysis, computational fluid dynamics, and OTHER problems [Note: Feb 97: it occurs to me that modelling of huge numbers of polarised molecules – as in water – might be possible – RW]
RW:   Tell me, have they got any useful results out of this? It seems extraordinary–
IC:   Yes!
RW:   Why not just approximate? Have they, are they just too lazy – if you take something like fluids, I'd have thought you could easily, yeh, if you've got a big column of stuff with concrete in, I'd have thought you could just bung in a bit more concrete and save yourself a billion dollars worth of Cray, wouldn't you? And it just seems absurd that–
IC:   Anyway, the copper mine industry–
RW:   Why do they do it?
IC:   In order to save money, not to waste money
RW:   But that sounds ridiculous, doesn't it
IC:   No. And if that's not true it doesn't matter, because you start saying why don't you just dig more holes under the sea and find the oil, rather than compute where it must be on the basis of reflections–
RW:   No, I'm not saying if your example is correct, if they just build a pillar–
IC:   It doesn't matter. That's what the guy told me. The point is I go round and I say what, in fact, are the problems that people are addressing? Now he can have lied to me and it doesn't matter
RW:   No you don't see the point. If in fact the Crays are just expensive toys, then they're not going to be INTERESTED in reducing the price are they, it's not, why should they bother. [NB: I was thinking of something I'd been told by one of the Ardleys, that weather forecasting in fact is just a side issue of [the Ministry of] Defence; the data is collected, but really the point is for defence information]
IC:   They're not expensive toys
RW:   Well, they are expensive toys! They are generally believed to be more expensive than most computers, aren't they? Isn't that a fair statement?
IC:   That's the only way you can do the problems that I'm interested in
RW:   Well I'm saying it's not the only way. I mean with finite element analysis, obviously the smaller the elements the longer it takes. If you cut it down by ten, it takes a thousand times as long, obviously. If you did an approximation just based on, just building a model, you get a hole in the ground, stick some stuff in, you need roughly–
IC:   You're talking nonsense
RW:   I'm not talking nonsense. In most cases ultimately they use a rule of thumb, don't they. You don't use a Cray to model the, a house, do you
IC:   Look. You've got a series of TV frames which has come down a certain channel and it's been degraded. You wanna try to restore it. Right. So you analyse this frame compared with the previous frame and the next frame, and you analyse that bit. And you're just coming along and saying, well why not be more approximate, and analyse a one inch square. I mean that's just a stupid remark–
RW:   It's not a stupid remark. If you | use a telephone, it's design is roughly speaking, the quality of it is enough so you can make out what people are saying. Aren't you. Isn't it. You don't need, you don't to make it any more precise.
IC:   So you don't need array processors. Fine!
RW:   Well, it may be that these people don't need, it may be that they're using their Crays inefficiently, they've bought the thing, it's no fucking use to them really, just as with electron microscopes
IC:   Yeh. Yeh
RW:   That MAY be the case. If that is the case, then it's understandable that they look at the stuff, they think, well you know, this bloke, he's missing–
IC:   That's right that's right
RW:   –the whole point. Well you don't seem to face this possibility. You're assuming they use their Crays sensibly, whereas from my point of view it seems highly doubtful that something as PIFFLING as you know building a little pillar, you know shoving a bit of concrete in it, actually needs–
IC:   Yes, just throw that one out
RW:   Well I'm just saying that might be a possibility, all the examples I've heard of the use of Crays, strike me as being people you know who have been funded and are playing around with these things – you know, good luck to them in a sense – they've got their simulations of the atmosphere which are known not to work properly, you know, which are only accurate to half a mile or so–
IC:   Anyway, you've understood that, so–
RW:   I think so
IC:   So I've said enough about that. Haven't I. So what's the next subject?
RW:   Let's talk about Hillman
IC:   Oh yeh. Do we watch the tape right now?
RW:   You owe me some money for this tape, by the way
IC:   Yeh. How much?
RW:   I dunno. How much do you want to pay me for it? Is a tenner OK?
IC:   Yes, all right.
RW:   .. computer architecture
IC:   Yes. He said something like, oh you know there's that business, they think, that in computer architecture, you might have lots of – you know how if two people are working on a databank it's a nuisance cos one might modify it while the other's dealing with it, you know, and he'd gone down that route, and and he's got all these processors and all these databanks, all the toing and froing, and he says, Of course, I'm a theoretician, and I don't bother about the hardware! Ha ha! So he's doing the theory of how lots of processors will access lots of data banks! And he's not interested in the hardware! Well, the point is, once he says that I just walk away. Because | because I mean supposing you were I mean the equivalent is you say I'm a I'm a
RW:   Builder, not interested in bricks?
IC:   Yeh, not interested in the physics of the brick. You know. In architecture–
RW:   How did he do it? Was it probabilistic or something? Queuing theory? That kind of business?
IC:   That kind of stuff. They end up with all these great big pages of mathematics and they send in their applications for the next year's grant and they have PhD researchers, and I've met those as well, and they're burrowing away. But the point is when it comes to actually building the machine, I went and built a machine, you know, they didn't build anything, they're just flotsam. Now that's computer architecture. If you go into electromagnetic theory, I know what all those guys know, the ones who are writing the textbooks. I mean Salaam, like he got a bit of a Nobel Prize for unifying the weak force and the you know, there's four forces. I mean | der | well I mean no way does he know any electromagnetic theory. Well, for one thing he's a mathematician. It's just, I mean, take wafer–scale integration. I had to know about the signals and so on, the noise and so on. It worked first time, right? Well, I had to understand how the signal worked. Well, you won't get that from Salaam. He wouldn't know where to begin, see
RW:   Hm
IC:   And and the point is one thing that's gone wrong is a man called Bohr, Born, no. He's quoted in my book Computer Worship. It's the Renaissance is you combine the theory with the practical. You know, and once you divorce them, then you're a scholastic. You're back in the Middle Ages. And and and that's nothing
RW:   Did you–
IC:   Um? and a lot of what's happening is they've gone back into the Middle Ages
RW:   Er did you ever considering re re relaunching, or redoing your Catt concept book?
IC:   Yeh, yeh.
RW:   It's a shame
IC:   Well, I can do it now, cos I can make, you know my book is twice as big as it ought to be. Well, I've, you've seen my new electromagnetism book
RW:   Oh yeh
IC:   Well now I can make it the proper size
RW:   Yeh, but that's not, I meant you know 'The Catt Concept' wasn't about electromagnetism
IC:   I'll just type it up, it won't take long; and then it'll be in print
RW:   Yeh, I think you should do that. Cos I mean you should publish it, or get it published
IC:   That's nothing to do with it. You just cut across what I was saying, didn't you?
RW:   No. I think you should relaunch The Catt Concept, that's what I'm saying!
IC:   That's nothing to do with the fact that the guy in Leeds, I'm told the ones you must go to visit round the country are the guy in Leeds and the guy in Glasgow. And when I go and visit the guy in Glasgow he's panicking. And when I visit the guy in Leeds he's panicking. Cos
[END OF TAPE]