History of Science and Technology Q&A (September 20, 2023)

All right. Hello, everyone. Welcome to another episode of Q and A about history of science and technology. And I'm happy to try and talk about. And it's things I know about, about general history of science and technology or particular things that I've personally been involved in, etc. Let's see. I see a number of questions saved up here. Aqua asks, was the 1970s truly the golden age of electronics? It's an interesting question. I mean, the, you know, just to sort of discuss kind of, you know, electricity became electricity in various forms. Like static electricity was known in antiquity, but electricity as something with wires and batteries and so on, that's kind of a thing of the. That really emerged in the early 1800s. And then the sort of uses of electricity for things like making light bulbs, telephones, and so on. That's kind of a late 1800s story. And then the whole understanding of kind of the relationship. Well, the idea of electromagnets, the idea that you could use electricity to make magnetism, that was something that I guess became known. Not sure exactly when solenoids and things like that were invented, but the idea of electromagnetic induction, the notion that you can induce a current with a magnetic field, induce a magnetic field with a current and so on, that got understood with folks like Faraday and then kind of mathematicized by folks like Maxwell in the 1870s, 1880s kind of timeframe, and then the expectation of things like radio from Heinrich Hertz, who famously, people asked him, what's this going to be good for? Well, it's just kind of an interesting validation of Maxwell's equations. But it was only subsequently that people understood that radio and the transmission of electromagnetic radiation through free space was an important thing in its own right. It's always interesting when there is a phenomenon in physics, for example, that gets recognized. And like, why is this useful? Nobody knows. Like, liquid crystals were discovered, I think, in the early 1900s, maybe long before the use for computer displays was understood. Ferrites, for example, were used for many forms of memory, disk drives, these kinds of things, before the age of solid state disks and so on. That ferrites were. I don't know when they were discovered, but I'm guessing certainly the materials were known for a long time. I think the phenomenon of ferrimagnetism is probably a phenomenon of the early 20th century, but certainly recognized long before the idea of storing large amounts of information. In a magnetic material. I mean, the big difference is in a ferromagnet, like iron is a ferromagnet. Iron Cobalt nickel are the famous ones. And then there are rare earths like gadolinium and so on, which are even stronger. Ferromagnets. The point there is that that's a place where all the atomic spins line up and you get sort of a big magnetic field for the whole sample. Ferromagnets, you can have these little domains where you have magnetism one way and one part another way and another part, which is really useful if you're trying to store a bunch of data on something like a magnetic disk. But any case, the sort of. That was. That was kind of the early evolution there. Then there was the invention of the vacuum tube. I think that was. Lee De Forest was the. Often usually attributed the invention of the vacuum tube. And that was kind of the idea of using the fact that a hot wire will emit a stream of electrons which can be sort of picked up elsewhere. But if that stream of electrons. And this only works if it's in a vacuum, because otherwise the electrons get absorbed by the air and other things. But you have this sort of hot wire and it's kind of streaming through and the electrons can be guided by a magnetic field, for example, or an electric field. But you can have kind of a gate that prevents the electrons from coming through and that allows you to make a switch. And that led to the things like triodes, which were, you know, they were big things, several inches tall, usually that were glass tubes that had vacuum inside and they glowed because there were actual hot wires that were emitting electrons. I know when I was a kid, there were still plenty of devices that had vacuum tubes. Vacuum tubes of the American name. In British English, they're called valves that. I'm not quite sure why they're called valves actually. Maybe because it's like opening and closing a valve, I'm not sure. But anyway, the triode was the quintessential version of this where you would have. There would be kind of. This current could flow from one place to the other with sort of a gate that would determine whether the current could flow or not. And that allowed you to make a switch and eventually an amplifier. Because the effect of that intermediate gate that you could make a small change there and there will be a big change in the amount of current that passed through. And that allowed you to effectively amplify the effect of what you were sending to the gate. So that was kind of the. There's a whole technology built up of vacuum tubes and they got very elaborate pentodes and that meant, you know, five legged creatures. And a whole elaborate system. And you know, the typical radio would have a few handful, two, three, whatever vacuum tubes in it as part of its electronic amplification mechanism. Okay, so that was the story until the 1940s, I think, if I remember correctly. Was it 1947, I'm not sure when the transistor was invented. And that was. The Bardeen Shockley Brittain. I think I certainly met Bill Bardeen. But the other two, not the Bardeen was subsequently was the person who figured out the. Was one of the people who figured out the BCS theory of superconductivity. So he got his two Nobel Prizes, one for transistors, one for superconductivity. And he was a professor at University of Illinois, actually at even overlapping with the time when I was there. But in any case, very dignified old fellow by the time I knew him. The. So in any case, the. The transistor was a. A thing invented by using semiconductors. Semiconductors were a type of material. Silicon and germanium are the two most famous that had been discovered oh long earlier, sometime in the 1800s, I think, to be something where kind of when you push them hard enough, they'll conduct electricity, but otherwise they won't. They have this gap where you need to push them with a certain voltage to make them conduct electricity. So then the idea for the transistor was to kind of mess around with that, to dope the semiconductor, mess around with that, with exactly where that transmission of electricity works to make a sandwich. I think the first transistors were probably NPN transistors with different doping in the three layers. And I think that's right. And that was something where the very first transistors were still, I think, fairly small compared to vacuum tubes, but were kind of this, you know, you could, you could see them. I mean, I remember I had a transistor radio when I was a kid circa 1965, 1966, that had. Was it one or two transistors in it. And they were kind of a things with a little, little, little piece of, you know, semiconductor, I guess. Very hard to see. Not, not tiny, you know, several millimeters across. And then a big sort of heat sink that was big piece of metal that was sort of dumping out the heat from the operation of this transistor. But that was. So transistors allowed for the possibility of sort of much more lower voltage kind of circuitry because vacuum tubes tend to use significant voltages, semiconductors tend to use, you know, six volts or whatever. The. Then I think the sort of the big innovation in the 1960s was the idea of microprocessors and the idea of integrated circuits. Because the original thing had been where you get a transistor. It comes an individual transistor with all three legs, and you would solder that into a circuit board or something, and that was the transistor. But then the idea arose, I think, at Texas Instruments, that one could make a. One could take one slab of silicon. Instead of doping the silicon just to have these layers for an individual transistor, you could dope the silicon differently in different places, and you could make a whole circuit on the piece of silicon. And that led to integrated circuits and microprocessors. And the technology was essentially, you make this photographic mask and then you optically just make the features, demagnify it, so to speak. And then you use ultraviolet light to have some acid, which either will cause you to etch the silicon or not, whatever. The whole elaborate these days, very elaborate and very expensive process to make microprocessors. But kind of there were a bunch of different technical things that had to exist for that to work. Like, you had to be able to make silicon crystals that were very pure and didn't have a lot of defects in them. You had to be able to control all of these processes really well. But the first microprocessors started to exist in the late 1960s, I guess, and I think they kind of were too late for the space program and things of that type. And I think the first place where most people got to see those things was in calculators, which kind of burst onto the scene around 1971, 1972, where you would have these four function calculators that were based on microprocessors and that ran off batteries and so on. And that was kind of a big place where many people got to see kind of microprocessors in action then. Well, in the decade that followed, sort of microprocessors showed up everywhere and in computers, I think, in lots of other kinds of devices. I think the kind of. The early idea was there were particular families of microprocessors and other sort of smaller scale things that are more like what would nowadays be called microcontrollers, Arduino style, things that showed up. And some of those technologies haven't actually changed much in a long time, but then started to show up all over the place. And then gradually the progressive technology of making microprocessors got better and better and kind of Moore's Law kicked in of every 18 months, double the capabilities. And that was a sort of huge engineering achievement over the course of many decades and leading to all sorts of. Leading to the semiconductor industry as it is today. I think in the, you know, there are different parts of the story. I mean, there's the story of microprocessors and how they advanced and the various times when people said we can do better, you know, silicon microprocessors, you know, we can use germanium instead, we can use gallium arsenide instead. None of those technologies really panned out big time. They all had the kind of force of economy of scale to do with silicon and CMOS and metal oxide semiconductor technology and so on. The force of the economies of scale in that area seem to exceed any potential advantages for these other technologies. I think some gallium arsenide stuff is used for very fast kinds of maybe optically connected microprocessor technology and so on. But there were, there were multiple attempts to use other technologies other than semiconductors to do electronics over the course of years. And in a sense, one could see the modern enthusiasm for quantum computing to be another iteration of that idea. Even some of the same technologies like superconducting, Josephson junctions and so on, are reappearing as they were. Things that had been discussed in the context of traditional microprocessors back in the 1980s and so on and hadn't really made it there and are now kind of being. Well, it's another thing where we could potentially feed different physics into making a computational system. Now another branch of the whole world of these things is analog electronics, digital electronics, where you have sort of data going around in discrete ones and zeros and so on. That's how essentially all computers work. But multiple times people have been enthusiastic about various kinds of analog approaches. And there's still analog circuitry in many kinds of devices, because in the end you've got some kind of electric motor. That's where it runs faster if you put higher current through it. Things like this, increasingly things that you might have thought were kind of analog circuitry, like the antenna aspects of radios and things increasingly that's as efficient to do digitally with software, radio and things of that kind. But. And sort of the, the experience has been that analog computing hasn't really been very successful. I mean, I remember I used an analog computer for a bit at Los Alamos National Lab back in the. When was. It must have been 1982ish timeframe. And it was kind of, you know, the idea there is Instead of doing everything in terms of bits and ones and zeros, you just have, and the voltage is 5.734 or something. And you're running that thing through an electric circuit that is a traditional circuit with inductors and capacitors and so on. And you're kind of taking, for example, a differential equation, and you're saying, I've got this differential equation I want to solve. I can convert that differential equation into something that corresponds to the differential equation for this electric circuit as an analog thing, as an analog circuit, so to speak. And then I can just see what the electric circuit does, and that will tell me things about the solution to my differential equation. That technology really never panned out in a serious way. It was very hard to be reproducible. It was, oh, it mattered whether the thing was hot or cold and things like this. And it just didn't seem very convenient. It wasn't the same kind of thing where you could say, oh, I've got this piece of software, let me give it to you on a floppy disk or whatever. The programming of an analog computer involved actual wires going from here to there. I think the one at Los Alamos had some whole elaborate scheme for doing something which was a little bit repeatable, but it seemed more like an old fashioned telephone switching system where the physical wire has to go from here to there, rather than a modern kind of virtual circuit digital switching system. So a few thoughts about that history. I think that the advance of kind of design for microprocessors and for analog circuitry, those have gone in two rather different tracks. I mean, microprocessor design, a lot of that. Early on, people thought about optimization of logic circuits. Oh, we've got this particular logic expression that it represents a piece of an. Adding an adder. For example, let's find the minimum set of gates, the minimum set of transistors that we can use to compute that particular function. That's one sort of direction. The other direction. By now, when there are billions of transistors, there are just all kinds of different issues about how you manage that design. And there are different levels of layers of design. Whether it's the kind of physical layer where you're worrying about are these two wires too close to each other? Or the logical layer where you're worrying about is there a bug in this algorithm that we're burning into our chip? And so on, And a lot of sort of development of kind of the methodology for designing microprocessors and large such systems. There was at various times there'd been various degrees of enthusiasm for the roll your own chip kind of approach. Silicon compilers was a thing in the 1970s particularly that people are very enthusiastic about. Then there are field programmable gated rays FPGAs, which are kind of an organized structure of circuit that you can pick, oh, I want these particular gates from this circuit to be used. And that's a way to kind of make in hardware something which is not an absolutely arbitrary optimized circuit, but it's something where you can put your particular kind of system and you can make it something which is implemented in hardware. And again, people had thought, oh, it's going to be so easy to make chips that everybody who has an algorithm is just going to make a chip for the algorithm. That really didn't pan out. Again, the sort of economies of scale to do with building general purpose microprocessors and the ease with which you could define new software really vastly exceeded the advantages that you got by sort of the roll your own chip kind of approach. An intermediate stage is microcode, which was a thing which one heard about more in past years than today, which is kind of like, well, you've got this raw microprocessor and it has all these gates and things in it and places where data goes from here to there. And there's a lot of detailed information about timing of things. And what people would do is, and I'm sure it's still done in a computer or a microprocessor. You would say, well, I've got this instruction which is actually going to be implemented in microcode. It's going to be implemented by having all these different gates open and close and signals sent from here to there at a very detailed hardware level. But at the level that's going to be exposed to software, it's just some particular instruction. And so that's another kind of intermediate approach. But I think in terms of analog electronics, sort of the design methodology for analog electronics has been very different from the design methodology for digital electronics, where it's all about logic, minimization and things like this. In analog electronics, it really is kind of, you've got these equations, you've got to solve them and it's not the same type of mathematics and so on. It's really calculus, continuous mathematics that's involved. And people try from time to time things like using genetic algorithms to say, let me invent a circuit for this. I think there's been some success in that of, you know, this, this thing I'm trying to achieve that has these particular characteristics. Can I invent a circuit that does that? That's going to have wires and capacitors and it's going to have an inductor with some particular, you know, inductance and so on, and can I make that thing? And I think there's been slow progress in that. It's kind of area where sort of early AI approaches have been fairly useful. But my impression is that if you were to compare sort of the level of sophistication of design for digital electronics to analog electronics, digital is far out ahead in that regard. And that's probably yet another reason why digital, in addition to the programmability, the fact that there's sort of software that you can exchange and so on, is why that has a great advantage there. Ah, let's see. We have all kinds of other questions here. Okay, Taz asks, what's the history of hacking? When did security risks become a prominent issue? Well, I know a few things about that. You know, I remember when I was a kid, probably must have been 1970ish timeframe, I remember hearing this rumor about how people, and I think it was MIT had discovered that if you were to type too many characters into a password field, that some of the data from that the code would assume that you're only going to type some number of characters. It's a buffer overflow exploit, so to speak, where sort of some of that extra stuff would get into the innards of the operating system and cause all sorts of bad things to happen. And I remember hearing that probably the first time I heard about any kind of exploit it was when I was like 10 years old or something. And I think the. I should explain that the term hacking was a term or hackers was a term that actually originated in the 60s, I think, at MIT and a bunch of people who I've known actually were involved in that. And it really just meant doing cool, surprising hacks with computers. It didn't mean breaking into computers. It just meant doing clever things with computers that nobody expected. And that led to, for example, a famous document called hackmem H A K M E M that I think came out in 1970, that I think my friend Bill Gosper was involved in that and several other people, Rich Chappelle maybe, person called Beeler maybe, who, I don't know, Michael Bieler perhaps. And that was this thing that's a list of, I don't know, 400 kind of cool things you can do with computers that's like, oh, this particular set of Machine instructions for a PDP10 will draw a totally weird fractal curve. Except they didn't call it fractals yet then. But. So that was kind of an early meaning to the term hacker and hacks was these cool things you could do with computers. I think then the sort of in public awareness, the really dramatic kind of moment when people became aware of sort of computer security in a big way was the Morris worm. Robert Morris, whose dad was a computer security researcher at AT&T@ Bell Labs. And I think I met the young Morris at some party when I was a consultant at Bell Labs when he was a high school kid. But he subsequently wound up as a student at Cornell and decided to build a computer worm that somehow got out. And this must have been 1988, 89, sort of got out and infected lots of computers around the world and started just replicating itself in the. Nothing could be alive. Nothing can. You know, if you self replicate, you must be alive. Well, this was an example of something that was only alive in the computer, so to speak, but replicated itself. And the main thing it did that was bad was that it just kept on grinding the CPUs of computers, sort of trying to replicate itself and trying to transmit copies of itself. And that meant the computers kind of ground to a halt, just running the worm rather than doing what they were supposed to do. Now, I suppose I must have known more about computer security by that time because one of the things we did as a company was we had a connection to the Internet back from 1986 when company was started. And I had arranged to make our gateway computer be a very obscure workstation computer made by Sony, whose operating system was based on Japanese and was just a very obscure computer that didn't make it commercially. But I guess I knew enough to decide that our gateway to the Internet should be that very obscure computer. And that turned out to be a huge win because the Morris worm was compiled for a bunch of different standard kinds of computers, but nobody would compile it for this incredibly obscure Japanese computer. And so, for example, our computers at our company didn't get infected by that worm. So I wonder how I knew to do that. It's a good question. I certainly had been. I mean, another sort of whole branch of computer security was cryptography. And that was something that I, for example, knew a decent amount about by that time. And I mean, just to give a little bit of the sketch of the history of cryptography even back in antiquity, it is said, and probably it's true, that Julius Caesar Used a Caesar cipher, which essentially takes every message and the letters of the Alphabet just get shifted 4 or 5 up or whatever it is. And the key is how many letters are you shifting the letters of the Alphabet by? And at a time when it's sort of an interesting question whether people would really not understand what that message meant because they just didn't have sort of the abstract thinking associated with, well, let's shift the letters down. But that was one early approach was just shift the letters in a message you give. Then there were vignette ciphers that came in, I don't know, 1400s, 1500s, things like that. They were things where the amount you shifted the letter would sort of gradually ratchet up as you went through the message. And then that led to all sorts of encryption devices that involved cipher wheels and so on. By the way, there was a whole other branch of using code books where you would say, you know, if you say 10, 4, it means whatever, where you would just have a book of code words. Another approach was the so called one time pad approach, where you would take some particular, well, either a sort of a secret book or even something where you just said, starting at this letter in this newspaper on this day, we're just going to take those letters and use those to determine the key. So if that next letter was a C, for example, then move the characters three letters forward. If it was a D, move them four letters forward or whatever else it was. So that was another kind of approach, approach to these things. I mean, there was one thing, I don't know if it still goes on, it certainly did during the Cold War was these very strange radio stations, numbers stations, I guess they were called, where you tune your radio to that frequency and it would just be somebody reading off a series of numbers. And I guess those were used for some kind of one time pad cryptography as part of the kind of whole Cold War espionage picture. But just a very strange thing to find as you kind of turn your radio dial through different frequencies. Anyway, in cryptography by the 1930s, there were these pretty elaborate machines that were being made that became famous in World War II in the Enigma machines. I think they had originated in countries like Sweden and so on. But where had there been a tradition of building these things? And there were kind of, you know, a series of wheels that would. And then sort of which made electrical connections and then some whole sort of way of scrambling those. And then the result was, you know, you press one button on an Enigma machine and wheels turn and somewhere it displays what the next letter that you should put in the ciphertext is. So that was kind of the state of the art in World War II. And there were slightly more elaborate versions than the Enigma machines that existed then. Famously, those were broken by the British at Bletchley park using these early electromechanical computers that would essentially just search through for different possible keys that might be used. And the question was given that you choose a particular key, is the message that you get out a plausible message, or is it gibberish? Is it German or is it gibberish? And that led people like Alan Turing to kind of think a lot about the statistics of messages. And I kind of suspect that Turing had pretty much figured out the sort of information theory idea that was later really developed by Claude Shannon. Turing and Shannon had certainly met. And at the end of the war, Shannon published all that stuff about kind of the redundancy of messages. You know, how much can you tell? If you've seen a cue, it must be followed by you. It's a very redundant thing to put the U there. But in general, in languages, there's a certain amount of redundancy. In English, it's maybe 50% or something. And that's the thing you use when you're doing cryptanalysis, because you need to know, is the thing that you are producing a plausible piece of text, or is it something that is sort of implausible? And you get to figure out plausibility from using this redundancy. So in any case, that was kind of the state of the art in. In World War II for cryptography. What emerged shortly after the war, shift registers, which were kind of a thing of electronics, where you have kind of a series of vacuum tubes or something, and you're constantly shifting the value always to the left, and then you feed it back. The one all the way to the left shifts shift to the right. And that's something you can use as a way of storing information. Because if it takes a while to shift down and it keeps on going around again, you can use that to just store the bits. But then the idea was developed in the 1940s, I think late 1940s, of having feedback shift registers where you have not just circulating everything around, but tapping into particular places in the shift register, taking the results and, for example, combining them together and using that to be the thing that you feed back. And then. So by the 1950s, it was mid-1950s, it was known that you could get very long sequences of seemingly random numbers. And Sol Galam, who I knew was the person who really figured out how to mathematicize that question and turn it into a question about polynomials over finite fields. It became sort of the big application area for polynomials over finite fields. Polynomials where the variables can just be 0 or 1 mod 2, things like this. And one thing that happened was that you could figure out these particular. You could represent the taps on the shift register by polynomials by just sort of saying the tap is at position 11 or something. So you have an X to the 11 and it's also position 7, so you have an X to the 7 and so on. And from the characteristics of those polynomials, polynomials, you could work out the cycle, how many different configurations the shift register could take on before it cycled back to the original one. And so then there were these maximal length feedback shift register sequences that got developed by Sol Galam. And those have, I mean, interestingly, those are probably the single most used algorithm in history is the idea of a maximal length linear feedback shift register. Those are used, you know, GPS satellites use them. That's what they're transmitting. Cellular telephony uses it for code development, multiplexing. The thing that it's even used a lot in, for example, transmitting data to computer displays and so on. Partly because very easy to decode a linear feedback shift register sequence. It has the nice feature that it's always sort of having ones and zeros sort of flapping around. And you never have something where there's zero voltage on the wire. And it's often very convenient in electronics to never have zero voltage or to not have zero voltage for very long. So it's better to send this thing encoded in this form where it's easy to decode, but it's always, it's always kind of not just making a zero. So in any case, feedback shift registers were a thing. I'm not sure to what extent they were used for cryptography. I kind of suspect that particularly in the Soviet Union, there were long linear feedback shift registers that were used for cryptographic Systems in the 1940s and 50s and so on and perhaps beyond that time. Then in the US the sort of. The next big step in cryptography was the Data Encryption Standard DES, which arrived around 1970 ish. And that was a thing that was kind of an electronic version of something, a little bit like an Enigma machine, but not quite the same setup. It had these sort of. You would have these blocks of data, zeros and ones, and you would have substitutions where you take one block and replace it with another block and you would permute blocks around and so on. And so that was one of the key parts of that technology were the so called S boxes, the substitution boxes, which would say, whenever you have this sequence of ones and zeros, replace it by that sequence of ones and zeros. One of the things that was very mysterious was why were the S boxes chosen the way they were? Because they just seemed like random kind of updates of collections of ones and zeros by other collections of ones and zeros. I eventually found out that in the late 1950s, the US National Security Agency, for example, had a whole bunch of mathematicians involved with this, investigating these kinds of sort of one dimensional systems where you would have a series of bits and you would take blocks of bits and replace them with other blocks of bits. Well, that turns out to be my favorite computational system of one dimensional cellular automaton, essentially. And so they had investigated and even tabulated lots of kinds of properties of one dimensional cellular automata, looking for ones that had certain kinds of reversibility properties. And some of the ones they found, I think turned into the s boxes of DEs because they were things that had sort of been identified as interesting in connection with linear feedback, shift registers and cellular automata. So in any case, that was for a long time the market was sort of flooded with kind of government subsidized encryption equipment that was based on des and that was sort of the standard for commercial encryption. There were then companies that popped up, I think particularly in Switzerland and Israel were the two sort of leading places where companies would pop up with various kinds of encryption techniques that they said, our encryption is better than other people's encryption. And look, it comes from this fine, fine, has this fine lineage. That was a time when a lot of the, my encryption is secure came from saying, and you don't know how my encryption works inside. And it had been a big deal that I think it had been folks in Poland who managed to get hold of a German Enigma machine in World War II. And that had been kind of a key part of being able to decode Enigma machines was knowing how they worked inside. And the kind of, the thing that was the, that was sort of some of the tradition in the 1970s and 80s even about encryption had to do with, well, you don't know how it works inside. I think there had been a time, this is a story, I think from the Korean War. There was a US ship, what was it called? USS Pueblo, I think that was captured in the Korean War. And it had a bunch of encryption equipment and it was big metal thing that was its encryption system and they really didn't want it to be captured. And so somebody took an ax to it and tried to destroy it, but it wasn't successfully destroyed. And I think that was one of the things that was sort of a big impetus for US military, for example, to get more involved with sort of small electronics that among other things, the next generation of encryption equipment. If somebody was about to capture your ship, you could just destroy the encryption equipment and people wouldn't. The other, the enemy wouldn't be able to get hold of it. But any case, by the 1970s, 1980s the kind of the, there were these commercial encryption outfits. There was a lot of kind of IBM was very involved in des, the data encryption standard. And that was sort of typical commercial encryption was done with destination and people. Then onto the scene came public key cryptography, the RSA algorithm. I think that was 1984 or so, maybe 1983. I myself actually invented a sort of crypto system based on the Rule 30 cellular automaton. @ that time it was sort of a big issue because there were sort of all sorts of questions about well, if you find a good strong encryption technique, should you tell the world about it? You know, is that okay to do that or do, is that something that, you know, only governments should be able to do? And I think the thing that emerged was partly through the RSA crypto system. You know, by the time there was an article in Scientific American which had a circulation of, you know, two and a half million or something in those days about this thing, it was kind of out. And, and so the kind of things flipped to the point where the idea that nobody knew how your encryption system worked wasn't the source of security. It was that the actual that even if you knew how the thing worked, it was still sort of mathematically secure. And that was a sort of key change. And by the way, a lot of the systems that are used in, I mean something like RSA cryptography is good for things like key exchange where you have small amounts of data that you have to deal with cryptographically. But when you have gobs and gobs of data, you end up using one of these stream ciphers or one of these other kinds of non public key systems like the Rule 30 system, or like derivatives of DES which like the SHA SHA systems for hashing and so on are related to those whole S box, et cetera kinds of methods. But anyway, so encryption was a thing which by the, I would say by the late 80s was a thing people sort of knew about. And it was a thing one routinely did. I have to say, you know, the Unix operating system, for example, developed in the, in the mid-70s, was it had a function called crypt that was based on essentially the Enigma machine or maybe the, I think maybe the purple cipher. And it was, you know, you could go see its code and it was, it was just doing that kind of encryption. A kind of a story to tell against myself is that when I worked on developing the system called smp, which was kind of a forerunner of Mathematica and Morphin language back in 1979-1981 kind of timeframe, one of my sort of clever ideas was let's encrypt the source code so that it won't kind of go walking away. So we did that using. And I said, but let's modify the UNIX scrypt program because if somebody is going to break encryption, they're going to break the generic UNIX scrypt program. So let's modify it to have our own version that is less likely to be broken quickly. So, okay, that story, of course, the next step is. And we lost the password. So decades later, there I am trying to talk to everybody who'd worked on it saying, do you remember the password? Well, nobody remembered the password. It then turned out actually about, what was it, eight years ago. So young fellow who actually now works at the company was managed to break the encryption without ever finding out the password by just using kind of statistical features of things, managed to reconstruct the source without ever determining what the password had been fed into the modified crypt program. Somewhat frustratingly, we had the encrypted code of the modified crypt program. So it was kind of a bootstrap. Can you find this from that? And so on. And we didn't know that. But anyway, that was fairly. I would say cryptography was a known thing. I will tell another story about cryptography from those days. The ED editor, which was a line editor that came with Unix, had in a piece of very terrible user experience. If you typed the single letter X when you were in a mode when the editor was kind of listening for editor commands rather than taking sort of text for your. For your file, when you typed, I think the single letter X, it would encrypt your file and the next things you typed would be masked and they will be used as the password. Okay, so I was actually, I was visiting Harvard, I think, and using a computer there. This must have been 1982 or some such other around that time. 81, maybe, probably 82. And there I am Writing some program about cosmology and trying to compute some interesting thing, I thought, at least at the time, about cosmic strings. And at some key moment I kind of thought I was typing into the file, but actually I was typing to the editor, so to speak. And I typed an X and then I typed random other letters, I don't know what other letters I typed. And my file got encrypted. And that was kind of a disaster for that project because there I had, you know, the code I'd been writing was encrypted. There was nothing I could do about it. I subsequently, when I was a consultant at Bell Labs, I ran into the person who'd written that aspect to the ED editor and made a vociferous, you know, user experience complaint about how ridiculous it was that a single character typed could wind up with an encrypted file and so on. But it was part of kind of the minimal ethic of UNIX design that something like that was possible. And it's also very much in sort of the UNIX design methodology of just trust the user and give them lots of power, which I think was an interesting approach, but it did have some bad side effects like that. Okay, so in any case, more on the history of computer security. I think the Morris Worm was really the thing that woke people up to sort of the possibility of, of computer security issues kind of on, on the Internet, so to speak. You know, many things have happened since. Many strange creatures have been released sort of on purpose by mistake things. There's a whole strange world of computer security. And from what I gather, it's becoming more and more difficult to make kind of generic computer security exploits. So a typical thing that's done in modern times is there's an operating system and somebody finds a bug in the operating system and that bug allows you to type the right incantation or do the right thing at the right time. That allows you to kind of take control of the operating system, take control of the computer and go and do things with it. And I would say that at the theoretical level what's happening is kind of interesting, which is that as soon as you can somehow get the computer to act as a universe, the program that you're interacting with, to act as a universal computer, you can get it to do anything. So for example, I remember an early hack that somebody I knew did was a printer. The printer had sort of connections from. This was a. Some kind of government facility where there was a. The printer had access from the non secure part and from the secure part. And it was one printer accessible from both sides. And what was done was something where somehow you took control of the printer. And even though it was just a printer, you could kind of get it to be computationally sophisticated enough that it could essentially run any program and then it could kind of crawl its way back onto the secure side and do bad things. And I think that the thing that is kind of relevant from a science point of view is kind of this principle of computational equivalence that I talk about a lot, that talks about how it's not very difficult to get a computational system to be capable of universal computation. That ends up being an important idea for computer security because it's kind of the thing that allows you, even though you think you have just a router, or you think you have just a firewall, or you think you have just a printer, really it's possible to sort of piece together the computational elements from that to do arbitrary computations, which allows you to do mischievous computations that were not intended, so to speak. It's kind of a long arm of Godel's theorem and undecidability that you're kind of being bitten with when you're kind of getting computer systems to do things they weren't intended to do. And that's sort of inevitable as soon as you give a certain amount of power to these computer systems. But what I'm told is that increasingly in developing operating systems and things like, that, the tools that exist for validating things like buffer overflow, for example, is really pretty. You have to not be paying attention to let your program be able to do buffer overflow, because they're pretty good validation tools and essentially theorem proving methods that allow you to show that that won't happen. I mean, it's still the case that a typical sort of way in to find a hack, to find a security vulnerability is techniques like fuzzing, where you say, I've got this input that I'm supposed to give, but let me just randomly change that input and see what happens and see whether suddenly some crazy thing happens. Now, also, in sort of the whole world of security, in things like smartphones, it's typical to just have these. These just different sections of capabilities that are kind of sequestered off from each other with only limited ways that they can communicate. And that helps make it really difficult to kind of break through and get to the wrong place, so to speak. But I think in. I forget now what the original question was here. Oh, yeah. I mean, the whole story of sort of vulnerability of computers with respect to computer Security, I think, was the thing that really came into prominence late late 80s, early 90s. You know, people have been concerned about cryptography from probably that we came into public awareness in the, in the probably early 1980s. And I think in terms of the more social engineering aspects of computer security, certainly they've been known to people in the biz for a very, very long time. But in terms of the general public, I don't know when that really became a thing. I mean, I certainly. Oh, gosh. I mean, as physical letters, one started getting, I started getting Nigerian scams in the, in what, the late 1980s, I think. And that was probably, you know, some of that kind of social engineering stuff. I'm sure that had happened earlier than that. Maybe I just wasn't the right mailing list, so to speak, to get those things. But the thing that was, I think those things are now, now have names that aren't specific to Nigeria. And I don't even know whether the things that one got back in those days actually came from Nigeria or whether that was part of the scam. But they tended to have the name, you know, they tended to have Nigeria in their name like French fries or something, at least back in those days. Well, let's see. I think, Yeah, no, I mean the whole question of things like fishing the ph and all these kinds of, of social engineering hacks are, You know, they have come to great, much more prominence in recent times. And it's always interesting, like for a company like ours, you know, we, we do testing about computer security systems and typically the technical side of it is fine and the human side of it is where things go wrong of people, you know, clicking on the link they shouldn't click on and so on. And we, we have instituted for a while now a fish of the month club of kind of highlighting sort of the most interesting phishing attempts that people report to our computer security team just as a way to raise awareness of what's out there. I don't know, maybe I could tell you some stories about. Let me think. Yeah, I think these are harmless. I mean, there are computer security exploits that we've been exposed to as a company and gosh, they get quite wild. And in some cases, I would say one we traced to Romania and one we're pretty sure came from North Korea and one longer ago came from a young kid in the US and all of these have had all sorts of elaborate interaction with various kinds of law enforcement. And I would say that it is interesting to see the levels of effort that have been Gone to at different times in different ways. I would say that the. Well, the very early one I'm remembering was from the early 1990s. And that was a. That was a case where this. This must have been. When was this. This was. Might be just after the web started to exist. Yeah, I think so, because I remember that. Well, it was a whole elaborate story which maybe I shouldn't tell, but maybe another time after I've thought through whether the story is all tellable, but it features the FBI and all kinds of folks and all kinds of strange events and all sorts of things like that. And other stories feature, well, various kinds of human hacking, which was defeated in various ways. I suppose one of. I'll say one thing, the one that ended up getting traced to Romania was. I think I was responsible for that piece of tracing, which was kind of the key thing was just you take apart the packets and you see that they have text in them. And what language is the text in? Well, these days would be easy to tell, but those. It wasn't quite as easy to tell. And gosh, it's in Romanian. Okay, that's a big clue. And that kind of unraveled that. I mean, there are other times in terms of computer security attacks, there are things like denial of service attacks where just people are throwing gazillions of packets at your computers. And it's not so much of a thing these days because there are all kinds of remediation techniques. But back in earlier years, that was a thing. And I remember one time that happened to us and we're like, why is somebody targeting us? It's not that interesting. And then we realized our IP address was one digit off from the IP address of an Internet casino. And that was the thing that was being attacked for ransom purposes. And we were just a typo away, so to speak. In any case, many, many things to say about that. Philip asked, I got to know Carver Mead at Caltech. Not as well as I would like. Certainly met him, but not as well as I would like. Let's see, maybe one more thing here. And then I think I need to go. Gosh, all right. There's a question here from Giovanni about what progress did anti gravity research? How far did it get in the 1950s and 1960s? Why did the research stop? Well, I don't know. In. In full detail, I will say that the sort of the. The original theory of gravity due to Isaac Newton in 1687. Newton's Universal Law of gravitation says, you know, bodies with. With masses attract each Other according to gravity with an inverse square law. And that was kind of the theory of gravity for a really long time. Then in 1915, Einstein came along with his general relativity, which was kind of a new relativistic theory of gravity, a new theory of gravity that included the effects of special relativity and the idea of space time. And general relativity was kind of an interesting theory. People worked on it mathematically, but it didn't get a lot of traction until really the 1960s. And it is a mathematically very difficult theory to work with. People didn't understand were there really black holes? What did that mean? That wasn't really understood until the 1960s. Was it a coordinate singularity? That was just kind of a glitch in the mathematics? Was it a real physical thing that you could see? That didn't really get understood until the 1960s. So by the 1960s, kind of the theory of gravity, Einstein's theory of gravity from general relativity was becoming an established thing. And there were starting to be all sorts of experimental tests of it. And there were sort of competing theories which added extra sort of detail. Like there was a thing called the Bran's Dicke theory, which was a theory that involved in addition to ordinary gravity, it would involve this scalar field, the graviton is a tensor field. And then that was largely excluded and so on. So in any case, sort of that kind of the theory of gravity didn't get established that well until the 60s and so on. People were still picking away at it for quite a while. I personally think that there are still challenges to it. And my guess is that in our models of physics things will work out a bit differently and that might allow one to understand things like dark matter and so on. It's worth commenting that that in the late 1970s I was involved in this. The kind of what became the inflation theory of the early universe involves sort of things with negative mass. Effectively, that's what dark energy. In the kind of modern view of cosmology, there is sort of a need to apparently to meet the experimental results, to have the idea that there is somehow in the universe something with negative mass that is causing things to be pushed apart rather than attracted together. That's a thing that's sort of very exotic and nobody knows what dark energy is, and maybe it doesn't even exist and so on. That's also a thing that sort of emerges in the inflationary theory of the early universe and so on. It's kind of a funky thing that theories of physics, particularly ones based on, well, early on, on spontaneous symmetry baking on this kind of hack that lets you kind of account for the masses of particles by introducing this field that suffuses the universe and that sort of gives mass to particles. But that whole sort of tower of physics has implications about sort of negative mass things that might exist, but those were not the motivations for anti gravity research in the 50s and so on. That is a modern reason why anti gravity wouldn't. Isn't necessarily, you know, if you take the orthodox view of physics today, it involves dark energy, which has anti gravity. So we don't happen to be able to get a bottle of dark energy. We don't know anything really about what it is, but that it isn't the case that, oh, physics proves that there can't be anti gravity. Not true. All the examples we know and have experience of are positive mass things which have positive gravity, not anti gravity. But in terms of sort of research in anti gravity, I think, well, there was a kind of the UFO time. Now these things have the new name of UAPs. But the time when people were there are strange things in the sky, that was probably had its peak, I think in the 1950s and maybe 1960s and possibly because there were strange things in the sky being tested, you know, various kinds of military things, various kinds of strange balloons and things like this, there perhaps were strange things in the sky. And you know, and it was something. A time when. I guess another thing that was true at that time was it was a time when the government in the U.S. for example, was the place that did more research. That was the place that had the most advanced science and technology. And the government was keeping some of those, kept things secret. And so there was kind of this vibe that if something was being discovered, it might be being discovered by the government and the government might be keeping it secret, as opposed to more modern times where the commercial sector has tended to be the thing that is most out in front in terms of most kinds of research and development. And that's a place where often the pinnacle of achievement shows up in a piece of consumer electronics. And it's a different vibe from the 1950s and 1960s where it was like the government behind closed doors is discovering the most, the most dramatic things. I think that probably made for a time when people felt that way. It was also the case that I think quite responsibly, the US government, for example, in the 50s and so on, wanted to investigate everything that might be conceivable because it was like, well, maybe the Soviets will be doing this, maybe they're using telepathy to control their drones or something. Maybe they're using anti gravity to do this. Maybe they're using. Maybe they've discovered how to wire up the brains of porpoises to be able to do this or that. Or all kinds of strange things that seem to us now like, oh, that didn't work. We know that isn't possible, but one didn't know it wasn't possible. And if one is kind of trying to make sure that one's military prowess, you know, is the best, so to speak, it makes perfect sense to investigate all those things. It probably, you know, you can ask what percentage of your budget should be spent investigating sort of wilder things. And I don't know the answer to that, but it's not zero because, you know, if it ends up being zero. And we've seen that in some commercial cases with large companies of, of oops, we forgot to investigate AI type thing. And that turns out to be sort of a bad idea. And I think it's. So, you know, the fact that these kinds of things were being investigated at that time is sort of not surprising. I don't really know what was done to make anti gravity. I mean, I think that people. Gosh, I'm trying to remember. I mean, there were. There were various kinds of perpetual motion machines, I remember, that were sort of mostly electrical and magnetic in nature. I don't know. Yeah, I don't know what was actually done to make the anti gravity machine. And I don't know whether it was kind of an electrical play or a thing spinning around and whatever. I mean, I think one of the things that gets tricky is in knowing sort of. Yeah, I'm trying to remember. There was one of these things. Oh boy. Somebody showed me at some point. And one of the things that's tricky is with things like gyroscopes or in general with. You can end up with something where sort of you pick it up and it does something you don't expect it to do because it's got the same momentum and it's kind of pulling itself in some weird direction and that can sort of make things. I'm trying to remember now that there was a particular example. Oh boy. Oh, gosh. Forgetting sometime. This must have been beginning of the 1980s or something. Some sort of claim of a. It weighs less than it should weigh. You put things in and you put more in and the thing weighs less when you put more in, so to speak, which is something that shouldn't be the case. The force of Gravity should increase when you put more stuff in there. And obviously you can have things if it's floating on the air, like a lighter than air, less dense than air, helium balloon or something. That's a different story. But this was a thing where in a vacuum you sort of put more in and it weighed less. I don't remember what that story was, but that was a, a time when those things. Well, that was a bit late for those things seeming plausible, but. Yeah, no, I think what really killed a lot of that stuff was when was this? This must have been the 70s, maybe the 80s, a certain senator Proxmire who started giving out these so called Golden Fleece awards for sort of crazy, apparently wasteful R and D efforts carried out by parts of the U.S. government or contractors for the U.S. government. And I think one of the famous consequences of that was there had been a SETI search for extraterrestrial intelligence effort that had been kind of a NASA type effort and that won one of these Golden Fleece awards and that was, oh, why are we spending money searching for little green men type thing. And that kind of killed government funding for SETI for a long time. And it sort of came back a bit in the form of astrobiology, which is kind of the modern polite version with slightly different objectives and slightly different methods. And most of the sort of funding and interest in seti, at least in the US went into kind of private, philanthropic, charitable kinds of entities. And not all that much was done. I actually myself was not that impressed with what was done. And I think it's one of these things that is a philosophically doomed problem that it's not a question of, you know, there's sort of a question of what it even means to have intelligence and then what it means to have communication and so on. And that's, you know, that's the, that's the thing to think about there, I think, Giovanni, commenting on crazy how much we still don't know. Yeah, I mean, I think one of the things to realize is that a lot of these things that have come up as, oh, it might be possible to have anti gravity, oh, it might be possible to have time travel. Oh, it might be possible to have telepathy. Oh, it might be possible, et cetera, et cetera, et cetera, in many cases, the story is more complicated than you think. The most obvious thing turns out not to be true. The most obvious thing, oh, you put this kind of paint on the bottom of your spacecraft and the thing will suddenly not feel the force of gravity that's probably not a thing. But to prove that it's impossible is basically impossible. In fact, even if we know, and we think we do know its framework, at least the fundamental theory of physics, to prove that anti gravity is impossible is something very unlikely to be done. Even for example. Well, one of the things, I looked at this a couple of years ago. Faster than light travel, which one sort of assumed based on kind of relativity and so on, it's kind of an axiom. Nothing goes faster than light. By the way, it's worth saying that one of the sort of stated axioms of relativity from 1905 is there is no preferred reference frame in the universe. There's no sense of absolutely being at rest or absolutely being in motion. It's all relative, so to speak. Well, that's a fine idea. Just happens that there is an absolute reference frame in our universe which is the center of mass of the cosmic microwave background, the center of mass of the radiation from the big bang. That is a thing where the earth is moving at 1000th the speed of light. Relative to that, there is a preferred frame of reference in the universe. Universe. Now one could say, well, the fundamental laws of physics don't depend on that. Well, that's largely true. There are very few things in the everyday world. Nothing really depends on the cosmic microwave background, at least in the current state of technology and where, you know, it is a background that is relevant for satellite communications, but the details of how fast we're moving relative to it and so on is not relevant to that. But the idea that the kind of the, the dogmatic statement there is no preferred reference frame in the universe is just not true. And it's not true also for these things that lead to the generation of mass through the spontaneous symmetry breaking, et cetera, et cetera, et cetera. It's a more complicated story, and it usually is. And so the statement nothing goes faster than light is again a more complicated statement in our models of physics where the structure of space is this giant network, this giant hypergraph. If you look at sort of the average characteristics of space, then yes, it will be true that nothing goes faster than light. But when you look at the micro details of how that network is put together. Well, if you found just the right piece of network to kind of trace down, you might very well find that it goes faster than light. It's kind of like the following thing. You've got molecules of air bouncing around in a room. Well, every molecule is basically going at the speed of sound. And. But if you are Kind of if you release a, you know, a scent somewhere in the room, it'll take a long time for that scent, the molecules in that scent, to diffuse over to the other side of the room. But in fact, every individual molecule, a micron at a time, is going at the speed of sound. It's just, it hits another molecule and it randomizes. That hitting of another molecule randomizes the direction it's going in. But if you could pick, if you were a molecular sized critter and you could pick. I'm going to ride this molecule now. Okay, I'm seeing it's going to have a collision with this one. I'm going to jump to this other molecule. I'm going to ride this one now. I'm going to ride that one now. You would actually be able to go at the speed of sound through the room, so to speak, by just hopping on the right molecule at the right time. Well, the same thing is true for the structure of spacetime. If you were able to sort of hop on the right atom of space at every moment, you would be able to go faster than the speed of light. Small problem you, aren't you. By the time you're sort of grinding down to the size of the atoms of space, so to speak, the atoms of space are the indivisible final points that space is made out of. And there's not a lot of you. You know, we might each be 10 to the 100 atoms of space. And by the time we're sort of going around at the level of atoms of space, There isn't enough. We don't get to stuff a whole 10 to the 100 atoms of space. Lump of something. That is us, the thinking human kind of entity, it's not us anymore. So as a practical matter for us, we presumably can't go faster than light unless we find some bizarre way of distorting space time or we find some space tunnels that I've sort of talked about a bit. It's the version of sort of every day going faster than light is only available to individual atoms of space, so to speak. And even then, not really because like the molecules in the air, they're sort of getting randomized every moment. But it is interesting, It's kind of that even these things where one sort of, where physics has in some level correctly dogmatically said this works, this doesn't work, and so on, there are footnotes that arise. It might take 50 years or 100 years, but there are footnotes. And sometimes you get to drive a giant truck through one of those footnotes sometimes. You know, I remember back in the late 1970s, people explaining that you just couldn't transmit computer data faster than 300 bits per second, that if you had a copper wire that was part of your telephone wiring or whatever, just wouldn't work. There was a theoretical reason why 300 bits per second, or maybe 1,000 bits per second was the fastest you'd ever transmit data. Well, there were loopholes. There were loopholes about. Well, there's this noise that gets generated, and there are patterns to the noise. And if you detect the patterns to the noise and make predictions of that, then you can suddenly send through more data and so on. And pretty soon we're sending through millions of bits per second through those same kinds of copper wires. And so we found the loophole, so to speak, and drove a giant truck through that loophole. And so these things where people say it's impossible to do an xyz, I'm always a bit skeptical about that, because, among other things, you have to define very precisely what the IT that you're trying to do actually is. And undoubtedly it is the case that the kind of skin depth, of transmission of an electrical signal of this or that kind in a copper wire is the way people said it was back then. But it turns out if what you want to do is transmit computer data, then you don't have to do it in that particular way, and so you evade that particular limitation. All right, well, I think we should wrap up, and I'm being told I'm late for something else. So thanks for those interesting questions, and I look forward to doing this another time. Bye for now.