WEBVTT - Thought Experiments and Philosophical Problems in Tech 0:00:04.400 --> 0:00:12.320 Welcome to Tech Stuff, a production from iHeartRadio. Hey there, 0:00:12.360 --> 0:00:15.360 and welcome to tech Stuff. I'm your host, Jonathan Strickland. 0:00:15.400 --> 0:00:18.720 I'm an executive producer with iHeartRadio, and how the tech 0:00:18.840 --> 0:00:22.079 are you? You know, often when we talk about tech, 0:00:22.360 --> 0:00:28.960 we reference various thought experiments, hypothetical situations, philosophical problems, and 0:00:29.360 --> 0:00:32.479 game theory. And this can get a little bit confusing 0:00:32.520 --> 0:00:35.199 if you've never actually studied any of those things, and 0:00:35.240 --> 0:00:39.040 people are just kind of off handedly spouting off terms. 0:00:39.040 --> 0:00:42.120 So today I thought i'd cover a small handful of 0:00:42.159 --> 0:00:46.040 them as a sort of foundation for future discussions. Keep 0:00:46.120 --> 0:00:50.520 in mind, there are tons of these, and I'm only 0:00:51.000 --> 0:00:54.840 covering like the teeniest, tiniest number of them. Some of 0:00:54.840 --> 0:00:58.360 these I have talked about extensively another episode, so I'll 0:00:58.400 --> 0:01:01.400 try to go a little light with them on this episode. 0:01:01.520 --> 0:01:03.520 But some of them are brand new to me, or 0:01:03.560 --> 0:01:05.520 I had only heard the name of them but had 0:01:05.560 --> 0:01:10.880 never actually researched the actual scenario or thought experiment. Now, 0:01:10.880 --> 0:01:14.520 these thought experiments in general, not the ones specifically we're 0:01:14.560 --> 0:01:18.640 talking about today, but the practice of thought experiments date 0:01:18.680 --> 0:01:22.000 back quite a ways, at the very least to ancient Greece, 0:01:22.160 --> 0:01:27.000 because we have records of them, so back then they 0:01:27.000 --> 0:01:32.000 were used to conceptualize complex mathematical problems and to give 0:01:32.040 --> 0:01:35.119 people a chance to consider consequences outside of a real 0:01:35.160 --> 0:01:39.920 world situation. But first up, I thought we would talk 0:01:40.120 --> 0:01:44.039 a little bit about game theory because we actually saw 0:01:44.080 --> 0:01:48.080 a real world version of game theory play out just 0:01:48.160 --> 0:01:51.240 a couple of weeks ago. And arguably you could say 0:01:51.240 --> 0:01:55.080 this is only tangentially related to technology, but it did 0:01:55.200 --> 0:01:58.280 have and continues to have a massive impact on tech. 0:01:59.000 --> 0:02:03.240 So the game theory situation that we're going to really 0:02:03.280 --> 0:02:07.560 talk about is known as the prisoner's dilemma. Folks were 0:02:07.600 --> 0:02:11.560 referencing this one in the wake of the Silicon Valley 0:02:11.600 --> 0:02:14.440 bank collapse, which is why I say this is only 0:02:14.520 --> 0:02:17.960 tangentially related to tech, because it's really about a run 0:02:18.040 --> 0:02:21.640 on the bank, and that bank just happened to be 0:02:21.680 --> 0:02:25.160 a really important bank to the tech industry. But the 0:02:25.240 --> 0:02:28.880 dilemma has its roots in the mid twentieth century, and 0:02:28.960 --> 0:02:32.680 the basic version goes something like this. A pair of 0:02:32.720 --> 0:02:37.080 suspected criminals are caught by police. They become the prisoners, 0:02:37.400 --> 0:02:41.040 and the police plan to interrogate each of the prisoners separately. 0:02:41.639 --> 0:02:46.080 The prison sentence for these criminals if they are found guilty, 0:02:46.440 --> 0:02:51.560 which is just taken as as the most likely outcome 0:02:51.919 --> 0:02:54.280 would be ten years. So if they are convicted, they 0:02:54.320 --> 0:02:58.280 get ten years in prison. However, each of them is 0:02:58.320 --> 0:03:03.600 told separately that there are different possible outcomes depending upon 0:03:03.639 --> 0:03:08.360 their cooperation or lack thereof with the police. They just 0:03:08.400 --> 0:03:11.520 aren't allowed to talk to each other, so the two 0:03:11.880 --> 0:03:14.919 prisoners have no way to communicate with one another. They 0:03:14.919 --> 0:03:18.919 have to make up their minds individually, and the four 0:03:19.040 --> 0:03:24.679 possible outcomes are this, neither suspect talks, neither prisoner confesses, 0:03:25.160 --> 0:03:28.400 and in that they each get only two years in 0:03:28.520 --> 0:03:31.480 prison because the cops won't have enough evidence to put 0:03:31.520 --> 0:03:35.200 them away for the more serious crime. So they'll have 0:03:35.240 --> 0:03:37.040 to go to prison, but i'll just be two years, 0:03:37.120 --> 0:03:41.160 not the full ten. However, if one of them stays 0:03:41.160 --> 0:03:45.320 silent but the other one confesses, well, then the one 0:03:45.320 --> 0:03:49.040 who confesses gets to go free and the silent prisoner 0:03:49.440 --> 0:03:52.720 has to serve the full ten years of the sentence. 0:03:53.480 --> 0:03:57.600 If both prisoners confess, then each of them will get 0:03:57.880 --> 0:04:02.200 five years. So you get your four possible outcomes. Suspects 0:04:02.240 --> 0:04:04.560 A and B keep the trap shut and each of 0:04:04.560 --> 0:04:08.520 them serve two years. Suspect A talks but Bee keeps quiet, 0:04:08.800 --> 0:04:11.800 which means A goes off scott Free and B goes 0:04:11.840 --> 0:04:16.080 to the pokey for ten years. Suspect A holds their tongue, 0:04:16.120 --> 0:04:19.320 but Suspect B sings like a canary, and that time 0:04:19.480 --> 0:04:22.320 Suspect B strolls out to freedom and Suspect A rots 0:04:22.320 --> 0:04:25.480 away in prison, or they both go blabbing and they 0:04:25.520 --> 0:04:30.320 both end up serving five years. Now, collectively, the most 0:04:30.400 --> 0:04:33.960 beneficial outcome is to serve only two years by not 0:04:34.120 --> 0:04:38.200 talking at all, and then the next best outcome for 0:04:38.360 --> 0:04:42.159 both individuals collectively would be that they both talk and 0:04:42.200 --> 0:04:44.440 they both have to serve five years, but not the 0:04:44.480 --> 0:04:50.360 full ten. However, individually, if we're not talking about collectively, individually, 0:04:50.400 --> 0:04:53.480 the best outcome is you talk and hope the other 0:04:53.520 --> 0:04:55.839 one doesn't. That way, you can put on your dance 0:04:55.880 --> 0:04:59.120 and shoes, just strut on out of the building and 0:04:59.160 --> 0:05:03.919 the other one's days behind. But worst case scenario, you talk, 0:05:04.080 --> 0:05:07.039 the other person also talks, and you just serve half 0:05:07.080 --> 0:05:11.760 the full sentence each. Actually, worst case scenario is you 0:05:11.880 --> 0:05:14.800 decide not to talk and the other person does talk, 0:05:14.960 --> 0:05:17.960 and then you're looking at ten years in jail. Well, 0:05:17.960 --> 0:05:21.520 when it came to Silicon Valley Bank, the real world 0:05:21.560 --> 0:05:25.719 scenario went like this. If everyone had remained chill, their 0:05:25.760 --> 0:05:31.040 money would have been safe. SBB had over extended its 0:05:31.160 --> 0:05:35.479 investment and government backed securities, which would take years to mature. 0:05:36.200 --> 0:05:39.680 And this was because SVB wasn't issuing as many loans 0:05:39.760 --> 0:05:44.119 once interest rates had gone up significantly. Venture capitalists weren't 0:05:44.160 --> 0:05:47.640 seeking loans so much. Plus they were already flush with cash, 0:05:47.680 --> 0:05:50.599 and loans are the way that banks make money for 0:05:50.640 --> 0:05:53.200 the most part. So instead the banks were investing in 0:05:54.480 --> 0:05:59.400 longer term investments that would have a modest payout, but 0:05:59.520 --> 0:06:03.359 it would have a payout once the investments matured. But 0:06:04.560 --> 0:06:07.880 that would mean that everyone If they just you know, 0:06:08.279 --> 0:06:11.839 cooled their jets and kept their money in place, things 0:06:11.880 --> 0:06:15.560 would probably have been fine. SBB would have likely survived, 0:06:15.880 --> 0:06:20.679 but instead the prisoners the customers of SVB in this case, 0:06:21.080 --> 0:06:24.160 chose to pull their money out because the thinking went 0:06:24.279 --> 0:06:27.719 something like this, if everyone else takes out their money 0:06:27.760 --> 0:06:31.159 and I don't, and then then SBB could shut down, 0:06:31.200 --> 0:06:34.760 and then my money will be stuck, it'll stop existing, 0:06:34.880 --> 0:06:36.600 or I won't be able to get to it, and 0:06:36.640 --> 0:06:39.400 I need my money. It's what lets me buy all 0:06:39.440 --> 0:06:43.920 that stuff like private jets and politicians. So I'm gonna 0:06:43.960 --> 0:06:46.320 go get my money out before i lose out on 0:06:46.360 --> 0:06:49.280 that option. The problem was is that some very big 0:06:49.320 --> 0:06:52.039 players who had a whole lot of money in SBB 0:06:52.400 --> 0:06:55.440 did this, including the heads of venture capitalist groups, who 0:06:55.520 --> 0:06:58.640 then urged all of their clients to do the same, 0:06:59.240 --> 0:07:01.320 and so there was a run on the bank and 0:07:01.560 --> 0:07:04.760 SVB could not cover all the withdrawals without selling off 0:07:04.800 --> 0:07:08.400 assets at a huge loss, and that put SVB in 0:07:08.480 --> 0:07:12.120 a very precarious situation, in fact, precarious enough that the 0:07:12.200 --> 0:07:14.560 US government had to swoop in and take over the 0:07:14.600 --> 0:07:17.640 bank and guarantee all the customers that they would still 0:07:17.680 --> 0:07:21.560 be able to access their money. So enough prisoners took 0:07:21.640 --> 0:07:25.320 the sure thing principle and they screwed over everybody else 0:07:25.320 --> 0:07:28.560 in the process, which not a big surprise because generally 0:07:28.600 --> 0:07:32.680 there's an agreement that taking the tactic of confessing makes 0:07:32.760 --> 0:07:36.120 the most sense from a game theory perspective, and that 0:07:36.240 --> 0:07:39.680 loyalty has no place in the game, that if you 0:07:39.720 --> 0:07:42.480 are loyal, the best you can hope for is two 0:07:42.600 --> 0:07:45.840 years in prison and the worst is ten. So it 0:07:45.920 --> 0:07:48.960 makes more sense to confess and either screw over the 0:07:49.000 --> 0:07:53.280 other person or you both get screwed. So that's the 0:07:53.400 --> 0:07:56.720 thinking behind the prisoner's dilemma, and like I said, we 0:07:56.840 --> 0:07:59.400 kind of saw it play out with the collapse of 0:07:59.400 --> 0:08:02.920 the Silicon Valley Bank. Now, one of my favorite thought 0:08:02.960 --> 0:08:07.080 experiments that has some connection to the tech industry is 0:08:07.200 --> 0:08:10.960 the Ship of Theseus, which dates back to at least 0:08:10.960 --> 0:08:13.600 a century of around five hundred to four hundred BC. 0:08:14.560 --> 0:08:17.280 And the Ship of Theseus idea goes like this, So 0:08:17.360 --> 0:08:20.160 you had the Greek hero Theseus. He had a ship, 0:08:20.760 --> 0:08:25.240 and he ended up docking that ship, and people were 0:08:25.280 --> 0:08:29.120 preserving the ship for his eventual return. And long after 0:08:29.400 --> 0:08:33.000 the hero himself had faded away, the ship remained preserved. 0:08:33.280 --> 0:08:36.120 But of course, over time, pieces of the ship need 0:08:36.160 --> 0:08:39.400 to be replaced. You know, maybe the sails rip and tear, 0:08:39.520 --> 0:08:41.360 so you need to put new sails on the ship. 0:08:42.080 --> 0:08:44.560 Maybe rot sets into part of the deck, so you 0:08:44.559 --> 0:08:46.480 have to rip that out and replace it with new 0:08:46.480 --> 0:08:50.720 planking and so on. And eventually, over the course of time, 0:08:50.880 --> 0:08:54.920 maybe it's decades, maybe it's even centuries, you gradually replace 0:08:55.240 --> 0:08:58.680 every single piece of the ship, so you ultimately arrive 0:08:58.720 --> 0:09:02.120 at a point where no element in the ship of 0:09:02.200 --> 0:09:08.120 Theseus is the original component. Theseus himself never touched anything 0:09:08.120 --> 0:09:11.160 on the ship at this point, So would you still 0:09:11.200 --> 0:09:15.120 call it the same ship? If not, when did it 0:09:15.160 --> 0:09:19.880 officially stop being the ship of theseus? Because obviously, if 0:09:19.880 --> 0:09:22.360 you took possession of the ship right after Theseus got 0:09:22.360 --> 0:09:26.800 tossed off a cliff by lack of medes, and then 0:09:26.920 --> 0:09:29.560 on your first day you had to replace a sale, 0:09:29.720 --> 0:09:31.760 you would still call it the ship of Theseus? Right, 0:09:32.000 --> 0:09:34.920 is just one sale that you replaced. The ship itself 0:09:34.960 --> 0:09:38.040 is still the same, And you know a single sale 0:09:38.080 --> 0:09:41.360 does not change the identity of the overall ship. But 0:09:41.600 --> 0:09:44.240 is there a point where that does happen where the 0:09:44.280 --> 0:09:49.720 ship's identity changes? In tech, one way this thought experiment 0:09:49.800 --> 0:09:54.160 can manifest is when a company undergoes digitalization. Companies that 0:09:54.200 --> 0:09:57.200 have been around for decades have various systems and processes 0:09:57.240 --> 0:10:02.319 in place that predate digital realization. And I always struggle 0:10:02.360 --> 0:10:04.320 over that word, so you're gonna hear me stumble a lot. 0:10:04.360 --> 0:10:08.240 But as a result you to modernize. The leaders of 0:10:08.280 --> 0:10:11.520 these companies have to decide when, if ever, to convert 0:10:11.600 --> 0:10:15.520 old processes into new ones in order to stay current 0:10:15.559 --> 0:10:19.680 and to avoid problems with outdated legacy components. Whenever I 0:10:19.720 --> 0:10:22.200 look at really big companies that have been around for 0:10:22.240 --> 0:10:25.720 like a century, I am often left wondering how they 0:10:25.800 --> 0:10:30.040 handled these transitions, or even if they tried to, because 0:10:30.040 --> 0:10:33.320 these legacy systems are often crucial to the business. The 0:10:33.360 --> 0:10:38.280 business grew around these systems, and so changing the system 0:10:38.480 --> 0:10:41.400 is hard because you have so much other stuff that 0:10:41.480 --> 0:10:45.240 grew up around it and depends upon it, and the 0:10:45.400 --> 0:10:49.120 hardware becomes outdated. It can even get difficult or sometimes 0:10:49.160 --> 0:10:54.760 impossible to maintain or replace old equipment simply because no 0:10:54.800 --> 0:10:58.640 one makes that thing anymore. You know that particular computer 0:10:58.679 --> 0:11:02.520 system may not even be available at any rate, so 0:11:02.559 --> 0:11:04.920 you have to figure out a different way to do things. 0:11:05.840 --> 0:11:09.800 Digitalization makes it easier to track progress and identify bobblenecks 0:11:09.800 --> 0:11:12.960 and such, but it also might mean having to take 0:11:13.040 --> 0:11:17.200 a slightly different approach to try and get a similar 0:11:17.240 --> 0:11:21.040 result as your legacy systems. So the new version isn't 0:11:21.520 --> 0:11:24.280 perfectly recreating the old one. It's just trying to reach 0:11:24.320 --> 0:11:28.760 the same conclusion. But in the process new things might 0:11:28.800 --> 0:11:33.880 pop up, unexpected complications or diversions, and thus he might 0:11:33.920 --> 0:11:36.840 be left wondering if the IBM of today is really 0:11:36.880 --> 0:11:39.720 the same company as the IBM that was founded in 0:11:39.840 --> 0:11:43.600 nineteen eleven. Actually nineteen eleven was the founding of the 0:11:43.600 --> 0:11:49.560 Computing Tabulating Recording Company, which was the precursor to IBM, 0:11:49.640 --> 0:11:52.520 But you get my point. There are some thought experiments 0:11:52.559 --> 0:11:55.840 that are specific to computing problems. One of those is 0:11:55.880 --> 0:12:00.440 called the two Generals problems, which focuses on the issues 0:12:00.440 --> 0:12:04.959 you face when you try to establish communications across unreliable connections. 0:12:05.240 --> 0:12:07.640 So when networks this becomes a big deal. Right. The 0:12:07.720 --> 0:12:11.000 basis of Internet connections largely falls to figuring out the 0:12:11.040 --> 0:12:13.960 most reliable way to deliver information to another system that's 0:12:13.960 --> 0:12:18.000 connected to that network. But the basic two General's problems 0:12:18.040 --> 0:12:20.880 goes something like this. You have an enemy that's in 0:12:20.920 --> 0:12:25.959 control of a centralized valley, and you have two generals. 0:12:26.040 --> 0:12:29.360 Each of them are in charge of their own army. 0:12:29.440 --> 0:12:34.000 Each army is in a valley that neighbors this central valley, 0:12:34.040 --> 0:12:36.800 So essentially you're flanking the enemy. You've got one army 0:12:36.840 --> 0:12:38.560 on the left, one army on the right. They're both 0:12:38.559 --> 0:12:40.800 in their own valleys, and the enemy is in the 0:12:40.880 --> 0:12:43.880 valley in between the two so your goal is to 0:12:44.000 --> 0:12:46.800 establish a time for both generals to attack the enemy 0:12:47.120 --> 0:12:50.080 in the middle at the same time, because the enemy 0:12:50.160 --> 0:12:53.120 is too well entrenched and too strong for either army 0:12:53.160 --> 0:12:55.720 to defeat it on its own. If only one of 0:12:55.720 --> 0:12:58.200 your army's attacks, it's going to get wiped out. So 0:12:58.320 --> 0:13:00.560 really the only hope for your victory is to have 0:13:00.600 --> 0:13:04.560 a coordinated attack. Now complicating things is that the only 0:13:04.600 --> 0:13:07.640 way the two generals are able to communicate with one 0:13:07.679 --> 0:13:11.800 another is to send a messenger across enemy lands to 0:13:11.840 --> 0:13:14.880 reach the other general, and any messenger runs the risk 0:13:15.040 --> 0:13:18.400 of being caught in the process. So let's say that 0:13:18.559 --> 0:13:21.160 you determine before they set out. The General A is 0:13:21.200 --> 0:13:24.960 in charge of establishing an attack time, and so General 0:13:24.960 --> 0:13:28.679 A writes, we attack at dawn in three days and 0:13:28.760 --> 0:13:32.040 sends a messenger out to travel to General B. Well, 0:13:32.160 --> 0:13:34.520 General A doesn't know if the messenger makes it to 0:13:34.600 --> 0:13:37.560 General B. So in three days at dawn, there's a 0:13:37.640 --> 0:13:40.559 risk the General B never got the message, and if 0:13:40.559 --> 0:13:44.240 General A attacks, it's going to mean a loss because 0:13:44.600 --> 0:13:49.640 B won't be participating in a simultaneous attack. But what 0:13:49.760 --> 0:13:53.240 if General B did get the message and then sends 0:13:53.320 --> 0:13:57.280 a confirmation back message received, we attack in three days 0:13:57.280 --> 0:14:00.600 at dawn. Well, that messenger might end up being intercepted. 0:14:00.960 --> 0:14:03.480 So now in three days time, General B isn't sure 0:14:03.480 --> 0:14:07.320 if General A knows that everything is good to go 0:14:07.520 --> 0:14:11.240 as scheduled. So maybe General B hesitates to avoid defeat 0:14:11.679 --> 0:14:14.240 because General B doesn't know if generally is aware of this. 0:14:15.200 --> 0:14:17.760 Of course, General AA, receiving the reply could try to 0:14:17.800 --> 0:14:21.560 send their own message back to General B. But maybe 0:14:21.600 --> 0:14:23.840 that messenger gets cut along the way, and this goes 0:14:23.880 --> 0:14:27.360 back and forth, and the challenges You cannot be confident 0:14:27.440 --> 0:14:30.520 that any one message made it to the correct destination, 0:14:30.680 --> 0:14:34.720 So how you design a communication system where you're reasonably 0:14:34.720 --> 0:14:38.960 certain that messages are going through becomes a challenge. The 0:14:39.040 --> 0:14:42.760 thought experiment shows that uncertainty in communication systems aren't a 0:14:42.800 --> 0:14:46.240 problem that can necessarily be outright solved, but it perhaps 0:14:46.280 --> 0:14:48.680 can be mitigated to a point where all sides are 0:14:48.800 --> 0:14:52.920 comfortably communicating. Okay, we've got more to say about thought 0:14:52.960 --> 0:15:05.840 experiments and philosophy, but let's take a quick break. All right, 0:15:06.160 --> 0:15:10.240 we're back. Let's talk a little bit more about computing 0:15:10.360 --> 0:15:14.600 thought experiments. We just mentioned that it's hard to be 0:15:14.800 --> 0:15:20.640 certain about communications in uncertain situations. You know, it's you 0:15:20.640 --> 0:15:24.600 can do the best you can and limit the chances 0:15:24.640 --> 0:15:29.480 for messaging to fall through, but you can't ensure that 0:15:29.560 --> 0:15:33.160 it is perfect. That's the purpose of the two generals 0:15:33.160 --> 0:15:35.280 thought experiment. But let's talk about a different one. Let's 0:15:35.320 --> 0:15:39.840 talk about philosophers and PISKEETI. Seriously, there is a thought 0:15:39.840 --> 0:15:45.360 experiment called the dining philosopher's problem. This one is more 0:15:45.400 --> 0:15:50.560 about the sharing of computational resources and avoiding a deadlock situation, 0:15:51.240 --> 0:15:56.280 or a situation in which one computational process is hogging 0:15:56.320 --> 0:15:59.440 all the resources and all the other processes that need 0:15:59.480 --> 0:16:02.960 to run that same machine can't. So to understand this, 0:16:03.720 --> 0:16:06.440 let's first recall that back in the old days, before 0:16:06.440 --> 0:16:10.960 we got to microcomputers and many computers, a computer system 0:16:11.000 --> 0:16:16.120 generally consisted of a big, centralized mainframe computer that you 0:16:16.120 --> 0:16:21.160 would access through a data terminal, a dumb terminal, which 0:16:21.520 --> 0:16:24.960 could include very basic input devices like a keyboard and 0:16:25.080 --> 0:16:28.760 a very basic output device like a monitor. But the 0:16:28.840 --> 0:16:33.200 dumb terminal wouldn't have any computational ability itself, like it 0:16:33.240 --> 0:16:36.320 would look kind of like a desktop computer, but instead 0:16:36.760 --> 0:16:40.040 it's literally just a monitor and a keyboard. It's connecting 0:16:40.600 --> 0:16:44.080 to this centralized computer, which likely has lots of other 0:16:44.240 --> 0:16:48.200 dumb terminals connected to it, with other people also accessing 0:16:48.200 --> 0:16:51.240 the centralized computer. So what you really have is a 0:16:51.280 --> 0:16:56.640 shared computational resource that is distributed across all these different 0:16:56.720 --> 0:17:01.760 dumb terminals. Computers typically handled all this by dealing with 0:17:01.840 --> 0:17:05.600 each terminal one at a time, in sequence, but at 0:17:05.640 --> 0:17:09.560 really fast speeds, so it felt that it was pretty 0:17:09.600 --> 0:17:12.399 responsive and that you were doing everything more or less 0:17:12.440 --> 0:17:16.399 in real time, and it was called time sharing. Every 0:17:16.400 --> 0:17:20.800 person at a data terminal was sharing time with this computer. 0:17:20.920 --> 0:17:24.400 Time was really precious with these things too. But how 0:17:24.440 --> 0:17:27.840 do you make sure that all the different processes slash 0:17:27.960 --> 0:17:30.919 terminals are able to access the computer fairly? How do 0:17:30.920 --> 0:17:34.320 you avoid situations where all the demands are coming in 0:17:34.359 --> 0:17:37.680 at a point that effectively locks the entire system where 0:17:37.680 --> 0:17:41.320 it can't do anything. This brings us to the dining 0:17:41.440 --> 0:17:46.600 philosopher's problem. So imagine we've got ourselves a big round 0:17:46.600 --> 0:17:51.240 table and we have placed five plates around this table. 0:17:51.880 --> 0:17:56.239 There's a chair at each plate, and in between the 0:17:56.320 --> 0:17:59.639 plates there is a single fork, so you have plate 0:17:59.720 --> 0:18:03.719 for plate, fork plate fork plate, etc. So five plates, 0:18:03.760 --> 0:18:10.080 five forks, five chairs. So far, so good, But the 0:18:10.160 --> 0:18:14.280 problem is that the philosophers who are coming to dine 0:18:14.600 --> 0:18:18.680 are there to eat spaghette. And the big old heap 0:18:18.760 --> 0:18:22.720 and plate of spaghett is glorious, but the only way 0:18:22.720 --> 0:18:26.239 to eat it is to use two forks simultaneously. So 0:18:26.280 --> 0:18:28.720 you need a fork in each hand in order to 0:18:28.800 --> 0:18:33.600 be able to wind up enough spaghett to shove into 0:18:33.640 --> 0:18:37.800 your gob and you can eat your spaghetti. When you're 0:18:37.840 --> 0:18:40.920 not eating, you can think, because you're a philosopher, so 0:18:40.960 --> 0:18:43.480 you're either thinking or you're eating. That's all you're doing 0:18:43.480 --> 0:18:46.800 at this table. But obviously if you go without eating 0:18:46.840 --> 0:18:50.080 for too long, you'll starve yourself to death. Now, since 0:18:50.080 --> 0:18:52.359 you need two forks to eat, and there are only 0:18:52.480 --> 0:18:56.040 five forks at the table, if you grab the fork 0:18:56.119 --> 0:18:58.240 on your left and the fork on your right, it 0:18:58.320 --> 0:19:00.240 means that the people sitting to your left and to 0:19:00.280 --> 0:19:03.920 your right they can't eat right they have access at 0:19:04.000 --> 0:19:06.800 most to one fork. They don't have access to the 0:19:06.800 --> 0:19:08.520 second one, because those are the ones that are in 0:19:08.560 --> 0:19:12.240 your hands. Once you put the forks down, they become available, 0:19:12.560 --> 0:19:15.440 and then the people on either side can pick that 0:19:15.600 --> 0:19:18.639 fork up and potentially eat unless of course, the fork 0:19:18.800 --> 0:19:21.520 on their opposite side has already been taken by the 0:19:21.600 --> 0:19:24.919 other two people at the table, so they could be 0:19:24.960 --> 0:19:27.440 out of luck, and you have to figure out how 0:19:27.440 --> 0:19:30.840 to juggle this. Worse than that, though, Let's say that 0:19:30.920 --> 0:19:33.480 you've set a rule that whenever a fork is laid down, 0:19:33.760 --> 0:19:36.399 you pick it up immediately, so that you're always at 0:19:36.480 --> 0:19:40.840 least half ready to start eating. But everyone follows this rule, 0:19:41.200 --> 0:19:44.200 and at the very beginning of the meal, everybody reaches 0:19:44.200 --> 0:19:47.240 over to their right and picks up a fork. Well, 0:19:47.920 --> 0:19:51.919 now all five forks are in hand, five different hands 0:19:51.960 --> 0:19:54.480 in fact, which means no one can eat or think 0:19:54.800 --> 0:19:57.760 because they only have one fork in their hand. They 0:19:57.800 --> 0:20:00.639 need two forks to eat. If they set down the fork, 0:20:00.960 --> 0:20:04.040 then they're going to lose it, so they're holding onto it. 0:20:04.040 --> 0:20:07.840 It deadlocks the whole system. So how do you fix this? Well, 0:20:07.880 --> 0:20:11.280 they are actually different solutions to this problem, and they're 0:20:11.280 --> 0:20:13.959 all meant to try and avoid deadlock. And then there 0:20:13.960 --> 0:20:16.680 are other solutions that are meant to ensure fairness, because 0:20:16.720 --> 0:20:20.040 that's not a guarantee in this system. For example, you 0:20:20.119 --> 0:20:23.000 might set a rule that ends up assigning a number 0:20:23.200 --> 0:20:26.080 to each one of the forks and maybe the rule 0:20:26.240 --> 0:20:30.159 is that you can only grab the lower number that's 0:20:30.920 --> 0:20:34.199 in front of you first, and then you can grab 0:20:35.560 --> 0:20:40.080 whichever fork has a higher number, and everyone grabs their 0:20:40.119 --> 0:20:43.600 lowest number. But someone's going to be left without being 0:20:43.600 --> 0:20:47.080 able to do that because they will only be left 0:20:47.119 --> 0:20:49.960 with the number five fork. There is no lower number four. 0:20:50.119 --> 0:20:52.960 They cannot follow the rule because the number four has 0:20:52.960 --> 0:20:56.199 been grabbed by someone else. This allows one person to 0:20:56.240 --> 0:20:58.720 grab the number five fork because they've already grabbed the 0:20:58.800 --> 0:21:01.240 lower number, and they can eat. Then they can sit 0:21:01.280 --> 0:21:05.119 down their fork, and this can then continue with everybody 0:21:05.160 --> 0:21:07.760 getting a chance, assuming you have other rules in place 0:21:08.280 --> 0:21:11.199 to help guide things. Now that's just one approach, mind you, 0:21:11.240 --> 0:21:13.480 there are lots of others, Like there's one where there's 0:21:13.480 --> 0:21:18.000 an arbiter who is there to determine when each person 0:21:18.119 --> 0:21:20.399 is allowed to eat. They essentially are the ones given 0:21:20.480 --> 0:21:26.600 permission to grant the privilege of eating too specific people 0:21:26.640 --> 0:21:29.200 and to make sure that no one overheats like. That's 0:21:29.240 --> 0:21:32.320 another approach. So the point of the whole thought experiment 0:21:32.400 --> 0:21:36.240 is to get people considering the challenges using a limited 0:21:36.320 --> 0:21:39.440 number of resources for multiple entities in such a way 0:21:39.480 --> 0:21:42.440 that no one goes without for too long, and there's 0:21:42.480 --> 0:21:45.119 a means of managing things. It's meant to give computer 0:21:45.240 --> 0:21:49.040 scientists heads up on things they have to consider when 0:21:49.080 --> 0:21:53.719 they're designing their systems. So it's really thought experiments that's 0:21:53.720 --> 0:21:55.840 where they're really valuable. Is that it's before you started 0:21:55.840 --> 0:22:01.119 to build anything, right, you haven't dedicated asset and time 0:22:01.160 --> 0:22:04.399 and effort to building something. You're thinking it through first 0:22:04.400 --> 0:22:09.240 and saying, how do I avoid this perceived problem so 0:22:09.280 --> 0:22:11.560 that we don't actually encounter it in the wild and 0:22:11.560 --> 0:22:13.359 then have to figure out a solution. How can I 0:22:13.400 --> 0:22:17.040 solve it just by thinking about it? That is what 0:22:17.160 --> 0:22:20.720 these thought experiments give you the opportunity to do, assuming 0:22:20.720 --> 0:22:24.520 that the thought experiment is constructed properly, which is not 0:22:25.119 --> 0:22:27.600 always the case. There are thought experiments that later on 0:22:27.720 --> 0:22:32.840 people picked apart and said, this thought experiment is predicated 0:22:32.920 --> 0:22:37.960 upon assumptions that we can't be certain are true, and 0:22:38.000 --> 0:22:43.240 therefore you can't really use this thought experiment without acknowledging 0:22:43.280 --> 0:22:46.080 that it could just all be for nothing because the 0:22:46.119 --> 0:22:51.320 actual primases aren't proven. But then there are the various 0:22:51.359 --> 0:22:54.640 thought experiments and ethics problems that come into play when 0:22:54.640 --> 0:22:58.880 you start to talk about artificial intelligence. Now, I've said 0:22:58.920 --> 0:23:02.000 many times in this show, AI covers a huge amount 0:23:02.040 --> 0:23:06.159 of ground. It's a very I think AI is a 0:23:06.240 --> 0:23:10.840 dangerous term. Not dangerous in the sense of it potentially 0:23:10.880 --> 0:23:14.280 being harmful to humans, but rather it's such a huge 0:23:14.359 --> 0:23:18.840 discipline that it's very easy to be reductive when you're 0:23:18.840 --> 0:23:22.399 talking about AI, and to think that when you say AI, 0:23:22.480 --> 0:23:26.159 are just talking about machines, thinking as if they were people. 0:23:27.080 --> 0:23:32.920 That's one version of what AI could be. It's generally 0:23:32.960 --> 0:23:36.680 referred to as strong AI. But AI covers a lot 0:23:36.680 --> 0:23:40.400 of ground. It is a multidisciplinary technology, and it encompasses 0:23:41.320 --> 0:23:47.080 relatively constrained concepts like computer vision or language recognition, and 0:23:47.119 --> 0:23:49.280 then it ranges all the way up to big ideas 0:23:49.359 --> 0:23:53.080 like strong or general AI capable of processing information in 0:23:53.080 --> 0:23:56.320 a way that it is at least human like. Well, 0:23:57.640 --> 0:24:01.280 one of the elements, in fact, one that's closest to 0:24:01.359 --> 0:24:05.880 strong AI, that's in the thought experiment world, is the 0:24:05.880 --> 0:24:09.280 thought experiment of an artificial brain or artificial mind. What 0:24:09.400 --> 0:24:14.879 would it take to produce an artificial brain? So something 0:24:14.920 --> 0:24:18.600 that we have created that is capable of some form 0:24:18.680 --> 0:24:23.160 of thought, something that we would recognize as thought. So 0:24:23.200 --> 0:24:25.639 there's a question about whether or not it's even possible 0:24:25.800 --> 0:24:29.840 to create an actual artificial brain and what that could entail. 0:24:30.320 --> 0:24:33.119 Some argue that what it will take is just a 0:24:33.160 --> 0:24:38.200 sufficiently complex computer system that's emulating how our brains work, 0:24:38.760 --> 0:24:41.600 so like an artificial neural network. If we were able 0:24:41.640 --> 0:24:44.280 to build an artificial neural network that was big enough 0:24:44.840 --> 0:24:51.480 and fast enough on powerful enough computer systems, then potentially 0:24:51.520 --> 0:24:55.119 we would see the formation of an artificial brain. That's 0:24:55.160 --> 0:24:59.479 how that argument goes. And maybe we wouldn't even need 0:24:59.520 --> 0:25:05.960