WEBVTT - The Lighter Side of Tech 0:00:00.160 --> 0:00:07.240 Brought to you by Toyota Let's Go Places. Welcome to 0:00:07.400 --> 0:00:14.440 Forward Thinking. Hey there, and wasn't a Forward Thinking The 0:00:14.560 --> 0:00:17.360 podcast that looks at the future and says revped up 0:00:17.400 --> 0:00:20.680 like a deuce. Another Runner in the Night. I'm Jonathan Strickland, 0:00:20.840 --> 0:00:24.040 I'm Joe McCormick, and our other host, Lauren Vogelbaum is 0:00:24.120 --> 0:00:28.960 not with us today, but she will be back soon. Yes, So, Joe, Uh, 0:00:29.720 --> 0:00:33.640 have you heard of Moore's law? Moore's Law? We've talked 0:00:33.640 --> 0:00:36.560 about it on this very podcast so many times now, 0:00:36.760 --> 0:00:39.120 so I guess that means you have, in fact heard 0:00:39.120 --> 0:00:41.640 of it. I've heard of it. I've talked about it. 0:00:41.720 --> 0:00:44.000 In fact, I know a lot of interesting things about it, like, 0:00:44.080 --> 0:00:46.440 for example, the fact that it's not really a law, 0:00:46.640 --> 0:00:49.440 is it. No, it's an observation. Gordon Moore made this 0:00:49.560 --> 0:00:52.760 observation in a paper that and I'm paraphrasing here because 0:00:52.760 --> 0:00:54.040 I didn't write it down in my notes, but it 0:00:54.080 --> 0:00:57.040 had something to do with something like cramming more components 0:00:57.080 --> 0:01:00.160 on a silicon chip is something along those line. But 0:01:00.200 --> 0:01:05.600 he was essentially observing that because of improvements in manufacturing 0:01:05.640 --> 0:01:09.000 processes and the fact that they were able to find 0:01:09.520 --> 0:01:14.399 uses for more powerful microprocessors, that we were seeing a 0:01:14.600 --> 0:01:18.480 doubling of the number of components on a chip every 0:01:19.480 --> 0:01:22.160 two years or so. It all depends upon the era 0:01:22.280 --> 0:01:25.160 that you look at Moore's law. It can go from 0:01:25.200 --> 0:01:28.199 as little as twelve months to as long as two years. 0:01:28.360 --> 0:01:30.400 I think it. I think it started as a year 0:01:30.440 --> 0:01:33.520 and has extended now. But the generalization often is like 0:01:33.600 --> 0:01:37.320 eight every eighteen months or so. Some people might say, um, 0:01:37.400 --> 0:01:40.440 you should see in the way this manifest practically is 0:01:40.640 --> 0:01:45.440 a doubling in computer power, Yes, computer processing power. So 0:01:45.920 --> 0:01:50.040 while it may not be a physical doubling of actual 0:01:50.080 --> 0:01:54.560 components on a microprocessor, the outcome now is that it's 0:01:54.600 --> 0:01:58.080 a doubling of the microprocessors computing power. This this is 0:01:58.120 --> 0:02:01.960 all about, how you know, through the entire history of 0:02:01.960 --> 0:02:05.080 what is known as Moore's law, we have tweaked the 0:02:05.120 --> 0:02:09.440 definition to suit whatever the current state was. That's true, 0:02:09.440 --> 0:02:11.680 we have extended the time period a little bit. But 0:02:11.880 --> 0:02:16.360 it is worth noting that this prediction has been remarkably consistent. Yes, 0:02:16.480 --> 0:02:18.360 and some people might say that it's sort of a 0:02:18.400 --> 0:02:22.600 self fulfilling prophecy that, but by stating this observation, it 0:02:22.639 --> 0:02:26.440 has helped us. Uh, not us like Jonathan and I, 0:02:26.560 --> 0:02:30.480 but much smarter people, computer scientists, much more educated and 0:02:30.520 --> 0:02:33.680 capable people, really keep this thing going, right. It's it's 0:02:33.720 --> 0:02:36.480 it's kind of a goal for people who work in 0:02:36.520 --> 0:02:40.320 the computer's field to say, look, we want to keep 0:02:40.440 --> 0:02:43.760 this this prediction true. We want to make sure that 0:02:44.160 --> 0:02:47.840 the devices that we're designing in two years time are 0:02:47.880 --> 0:02:50.040 going to be twice as powerful as the ones we're 0:02:50.080 --> 0:02:53.440 designing today. Therefore, we've got this goal to shoot for, 0:02:53.720 --> 0:02:58.160 and it means making some pretty amazing leaps and engineering. 0:02:58.480 --> 0:03:01.400 Of course, the sad thing about worse law is that 0:03:01.440 --> 0:03:04.919 the honeymoon can't last forever. Yeah. So Moore's law is 0:03:04.960 --> 0:03:07.720 one of those things that has been predicted to be 0:03:07.919 --> 0:03:12.359 on death's door ever since the probably nineteen eighties, maybe 0:03:12.400 --> 0:03:15.320 even earlier. Yeah, and it hasn't been so far, but 0:03:15.560 --> 0:03:18.720 we know that at some point it has to, at 0:03:18.800 --> 0:03:22.160 least if we're going to be building our computers based 0:03:22.240 --> 0:03:25.919 upon the same architecture that we've been using up till now. Right, 0:03:25.960 --> 0:03:28.200 And so there are a lot of different things that 0:03:28.280 --> 0:03:31.440 you might sort of notice will end up working against 0:03:31.520 --> 0:03:33.639 Moore's law. One of them, I think is just sort 0:03:33.680 --> 0:03:36.640 of a funny observation that has been consistent for a 0:03:36.640 --> 0:03:40.240 while now. It's referred to as pages law. I think 0:03:40.280 --> 0:03:43.320 it was coined by Sergey Brynn Right of Google. Uh, 0:03:43.360 --> 0:03:45.920 it may have been. I'm not familiar with the actual 0:03:45.960 --> 0:03:48.360 origin of it. I do know what the law is, however, 0:03:48.400 --> 0:03:50.720 I think it was well, yeah, so what is the law. 0:03:50.720 --> 0:03:53.680 The law is that as we get more powerful hardware, 0:03:54.040 --> 0:03:57.720 software will rise to meet the challenge to make that 0:03:57.800 --> 0:04:00.640 hardware just as slow as what we're used to exactly. 0:04:00.680 --> 0:04:04.160 So it's like if every eighteen months your computer hardware, 0:04:03.960 --> 0:04:06.560 you know, a new computer will get twice as fast, 0:04:06.880 --> 0:04:10.280 every eighteen months software gets twice as slow. Right, the 0:04:10.320 --> 0:04:14.000 software becomes more complex, more bloated, so that crammed with 0:04:14.080 --> 0:04:17.279 features you don't want that I could take longer to execute. 0:04:17.400 --> 0:04:20.880 And this maybe one explanation for why it just seems 0:04:20.920 --> 0:04:23.240 like that computer that used to be so speedy is 0:04:23.320 --> 0:04:26.200 really sluggish now, or or even just that the brand 0:04:26.200 --> 0:04:28.960 new computer you bought is as fast as the brand 0:04:28.960 --> 0:04:31.200 new computer you bought two years ago. It's just now 0:04:31.200 --> 0:04:34.560 it's running software that's two years further along as well, 0:04:34.920 --> 0:04:38.080 So everything has evened out. The playing field has not 0:04:38.279 --> 0:04:43.080 changed at all. In other words, if software manufacturers made 0:04:43.200 --> 0:04:46.400 a product and then never updated it, so it was 0:04:46.480 --> 0:04:50.120 the same. Like, let's say that there's a productivity suite 0:04:50.200 --> 0:04:52.719 that you like and it has never been updated. It's 0:04:52.760 --> 0:04:54.960 the same suite year after year after year. You can 0:04:55.000 --> 0:04:57.320 still get it, but it's no different than it was 0:04:57.360 --> 0:05:01.000 when it premiered. Your computer would run that it was nothing. 0:05:01.080 --> 0:05:03.000 It would be just lightning fast. But that's not the 0:05:03.000 --> 0:05:05.520 way the world works. In order to make money, these 0:05:05.680 --> 0:05:08.640 these companies continue to upgrade their software, which means that 0:05:08.720 --> 0:05:12.320 you are now running a more um complicated, bloated you 0:05:12.400 --> 0:05:15.920 might say, software suite, and it runs just as quickly 0:05:16.120 --> 0:05:20.120 or slowly. There's your features, Jonathan, their features. They make 0:05:20.160 --> 0:05:24.640 it better, right because because more is always better, right, Like, 0:05:24.839 --> 0:05:27.400 so I definitely want to have that feature that no 0:05:27.640 --> 0:05:32.080 person ever in the history of mankind has ever actually used, 0:05:32.440 --> 0:05:34.080 but I want to be there just because that way 0:05:34.120 --> 0:05:37.159 I have the option. Yeah. So there is pages law, 0:05:37.160 --> 0:05:40.080 and that's been tracking More's law for a while now. 0:05:40.120 --> 0:05:44.840 But there are also some physical limitations More's law might encountered. 0:05:45.120 --> 0:05:49.880 So not just imposed by over ambitious or greedy software developers, 0:05:50.440 --> 0:05:54.240 but by the laws of physics themselves. Yeah. So one 0:05:54.279 --> 0:05:58.799 of those is called dinnered scaling or Denard scaling, depending 0:05:58.880 --> 0:06:01.480 upon how you prefer to pronounce it. Right. Well, technically 0:06:01.480 --> 0:06:05.800 we're talking about not denards scaling, but the end of right. 0:06:06.000 --> 0:06:10.320 So this scaling is all about the way these these 0:06:10.360 --> 0:06:14.840 microprocessors receive power. Right. And then as long as uh, 0:06:15.040 --> 0:06:18.560 the scaling holds true, then the amount of power we 0:06:18.640 --> 0:06:23.480 need to keep these micro transistors working microprocessors rather not 0:06:23.520 --> 0:06:28.320 transistors working, will will be neck connect so that the 0:06:28.360 --> 0:06:33.320 development will maintain, so we continue to have optimal efficiency 0:06:33.800 --> 0:06:37.479 as far as we can based upon electronics. Yeah, it 0:06:37.520 --> 0:06:40.160 should be obvious, but I guess worth saying. The chip 0:06:40.200 --> 0:06:43.440 in your computer is an electronic device. It needs electrons 0:06:43.480 --> 0:06:47.360 flowing through it to work. Yeah. So here's the issue. 0:06:47.800 --> 0:06:50.800 The problem is that once we got to having components 0:06:50.880 --> 0:06:53.160 of a certain size. Remember the whole idea here is 0:06:53.200 --> 0:06:55.760 that we're maturizing components so that we can fit more 0:06:55.800 --> 0:06:58.440 of them into the same physical space. Right. Your laptop 0:06:58.520 --> 0:07:02.159 doesn't get twice as fast becoming twice as big because 0:07:02.440 --> 0:07:04.880 twice as fast staying the same size. Right, And that's 0:07:04.960 --> 0:07:08.239 because we have managed to find new architectures that either 0:07:09.240 --> 0:07:13.240 either use the the features that we have designed more effectively, 0:07:13.920 --> 0:07:17.680 or they have decreased the size of those individual components 0:07:17.720 --> 0:07:20.120 so that you can fit more of them onto the chip. This, 0:07:20.200 --> 0:07:23.960 by the way, is similar to the tik talk approach 0:07:24.040 --> 0:07:28.320 with Intel. Intel calls their development TikTok. The tick is 0:07:28.360 --> 0:07:33.240 when they end up creating brand new majorized components. The 0:07:33.320 --> 0:07:36.400 talk is when they have figured out the ideal architecture 0:07:36.520 --> 0:07:38.520 to lay out those components so that they are at 0:07:38.520 --> 0:07:41.680 their most efficient. Each ends up meaning that you get 0:07:41.680 --> 0:07:44.880 a boost in your computer's performance. The first is just 0:07:44.960 --> 0:07:48.720 because it's just got more power, more horsepower in a sense. 0:07:49.040 --> 0:07:52.960 The second is because you have, uh, you have optimized 0:07:53.000 --> 0:07:55.280 the layout so that you get the most out of 0:07:55.280 --> 0:07:59.560 that horse power. Well, the problem with this Denard scaling 0:07:59.600 --> 0:08:01.960 coming to an end is that once you get to 0:08:02.440 --> 0:08:08.440 transistors below ninety nanometers in size, which we're already there, um, 0:08:08.840 --> 0:08:12.840 you start getting electrical current leakage issues, which means that 0:08:12.880 --> 0:08:19.240 the efficiency overall of your system decreases. So maybe, so 0:08:19.440 --> 0:08:22.640 that's saying like, even if you could continue shrinking the chip, 0:08:23.760 --> 0:08:26.360 you might not be getting a return on that exactly, 0:08:26.360 --> 0:08:30.080 and that would tie right back into Gordon Moore's observation. 0:08:30.280 --> 0:08:33.800 He had said that this was really an element of 0:08:33.880 --> 0:08:38.199 manufacturing and economics and less of some sort of innate, 0:08:38.760 --> 0:08:44.440 you know law that that guides computers becoming twice as 0:08:44.480 --> 0:08:47.720 powerful over a set amount of time. He was saying 0:08:47.760 --> 0:08:50.960 that it only holds true if the manufacturing processes and 0:08:51.000 --> 0:08:54.480 the economic returns makes sense. If they no longer make sense, 0:08:54.840 --> 0:08:58.760 the law falls flat. So it may come a time where, 0:08:58.840 --> 0:09:03.160 because the efficiency has dropped to such a point, it 0:09:03.200 --> 0:09:06.880 doesn't make sense for us to make electronic components any smaller, 0:09:07.000 --> 0:09:09.800 even if we are able to. So we might say, 0:09:09.920 --> 0:09:13.280 you know, engineering wise, we can make a microchip that 0:09:13.320 --> 0:09:18.920 has even smaller components, but because of electrical leakage, it's 0:09:18.960 --> 0:09:21.400 not going to perform any better than the chips we 0:09:21.480 --> 0:09:25.440 already have. Therefore, it makes no sense to actually do that. 0:09:25.640 --> 0:09:30.360 But Jonathan, Yeah, what about a crazy alternative. I like 0:09:30.440 --> 0:09:32.600 that you did jazz hands. Is this like a musical? 0:09:32.760 --> 0:09:34.800 Like like, we have computers that sing. I do them 0:09:34.800 --> 0:09:37.200 because the listeners cannot see them, and it brings me 0:09:37.240 --> 0:09:40.679 pleasure to have that secret. Okay, sorry, I I just 0:09:41.000 --> 0:09:43.840 ignore that. Listeners just pretend like I didn't say anything. Okay, 0:09:43.880 --> 0:09:49.680 this crazy alternative is today's microchip architecture, and our transistors 0:09:49.720 --> 0:09:54.000 are based on electronics. So you're you're using electrons, these 0:09:54.000 --> 0:09:56.440 little particles that are part of an atom. That's the 0:09:56.480 --> 0:10:01.000 same way all our other electronics. So negative to h 0:10:03.920 --> 0:10:09.200 your puns are strong. Okay, what if instead of using 0:10:09.280 --> 0:10:14.559 electronics we turned our attention to photonics. Oh so you're 0:10:14.559 --> 0:10:18.160 talking about fundamental particles of light, right, so instead of 0:10:18.280 --> 0:10:22.000 messing around with electrons, you mess around with photons. That 0:10:22.200 --> 0:10:25.360 is an interesting idea that a lot of people are very, 0:10:25.480 --> 0:10:29.200 very smart people are concentrating on right now. Yes, I 0:10:29.240 --> 0:10:31.880 want to clarify this transition. It was not my idea. 0:10:32.160 --> 0:10:34.960 It is something we have read about. Yes, this is 0:10:35.000 --> 0:10:37.520 something that we wanted to talk about because it's really 0:10:37.559 --> 0:10:43.880 an exciting proposition, especially for certain types of computing. It 0:10:43.920 --> 0:10:47.840 may it may turn out that there's never a practical 0:10:47.840 --> 0:10:52.360 way for this to become like the average computers architecture. 0:10:52.679 --> 0:10:55.080 On the other hand, it could very well become the 0:10:55.120 --> 0:10:58.280 future of computing. It could, it could. It's really early 0:10:58.320 --> 0:11:01.920 to say. Right now. The feel this still developing. We 0:11:01.920 --> 0:11:05.400 we are still working on creating the fundamental building blocks 0:11:05.440 --> 0:11:08.720 that this sort of computer system would depend upon in 0:11:08.800 --> 0:11:12.040 order for it to work as we understand computers to 0:11:12.080 --> 0:11:14.840 work today. And here's part of the issue before we 0:11:14.880 --> 0:11:19.439 even get into all the details. Part of the problem 0:11:19.480 --> 0:11:24.160 is that we've we've really written that electronic computer train 0:11:25.400 --> 0:11:28.520 really far right. We have advanced to such a point 0:11:29.160 --> 0:11:33.600 that in order to transition to a completely different computing platform, 0:11:33.679 --> 0:11:36.400 or at least a different architecture, we would need to 0:11:36.480 --> 0:11:38.839 have that architecture be advanced to the point where it 0:11:38.880 --> 0:11:42.560 could at least be a lateral move uh initially and 0:11:42.600 --> 0:11:46.760 then to continue to develop in order for Moore's law 0:11:47.280 --> 0:11:50.280 to remain at all relevant or at least, you know, 0:11:50.960 --> 0:11:54.760 applicable in some way. So, oh, I see, so like 0:11:55.160 --> 0:11:57.320 it would be very frustrating if there were a twenty 0:11:57.400 --> 0:11:59.880 year gap in Moore's law while we're waiting for for 0:12:00.000 --> 0:12:03.000 potonic computers to catch up to electronics, right, we would 0:12:03.080 --> 0:12:06.120 we would essentially have plateaued with electronic computers. We would 0:12:06.160 --> 0:12:09.839 not be able to really get measurably faster. We might 0:12:09.840 --> 0:12:11.880 be able to do some tweaking here and there, and 0:12:11.920 --> 0:12:15.160 we'll talk about some at least one alternative towards the end. 0:12:15.160 --> 0:12:18.440 Of this episode to photonics. That could have us see 0:12:18.520 --> 0:12:23.880 some some development continue using a tweak on traditional architecture. 0:12:24.320 --> 0:12:26.920 But it would mean that we would see a cease 0:12:27.160 --> 0:12:33.480 in the rapid escalation of microprocessor performance until photonics got 0:12:33.600 --> 0:12:36.920 fast enough so that it could take over and be 0:12:37.320 --> 0:12:40.040 the next new thing. And maybe that both are working 0:12:40.040 --> 0:12:43.240 in parallel for a while. Um, we don't know, because 0:12:43.280 --> 0:12:47.520 photonics still are relatively young when it comes to optical computers. 0:12:47.520 --> 0:12:51.600 But we have been working in photonics for decades. Oh yeah, 0:12:51.640 --> 0:12:55.400 I mean you have already interacted almost definitely with tons 0:12:55.440 --> 0:12:58.800 of photonic devices. One of the big ones I would 0:12:58.800 --> 0:13:03.199 like to bring up is just optical fiber. Absolutely so 0:13:03.600 --> 0:13:08.440 it's really becoming like an industry standard. Yeah. I come 0:13:08.520 --> 0:13:13.880 from Chattanooga, Tennessee, where the the electric powerboard, the local 0:13:13.920 --> 0:13:18.000 one has optical fiber to the home. Uh. It's people 0:13:18.120 --> 0:13:22.320 love it there. It's super super fast, awesome throughput. It 0:13:22.480 --> 0:13:26.280 is a good system. People are really happy with it. Yeah. 0:13:26.320 --> 0:13:29.840 So optical fiber. Uh, this is this is kind of 0:13:29.880 --> 0:13:33.360 what has provided sort of a foundation for the photonics industry. 0:13:33.559 --> 0:13:36.920 The two two big developments that happened in the twentieth 0:13:36.920 --> 0:13:40.880 century that have made photonics a an interesting area of 0:13:40.920 --> 0:13:44.040 study was the development of lasers and the development of 0:13:44.040 --> 0:13:49.360 optical fiber. So fiber optics are it's phenomenal in that 0:13:49.400 --> 0:13:54.559 you can really find creative ways to transmit data, multiple 0:13:54.600 --> 0:13:59.160 data streams across a single optical fiber simultaneously. And if 0:13:59.200 --> 0:14:02.320 you think about just how optical fiber works, it's kind 0:14:02.360 --> 0:14:06.080 of amazing. I mean talking about beaming these uh, these 0:14:06.120 --> 0:14:10.080 wave particle pulses of light, pulses of light refracting through 0:14:10.120 --> 0:14:14.079 a tiny essentially like a glass kind of cord across 0:14:14.200 --> 0:14:18.760 miles and miles. It's really strange. So so you've essentially 0:14:18.760 --> 0:14:21.200 covered the basis. It's the idea of you've got I mean, 0:14:21.360 --> 0:14:23.760 if you want to be really really simple, if you 0:14:23.800 --> 0:14:27.560 were to really boil this down to a very simple idea, 0:14:27.680 --> 0:14:31.400 imagine that, uh, that you and a friend are standing 0:14:31.400 --> 0:14:35.280 across from a football field and you have created a code. 0:14:35.360 --> 0:14:37.360 Let's even say it's just Morse code. You haven't created it, 0:14:37.360 --> 0:14:40.600 you're just relying on Morse code. You each have a flashlight, uh, 0:14:40.600 --> 0:14:43.640 and you flashed the light on and off in uh 0:14:43.680 --> 0:14:48.000 in you know, using the code as your basis, and 0:14:48.080 --> 0:14:50.720 you send messages to each other essentially with a fiber 0:14:50.720 --> 0:14:53.880 optic cable. You have a very high tech version of this, 0:14:53.960 --> 0:14:57.400 although it gets a lot more complex. So you have 0:14:57.760 --> 0:15:02.040 emitters and sensors on either end that are converting these 0:15:02.160 --> 0:15:06.160 light messages back into electronic messages. Or take an electronic 0:15:06.160 --> 0:15:08.640 message converted to light, send it across the optical fiber, 0:15:08.680 --> 0:15:11.280 where when it gets to its destination it converts back 0:15:11.320 --> 0:15:15.280 into electric messages so that a computer can understand it. Right. 0:15:15.320 --> 0:15:18.840 So the neat thing is not only does this travel 0:15:18.840 --> 0:15:21.760 at the speed of light across that optical fiber, not 0:15:21.880 --> 0:15:24.080 only do you not have to worry so much about 0:15:24.120 --> 0:15:29.000 interference with other types of signals because photons are pretty 0:15:29.080 --> 0:15:32.240 good about not getting mixed up with that kind of thing. 0:15:32.320 --> 0:15:35.200 They're not into, you know, being influenced by the wrong crowd. 0:15:36.000 --> 0:15:38.520 You don't have to worry about electromagnetic interference with photons. 0:15:39.160 --> 0:15:41.480 The other cool thing is that you can use different 0:15:41.520 --> 0:15:44.400 polarities of light, because you know light you can polarize 0:15:44.400 --> 0:15:47.600 in different ways, or you can even use different wavelengths 0:15:47.680 --> 0:15:50.560 of light, which we perceive as color right in the 0:15:50.640 --> 0:15:53.960 visible spectrum. Anyway, you can use different wavelengths of light, 0:15:54.080 --> 0:15:57.440 and then you can transmit multiple messages across the same 0:15:57.480 --> 0:16:00.920 optical fiber simultaneously and they and interfere with each other. 0:16:00.960 --> 0:16:04.480 Photons won't mess with other photons, so you could have 0:16:04.920 --> 0:16:08.400 you know, ten different streams of data going across one 0:16:08.480 --> 0:16:11.040 single optical fiber. You could even have it going in 0:16:11.080 --> 0:16:14.640 both directions at once. It's bidirectional, so you could have 0:16:14.720 --> 0:16:17.600 this communication going on. It's superspeeds. That's why you're able 0:16:17.600 --> 0:16:21.360 to get that huge data throughput. So the neat thing 0:16:21.480 --> 0:16:25.200 is that we've even seen cables for computers come out 0:16:25.480 --> 0:16:29.560 using this technology. Things. Uh, the original development name was 0:16:29.640 --> 0:16:32.800 light Peak, and then Apple bought it and and rebranded 0:16:32.840 --> 0:16:38.080 it as Thunderbolt, So the initial Thunderbolt cables, I believe 0:16:38.120 --> 0:16:40.880 we're all still copper. But the plan was to move 0:16:40.880 --> 0:16:43.520 to fiber optics using this technology, where you could do 0:16:44.000 --> 0:16:47.520 multi threading through fiber optics and have really rapid data 0:16:47.520 --> 0:16:50.200 transfers like twenty gigabytes in just a couple of seconds. 0:16:50.240 --> 0:16:54.920 It's crazy how fast we're talking. Now, what if see 0:16:55.080 --> 0:16:56.520 you still have a slow point, You still have a 0:16:56.520 --> 0:16:59.440 bottleneck at either end of that cable, right because you 0:16:59.480 --> 0:17:03.000 still have to convert the electricity to light and on 0:17:03.040 --> 0:17:06.040 the other end, convert that light back into electricity so 0:17:06.119 --> 0:17:09.560 that it can communicate with our devices. Because our devices 0:17:10.000 --> 0:17:14.200 rely on electronic transistors. But if we could change that, 0:17:14.680 --> 0:17:19.919 if we could make the fundamental components of our devices 0:17:20.160 --> 0:17:24.720 communicate through light rather than through electricity, everything we had 0:17:24.760 --> 0:17:26.719 would have that kind of throughput. It would be kind 0:17:26.760 --> 0:17:29.679 of like having a universal currency instead of having to 0:17:29.680 --> 0:17:32.280 go to the exchanger. Yeah, it would just mean that 0:17:32.359 --> 0:17:34.880 everything would be able to move at this incredible rate 0:17:34.920 --> 0:17:38.879 with these incredible options. But again you'd have to figure out, all, right, 0:17:38.960 --> 0:17:41.720 how do we achieve that, how do we build these 0:17:41.760 --> 0:17:45.160 fundamental blocks that we've depended upon. I mean, the development 0:17:45.160 --> 0:17:47.879 of the transistor dates back to the nineteen fifties, and 0:17:47.920 --> 0:17:51.520 then we've seen the transistor improve over time since then. 0:17:52.000 --> 0:17:55.280 So we would essentially have to reinvent the transistor. And people, 0:17:55.440 --> 0:17:58.600 by the way, have done that with photonics instead of 0:17:58.600 --> 0:18:02.480 with electricity. Okay, so I'm trying to imagine the photonic 0:18:02.520 --> 0:18:05.119 computer or the future. I'd assume you'd still base it 0:18:05.200 --> 0:18:08.800 on some kind of binary and coding of data, right, 0:18:08.960 --> 0:18:13.280 So unless you were going to completely change computer science 0:18:13.320 --> 0:18:16.119 and information theory. Unless you were to go to a 0:18:16.119 --> 0:18:18.959 point where you say, all right, well let's abandon boolean logic, 0:18:19.080 --> 0:18:24.280 let's abandon the basis for computers and create something entirely new. 0:18:25.040 --> 0:18:28.600 Then what the most logical approach is to say, how 0:18:28.640 --> 0:18:34.320 can we achieve a boolean logic based computer system but 0:18:34.520 --> 0:18:38.600 use photonic equipment rather than electronic equipment. That sounds kind 0:18:38.600 --> 0:18:41.080 of fancy, but what that really really boils down to 0:18:41.320 --> 0:18:44.800 is inside your computer, everything is a matter of a 0:18:44.960 --> 0:18:48.680 long series of on and off switches. Yes, those transistors, 0:18:48.720 --> 0:18:51.960 and those transistors are what we're essentially talking about as 0:18:52.040 --> 0:18:55.239 on off switches. Their gates that allow the flow of 0:18:55.440 --> 0:18:59.199 electrons in an electronic based system, but it would be 0:18:59.280 --> 0:19:02.679 photons in an optical based system. It would be a 0:19:02.800 --> 0:19:05.679 gate that would either allow photons to pass through or 0:19:05.760 --> 0:19:08.159 would prevent them in some way in order for it 0:19:08.200 --> 0:19:12.159 to still represent that one or zero, that bit. That 0:19:12.320 --> 0:19:16.280 is the basis of information theory in the modern computing age. 0:19:16.640 --> 0:19:20.399 So you would have to rebuild those basic components you 0:19:20.480 --> 0:19:25.000 find in computers, things like transistors, which if you arrange 0:19:25.040 --> 0:19:28.320 transistors in a proper way, you can create logic gates. 0:19:28.440 --> 0:19:32.000 This is like the and or nor If you ever 0:19:32.040 --> 0:19:35.040 study any basic logic, these are the basic functions that 0:19:35.119 --> 0:19:38.080 allow you to make statements and then evaluate those statements 0:19:38.280 --> 0:19:41.359 to get to the proper conclusions. That's the basis of 0:19:41.400 --> 0:19:44.520 computing today. That's exactly how computers work. So you would 0:19:44.560 --> 0:19:46.960 have to be able to make those logic gates using 0:19:47.320 --> 0:19:51.640 these photonic elements. Uh, and so right now, the state 0:19:51.720 --> 0:19:53.960 of the art as it stands today is all about 0:19:54.280 --> 0:19:58.560 trying to find new ways to create those those fundamental 0:19:58.600 --> 0:20:03.920 components that will make up an optical computer and be useful. 0:20:04.160 --> 0:20:07.480 Like we we've seen several several developments that are really interesting, 0:20:07.920 --> 0:20:11.040 but they are mostly proof of concept things that wouldn't 0:20:11.040 --> 0:20:13.320 be practical. I guess one of the big issues has 0:20:13.320 --> 0:20:16.480 got to be miniaturization. That would drinking it down. That's 0:20:16.520 --> 0:20:20.200 exactly right, Yeah, getting these components to be really, really tiny. 0:20:20.200 --> 0:20:22.960 I mean you're talking about like when you're talking about 0:20:22.960 --> 0:20:26.600 a microprocessor. Just think about those tiny computer chips you've seen, 0:20:27.200 --> 0:20:30.800 and keep in mind that the components, the individual components 0:20:30.840 --> 0:20:35.760 on that chip are at forty five nanometers or even smaller. 0:20:35.880 --> 0:20:38.640 And a nanometer is a billionth of a meter. So 0:20:38.840 --> 0:20:43.320 you're talking about billions of components on a single microprocessor. 0:20:43.880 --> 0:20:46.280 Now you have to create something that can emit light 0:20:47.040 --> 0:20:51.920 and sense light, and and be able to to essentially 0:20:52.000 --> 0:20:55.840 do that without becoming an enormous machine. So, in other words, 0:20:56.520 --> 0:20:58.639 a lot of the work that's done in labs, you 0:20:58.720 --> 0:21:02.160 might be able to create system that is a good 0:21:02.160 --> 0:21:04.600 proof of concept. But what if you if you made 0:21:04.600 --> 0:21:07.840 an entire computer that was of equivalent power to one 0:21:07.880 --> 0:21:11.840 of today's PCs using that technology, it would go back 0:21:11.840 --> 0:21:14.280 to the old days where your basic computer was the 0:21:14.320 --> 0:21:18.639 size of a building. So that would mean that it 0:21:18.640 --> 0:21:22.200 wouldn't be terribly practical on your next business trip. So 0:21:22.480 --> 0:21:26.760 who's actually built like a photonic transistor so far? Well, 0:21:27.200 --> 0:21:29.680 early works started in the nineteen eighties, so there have 0:21:29.760 --> 0:21:31.159 been a lot of people who have worked on it 0:21:31.200 --> 0:21:33.920 in various ways. But one of the more recent interesting 0:21:34.080 --> 0:21:37.440 stories was about some researchers that they might t back 0:21:37.440 --> 0:21:39.800 in July of two thousand and thirteen, they built an 0:21:39.800 --> 0:21:43.600 all optical transistor that could switch between zero and one 0:21:43.800 --> 0:21:46.800 using a single photon. So we're not talking about a 0:21:46.840 --> 0:21:48.680 beam of light, we're not talking about a laser beam. 0:21:48.680 --> 0:21:52.480 We're talking about one photon. Remember that a photon represents 0:21:52.520 --> 0:21:55.720 a particle of light, but that light connect both as 0:21:55.720 --> 0:21:59.040 a particle and a wave, which means it has some 0:21:59.200 --> 0:22:07.120 quantum uh features, some features we're talking about stuff when 0:22:07.160 --> 0:22:10.000 I say some quantum features, essentially, be prepared to have 0:22:10.119 --> 0:22:13.399 your mind bent. Yeah. The best way to understand this 0:22:13.440 --> 0:22:15.639 if you're not familiar with this concept is just to 0:22:15.760 --> 0:22:17.680 like go on the internet and look up the double 0:22:17.720 --> 0:22:22.120 slit experiment right where you know the is that's certainly 0:22:22.200 --> 0:22:26.359 a mind bending experiment where you know things that you know, 0:22:26.400 --> 0:22:28.600 you would think, you know, you would think it should 0:22:28.920 --> 0:22:31.080 behave a certain way and always behave that way, and 0:22:31.119 --> 0:22:33.719 it turns out that's that's not right. So in this 0:22:33.760 --> 0:22:38.240 particular implementation, what they had was to a pair of mirrors, 0:22:38.440 --> 0:22:42.720 okay uh, and the mirrors could perform in two different ways. 0:22:42.880 --> 0:22:45.320 When it was in the on position, let's say that's 0:22:45.359 --> 0:22:49.440 that represents a one in bit language, a photon could 0:22:49.440 --> 0:22:52.879 pass straight through both mirrors the pair of mirrors that 0:22:52.920 --> 0:22:56.440 are placed very closely together, extremely closely, as I'll mention 0:22:56.480 --> 0:22:59.760 in a second, when it's in the off position, only 0:22:59.800 --> 0:23:02.160 to any percent of the light would actually pass through 0:23:02.200 --> 0:23:05.160 the mirrors. In other words, the photon would get bounced back. 0:23:05.200 --> 0:23:08.119 A single photon would not make it through. And here's 0:23:08.160 --> 0:23:11.439 the reason why this works. Light does behave both as 0:23:11.480 --> 0:23:14.000 a particle on a wave. Now, if it acted only 0:23:14.040 --> 0:23:17.600 as a particle, it wouldn't matter whether the switch was 0:23:17.640 --> 0:23:20.119 on or off. It would encounter that first mirror and 0:23:20.200 --> 0:23:22.800 bounce off. It would be like you know, you running 0:23:22.800 --> 0:23:25.280 out a wall. You're not You're never not going to 0:23:25.400 --> 0:23:28.200 bounce off that wall unless the walls made of something 0:23:28.240 --> 0:23:30.760 really weak and you crash through cool a man style. 0:23:30.840 --> 0:23:34.400 But let's say let's say it's a cinder block solid wall. 0:23:34.520 --> 0:23:37.680 You would kind of splat slash bounce off of it. 0:23:37.680 --> 0:23:41.880 It would not be pleasant. However, light also acts as 0:23:41.880 --> 0:23:44.720 a wave, and so if you were to look at 0:23:45.200 --> 0:23:48.919 a wavelength of light, the electromagnetic field of a single 0:23:48.960 --> 0:23:52.320