WEBVTT - Gaming the System 0:00:00.160 --> 0:00:07.160 Brought to you by Toyota. Let's go places. Welcome to 0:00:07.360 --> 0:00:15.080 Forward Thinking, either everyone, and welcome to Forward Thinking, the 0:00:15.160 --> 0:00:18.600 podcast that looks at the future and says, these foolish 0:00:18.640 --> 0:00:22.120 games are tearing me apart. I'm Jonathan Strickland, I'm Laura, 0:00:22.600 --> 0:00:26.800 and I'm Joe McCormick. Hey, Joe, do you do you 0:00:26.840 --> 0:00:33.440 ever play games? Joe? No, In fact, I don't play games. 0:00:33.600 --> 0:00:41.839 Joe is extremely serious. Lauren, do you play games? Lauren? Uh? Sometimes, yes, yes. 0:00:42.000 --> 0:00:44.520 You know, this is a question that I don't ask computers. 0:00:44.560 --> 0:00:47.240 I get scared to ask computers if they play games 0:00:47.280 --> 0:00:51.080 because I was brought up in the eighties when the 0:00:51.080 --> 0:00:55.480 the amazing film War Games came out. So I gotta 0:00:55.720 --> 0:00:59.880 be honest, I've never seen it, actually me neither. Wow. Okay, guys, 0:01:00.200 --> 0:01:03.680 prepare yourselves because this movie is amazing. So what now 0:01:03.720 --> 0:01:06.080 We're just gonna watch this movie and the podcast booth 0:01:06.120 --> 0:01:07.720 and we don't have time for that. I'm just gonna 0:01:07.720 --> 0:01:10.920 give you. I'm gonna sum up. So War Games is 0:01:10.959 --> 0:01:15.679 specifically the story of a a uh a hacker. A 0:01:15.720 --> 0:01:20.440 young hacker played by Matthew Broderick, who has Fred Savage 0:01:20.520 --> 0:01:24.560 not Fred said this is pre Fred Savage. Um, before 0:01:24.600 --> 0:01:27.680 Fred Savage, we had Matthew Broderick, right, Yeah, Matthew Broderick 0:01:27.760 --> 0:01:32.840 was the proto Fred Savage was. So Matthew Broderick is 0:01:32.840 --> 0:01:35.000 playing a part of a of a hacker and he 0:01:35.440 --> 0:01:38.360 he likes these particular computer games, so he hacks into 0:01:38.400 --> 0:01:41.399 a system in order to get a chance to play 0:01:41.440 --> 0:01:45.319 some of these games that have not been released, and 0:01:45.400 --> 0:01:47.559 one of them, one of them is Tic Tac Toe, 0:01:47.680 --> 0:01:53.560 but another one is Global Thermonuclear War. And it turns 0:01:53.560 --> 0:01:56.800 out that the AI developed by the computer scientists who 0:01:56.800 --> 0:02:00.240 created the games isn't really really sophisticated, and you can 0:02:00.280 --> 0:02:06.880 play these simulated global war scenarios. But spoiler alert, the 0:02:06.960 --> 0:02:12.760 program gets installed into an actual military facility. It essentially 0:02:12.800 --> 0:02:16.680 infects a military facility like a virus and then starts 0:02:16.720 --> 0:02:20.800 to take over actual defense systems and prepares to launch 0:02:21.000 --> 0:02:25.480 a full attack on the then Soviet Union. So it 0:02:25.560 --> 0:02:29.240 becomes the role of Matthew Broderick's character, along with the 0:02:29.240 --> 0:02:34.160 original programmer, to figure out a way to convince the computer, hey, 0:02:34.360 --> 0:02:37.080 you don't want to do that because it would be 0:02:37.120 --> 0:02:40.520 bad for everybody and you don't need to win this game. 0:02:40.560 --> 0:02:43.440 Computer you can let it go. They actually teach that 0:02:43.440 --> 0:02:45.720 that there is no way to win. In fact, the 0:02:45.720 --> 0:02:49.280 computer says, strange game. The only way to win is 0:02:49.320 --> 0:02:52.920 not to play. And the way that the computer figures 0:02:52.960 --> 0:02:55.760 this out is by using an analogy. It actually starts first. 0:02:55.760 --> 0:03:00.520 It starts with the various simulations of nuclear war uh, 0:03:00.560 --> 0:03:03.280 and it runs the simulations over and over and over again. 0:03:03.360 --> 0:03:05.640 Hasn't launched anything. It's just simulating. And so you're on 0:03:05.680 --> 0:03:08.400 the big like Norad screen you see these things playing 0:03:08.400 --> 0:03:11.440 out this which is the tic Tac Toe. The reason 0:03:11.440 --> 0:03:14.400 it switches to tic tac Toe is because if you 0:03:14.440 --> 0:03:17.720 play Tic tac Toe perfectly on both sides, you will 0:03:17.760 --> 0:03:21.160 always end in a draw. And so the lesson is 0:03:21.480 --> 0:03:24.560 there is no way to win this game. Playing this 0:03:24.600 --> 0:03:27.960 game makes no sense because you cannot win it if 0:03:27.960 --> 0:03:31.880 you are doing everything correctly. So I just saved you 0:03:31.960 --> 0:03:34.800 the the time of actually watching the movie. But it 0:03:34.920 --> 0:03:36.920 is a good film, guys, I actually really like it. 0:03:38.440 --> 0:03:40.480 All the joy in any Matthew Broderick film is not 0:03:40.520 --> 0:03:43.440 the plot. It's it's Matthew Brodick, right. So at that 0:03:43.520 --> 0:03:49.840 point the computer learns what we know that did you 0:03:50.160 --> 0:03:54.960 take that back kidding. Matthew Broderick is a Broadway superstar. 0:03:55.720 --> 0:03:59.560 Um No that that in mutually assured destruction. Once you 0:03:59.760 --> 0:04:03.600 enter a conflict, it's game over and nobody wins. It's 0:04:03.680 --> 0:04:06.720 kind of right there in the name. Yeah, So comparing 0:04:06.920 --> 0:04:11.360 nuclear war to tic tac toe is absurd but also 0:04:11.600 --> 0:04:16.200 kind of awesome. That's able to make this abstract connection 0:04:16.240 --> 0:04:19.480 between the two. Uh, and tic tac toe is what 0:04:19.520 --> 0:04:21.680 we would call a solved game. And we'll talk more 0:04:21.680 --> 0:04:25.839 about what solved versus unsolved means a little bit later. Uh, 0:04:25.880 --> 0:04:28.360 And we wanted to talk a little bit more about 0:04:28.480 --> 0:04:32.440 games and computers and whether or not we humans are 0:04:32.520 --> 0:04:35.359 doomed to always come in second place to computers in 0:04:35.360 --> 0:04:38.159 the long run. Will computers get to a point where 0:04:38.160 --> 0:04:41.200 they will always be better at whatever games we pick 0:04:41.360 --> 0:04:43.599 than we are? Well, maybe the stakes are high for 0:04:43.600 --> 0:04:45.880 the computers too. I mean, I'm sure that neither of 0:04:45.880 --> 0:04:49.159 you guys ever watched the nineties CG Cartoon Show reboot, 0:04:49.240 --> 0:04:52.280 because like five people in the world watched the nineties 0:04:52.320 --> 0:04:54.719 cartoon show Reboots. I had a friend who is into it, 0:04:54.760 --> 0:04:57.239 and I tried to watch it once. It's real, silly, 0:04:57.320 --> 0:05:01.200 it's real, real, silly, I never tried to watch it. Okay, okay, 0:05:01.240 --> 0:05:04.920 in reboot, you guys. The premise here is that when 0:05:04.920 --> 0:05:07.880 a user runs game software on a computer, all of 0:05:07.920 --> 0:05:10.760 the citizens of the computer are that are caught in 0:05:10.800 --> 0:05:13.359 the game's path must fight for their lives according to 0:05:13.400 --> 0:05:16.440 the rules of the game, and losing the game to 0:05:16.480 --> 0:05:20.480 the user means decimation or worse than one tenth destruction 0:05:20.839 --> 0:05:23.000 for all of the ones and the zeros and the 0:05:23.040 --> 0:05:26.600 sprites who make the computer run. So it's kind of 0:05:26.640 --> 0:05:30.559 like Tron. Actually, when you play a game, the user 0:05:30.600 --> 0:05:33.400 plays the game, but the people inside the computer live 0:05:33.520 --> 0:05:37.120 or die. Yeah, in this case, the people fight against 0:05:37.279 --> 0:05:41.280 the users, not for the users. Okay, so that's probably 0:05:41.279 --> 0:05:45.000 not really what's happening in in reality, but at any rate, 0:05:45.560 --> 0:05:48.200 so we wanted to talk about this, and in order 0:05:48.200 --> 0:05:51.760 to do so, we wanted to first mention the concept 0:05:52.000 --> 0:05:55.720 of game strategy or game theory. So there are volumes 0:05:55.839 --> 0:06:00.359 written on game theory, like textbooks of in formation on 0:06:00.440 --> 0:06:03.960 logic and game theory. But ultimately, if you really want 0:06:03.960 --> 0:06:07.320 to boil it down to its basics, game theory comes 0:06:07.360 --> 0:06:10.719 down to using strategy in order to make decisions to 0:06:10.839 --> 0:06:14.520 achieve a goal. That's really what game theory comes down to, 0:06:14.600 --> 0:06:17.520 once you get rid of everything else. And not all 0:06:17.600 --> 0:06:20.680 games actually rely on game theory. Some of them are 0:06:20.760 --> 0:06:24.320 more about random chance and less about strategic decisions made 0:06:24.400 --> 0:06:27.279 by the players. Yeah, I mean there are slot machines. 0:06:27.400 --> 0:06:29.400 I guess that counts as a game. There is no 0:06:29.440 --> 0:06:32.000 game theory for slot machines. Yeah. No. You you might 0:06:32.080 --> 0:06:34.800 think that there's a system to win at a slot machine. 0:06:34.839 --> 0:06:38.440 You are incorrect. Uh, it's all based on timing. And 0:06:38.480 --> 0:06:41.239 the timing we're talking about is down to like hundreds 0:06:41.360 --> 0:06:44.440 or thousands of a second, so there's no way impact 0:06:44.560 --> 0:06:47.400 whatsoever on the outcome of the machine. It's pure chance 0:06:47.440 --> 0:06:49.400 if you have pushed the button or pulled the lever 0:06:49.600 --> 0:06:53.360 at just that right moment, or something like like candy Land, 0:06:53.360 --> 0:06:56.560 where you're just rolling dice and moving pieces. Yeah. Yeah, 0:06:57.000 --> 0:07:02.000 candy Land is like a progressive slot machine. Yeah. I 0:07:02.279 --> 0:07:03.960 wish I could make a candy line joke, but it's 0:07:04.000 --> 0:07:06.599 been a long time since I've played that game. Alright. So, 0:07:06.600 --> 0:07:09.520 so games that do involve strategy, you should try it sometimes. 0:07:09.520 --> 0:07:14.360 It's pretty sweet, nice, nice, all right. So anyway, games 0:07:14.360 --> 0:07:17.640 that are have straightforward strategies include stuff like chess. That's 0:07:17.640 --> 0:07:23.320 pretty obvious, right, Checkers or drafts if you're English und 0:07:24.120 --> 0:07:29.640 draft Yeah, drafts, I've never heard this. Draft Checkers, that 0:07:29.760 --> 0:07:33.520 is the pro that's the proper name for Checkers. Actually, yeah, 0:07:33.840 --> 0:07:36.640 the the actual older name for Checkers as drafts. Do 0:07:36.720 --> 0:07:40.280 they have cracker barrels in the UK? I hope not. 0:07:41.040 --> 0:07:45.000 I doubt it. Yeah, I don't. I've never seen one. 0:07:45.600 --> 0:07:48.560 Uh if if you work at a cracker barrel in 0:07:48.600 --> 0:07:52.040 the UK, give us a shout. So other games that 0:07:52.120 --> 0:07:59.240 also involved strategy connect for Tic Tac toe go is 0:07:59.320 --> 0:08:02.240 another one. There are lots and lots of these, and 0:08:02.360 --> 0:08:06.240 some of the games have enough strict parameters or rules 0:08:06.560 --> 0:08:10.680 in play that make them solvable. Right, And so this 0:08:10.800 --> 0:08:13.440 is a concept in in game strategy that there are 0:08:13.440 --> 0:08:16.480 some games that are solved in other games that are 0:08:16.520 --> 0:08:20.480 not solved. And basically, a solved game is a game 0:08:20.640 --> 0:08:25.280 where if you know the starting conditions and players play 0:08:25.320 --> 0:08:29.480 at optimal performance, meaning they all make the perfect or 0:08:29.600 --> 0:08:33.320 best decision every time. Yeah, you can predict exactly what 0:08:33.360 --> 0:08:35.200 the game is going to be. Right, this would be 0:08:35.280 --> 0:08:38.280 This would be a ken to having two people sit 0:08:38.320 --> 0:08:41.920 down at a chessboard and before anyone's even touched a piece, 0:08:41.960 --> 0:08:45.280 one person says mate, and twelve moves and the other 0:08:45.280 --> 0:08:47.840 guys like, very well done, and they shake hands they leave. 0:08:47.920 --> 0:08:50.280 But of course chess is not solved, No, it is not. 0:08:50.440 --> 0:08:53.920 Chess is much too complicated to have solved. You can't 0:08:54.040 --> 0:08:57.240 know what all of the possible moves in chess are 0:08:57.360 --> 0:09:01.559 and not like our current computing powers, right, Uh, yeah, 0:09:01.640 --> 0:09:04.960 I mean there's so many possible series of moves in chess. 0:09:05.000 --> 0:09:07.679 I kind of doubt that could ever be calculated. You 0:09:07.679 --> 0:09:10.320 are not the only person to doubt that. As for 0:09:10.440 --> 0:09:14.640 solved games, there are three broad categories of solved games. 0:09:15.120 --> 0:09:19.160 There's ultra weekly solved games, weekly being w E a K. 0:09:19.720 --> 0:09:25.080 Not not like every week a game this really really 0:09:25.120 --> 0:09:28.160 comes out every week. This would be this is really 0:09:28.200 --> 0:09:31.640 really not strong. Uh. You can only predict the outcome 0:09:31.679 --> 0:09:35.920 accurately from the initial position, meaning before any player has 0:09:36.000 --> 0:09:39.000 made a move. And uh, at that point, it would 0:09:39.040 --> 0:09:41.839 just say like, all right, well, assuming that everyone plays optimally. 0:09:42.559 --> 0:09:45.640 For example, with Connect four, the first player is always 0:09:45.679 --> 0:09:49.400 gonna wine if you always you know, if no matter 0:09:49.480 --> 0:09:52.720 if both players are playing perfectly, player one always wins 0:09:52.760 --> 0:09:57.800 Connect four. Uh. Weekly solved games include not just the 0:09:57.840 --> 0:09:59.920 prediction of the outcome, but also a strategy for its 0:10:00.000 --> 0:10:03.679 eiaving it from the initial positions. So essentially saying, here 0:10:03.880 --> 0:10:07.480 is the the process through which you will always hit 0:10:07.559 --> 0:10:12.640 this particular outcome. Uh. Strongly solved games include strategy to 0:10:12.720 --> 0:10:15.920 achieve the best possible outcome from any point of the game, 0:10:16.040 --> 0:10:18.240 even parts of the game where someone may have made 0:10:18.240 --> 0:10:22.040 a mistake earlier on. UH. So this would be where 0:10:22.040 --> 0:10:23.480 if you would look at a game that's already in 0:10:23.520 --> 0:10:25.800 progress and say, all right, from this point forward, we're 0:10:25.800 --> 0:10:27.840 going to have perfect play on both sides and you 0:10:27.840 --> 0:10:30.760 could still predict who was going to win. So you 0:10:30.800 --> 0:10:33.199 gave the example of tic tac toe in the movie 0:10:33.200 --> 0:10:37.280 War Games. But this is a solved game if players, 0:10:37.360 --> 0:10:41.400 If both players are playing perfectly, and there are perfect 0:10:41.480 --> 0:10:44.560 ways to play, there the optimal moves you can make 0:10:44.640 --> 0:10:47.920 in any given game, then it's always going to end 0:10:47.920 --> 0:10:50.360 in a draw. That's exactly right, which is why I 0:10:50.400 --> 0:10:53.240 say that's why the computer it comes to the conclusion 0:10:53.320 --> 0:10:57.040 that there is no point in playing because there's no 0:10:57.080 --> 0:10:59.880 way to win. So winning a tic tac toe do 0:11:00.200 --> 0:11:04.600 pens on your opponent making a mistake right exactly. You 0:11:04.640 --> 0:11:08.960 also mentioned Connect four. Yes, that one was solved in twice. 0:11:09.840 --> 0:11:13.600 It was by two different independent researchers, and it was 0:11:13.640 --> 0:11:16.440 weekly solved to show that the first player can always 0:11:16.440 --> 0:11:20.000 force a win given perfect play, and in it was 0:11:20.040 --> 0:11:23.000 actually strongly solved to the point where at any given 0:11:23.040 --> 0:11:25.160 moment within a game, if you took over from that 0:11:25.200 --> 0:11:27.679 point forward, you could you could predict who was going 0:11:27.720 --> 0:11:30.960 to win based upon that previous positioning. It wouldn't always 0:11:30.960 --> 0:11:34.280 be player one in that case, because they may have 0:11:34.320 --> 0:11:38.640 made mistakes. So next we have nine Men's Morris. I 0:11:38.679 --> 0:11:41.560 don't know what that is. If you played Assassin's Creed 0:11:41.600 --> 0:11:44.400 Black Flag, you would because it's a game that exists 0:11:44.440 --> 0:11:47.760 within that game and you get achievements for playing it 0:11:47.800 --> 0:11:51.079 and winning. So I played a lot of nine Men's Morris. 0:11:51.880 --> 0:11:56.760 Nine Men's Morris involves it is a not no, it's 0:11:56.760 --> 0:11:58.880 not specifically a pirate game. There's a board game in 0:11:58.920 --> 0:12:02.280 which you have spaces where you can place a piece, 0:12:03.040 --> 0:12:05.840 and so the first part of the game involves placing 0:12:05.920 --> 0:12:09.000 your pieces on the board strategically. The second part of 0:12:09.000 --> 0:12:13.040 the game involves moving those pieces along specific pathways that 0:12:13.080 --> 0:12:16.280 are available to you and you're trying to line up 0:12:16.480 --> 0:12:18.960 three pieces in a line, and if you can line 0:12:19.000 --> 0:12:21.320 up three pieces in a line, then you are allowed 0:12:21.360 --> 0:12:24.760 to take one of your opponent's pieces off the board. Well, 0:12:24.840 --> 0:12:26.920 whereas your opponent is trying to do the same thing 0:12:26.960 --> 0:12:30.720 while also preventing you from lining up three three pieces 0:12:30.720 --> 0:12:33.200 in a line. So one of the things I've noticed 0:12:33.240 --> 0:12:35.600 so far about all of these examples is that they 0:12:35.640 --> 0:12:39.840 don't include any element of luck. Yeah, these are all 0:12:39.880 --> 0:12:43.040 again very much strategic games. These are games where it 0:12:43.120 --> 0:12:46.080 requires the player to make a decision, and then the 0:12:46.120 --> 0:12:49.720 player's decision is uh, you know, whatever happens from that. 0:12:49.840 --> 0:12:52.440 The consequences of that are all based upon the strategy 0:12:52.440 --> 0:12:55.719 of the other player as well. So I'm sure we'll 0:12:55.760 --> 0:12:58.240 talk more in later in the episode about games that 0:12:58.280 --> 0:13:01.720 are a combination of skill in law. Yes, but but 0:13:01.840 --> 0:13:06.480 checkers or or droughts, drafts, Well, we're we're American. We 0:13:06.520 --> 0:13:08.960 can say drafts, some of the British douce drafts. But 0:13:09.920 --> 0:13:13.200 so you're saying checkers like they play at the cracker barrel? 0:13:13.840 --> 0:13:16.880 Was that's a solved game? Solved in two thousand seven. 0:13:17.640 --> 0:13:22.199 It had been close to being solved for much longer 0:13:22.240 --> 0:13:25.960 than that, but officially solved in two thousand seven, which 0:13:26.080 --> 0:13:30.160 was proving that from the initial standpoint standpoint, both players 0:13:30.160 --> 0:13:33.480 can play to a draw with optimal play. So so 0:13:33.520 --> 0:13:36.360 even if you go second, yeah, you can at least 0:13:36.400 --> 0:13:38.120 play to a draw. You may not win, but you 0:13:38.120 --> 0:13:40.480 can at least play to a draw. So if you 0:13:40.600 --> 0:13:43.920 are player too and you lose, it's your fault. Uh. 0:13:44.000 --> 0:13:48.079 It took. It took eighteen years to solve checkers. At 0:13:48.080 --> 0:13:52.200 one point they had two hundred computers working on this. Yeah, 0:13:52.240 --> 0:13:56.920 that's because there are five hundred billion billion possible arrangements 0:13:56.920 --> 0:13:59.800 that could appear on an eight pie eight checkerboard. So 0:14:00.200 --> 0:14:03.280 it's real impractical to analyze every single one of those 0:14:03.640 --> 0:14:06.080 that this The solving of it was the odyssey of 0:14:06.200 --> 0:14:09.880 one Jonathan Schaefer, who's a comp scientist who began with 0:14:10.000 --> 0:14:13.640 just sixteen megs of memory on a computer in and 0:14:13.640 --> 0:14:17.600 and built out this this solvability issue. Uh it took. 0:14:17.600 --> 0:14:19.680 It took really the scaling up of computer power to 0:14:19.720 --> 0:14:22.600 make it possible. This is you know this, this is 0:14:22.600 --> 0:14:26.320 not simple stuff. Obviously, it's interesting because a lot of 0:14:26.360 --> 0:14:29.240 these games are very easy for us to grasp as players. 0:14:29.280 --> 0:14:31.920 We understand the rules and the basics and the and 0:14:32.120 --> 0:14:37.280 general strategies pretty intuitively. But when it comes to proving 0:14:38.320 --> 0:14:40.960 the you know that you have solved the game, that 0:14:41.080 --> 0:14:45.280 you know definitively how the outcome will will be based 0:14:45.400 --> 0:14:49.920 upon perfect play, that's a lot trickier. Some games are 0:14:49.960 --> 0:14:53.920 only partially solved, So chess is partially solved, but only 0:14:54.000 --> 0:14:58.280 from an end game standpoint. Yeah. Yeah, So if you say, um, 0:14:58.320 --> 0:15:01.240 you know, the only pieces left on the board, or 0:15:01.280 --> 0:15:05.160 these five or something, and they're in these positions, there 0:15:05.320 --> 0:15:08.040 is a perfect way to play. And as you get 0:15:08.240 --> 0:15:11.520 more pieces on the board, it gets way more complicated 0:15:11.520 --> 0:15:14.880 because you have more options, more variables, and so I 0:15:14.920 --> 0:15:18.000 think when you get up to about seven pieces, it's 0:15:18.080 --> 0:15:20.760 really that's really like the limit of how far you 0:15:20.800 --> 0:15:24.840 can solve. And not all of those uh permutations are solved, 0:15:24.920 --> 0:15:27.520 yet some of them are, because again it depends on 0:15:27.560 --> 0:15:30.840 the combination of pieces. Pieces move in different ways depending 0:15:30.920 --> 0:15:33.240 upon there that what piece you're talking about, whether it's 0:15:33.280 --> 0:15:36.320 upon or a bishop or a rook or whatever. Uh. 0:15:36.360 --> 0:15:38.480 And then there's some other ones that have been solved 0:15:38.480 --> 0:15:41.480 for smaller versions of the game, like Go Go as 0:15:41.520 --> 0:15:43.200 a game, which is really interesting, and that you can 0:15:43.200 --> 0:15:46.280 play it on different size boards that determine how many 0:15:46.320 --> 0:15:49.120 pieces are in place. So a five by five board 0:15:49.240 --> 0:15:52.480 of go has been solved. There's a perfect way to 0:15:52.480 --> 0:15:54.800 play it, Yes, but most people play it on something 0:15:54.840 --> 0:15:59.240 like a nineteen by nineteen board, which is nowhere close 0:15:59.320 --> 0:16:04.560 to being solved. So again, is partially because of just 0:16:04.600 --> 0:16:08.320 the amazing complexity of the game at that scale. Sure, Now, 0:16:08.440 --> 0:16:11.640 whether or not a game has been solved doesn't necessarily 0:16:11.680 --> 0:16:16.880 mean that humans can beat a computer at playing it consistently. Yeah, 0:16:16.920 --> 0:16:19.400 as it turns out, computers don't have to have a 0:16:19.520 --> 0:16:22.200 solved that they don't have to know quote unquote the 0:16:22.240 --> 0:16:25.200 solution in order to still beat the pants off a 0:16:25.280 --> 0:16:28.880 human opponent, even a really good human opponent. Yeah. So 0:16:29.200 --> 0:16:33.680 chess is not solved, but we've gotten to the point 0:16:33.720 --> 0:16:37.160 where computers will always be the best human chess player, 0:16:37.440 --> 0:16:41.360 even way back. Yeah, this was the famous case. It's 0:16:41.360 --> 0:16:44.640 sort of like the people people likened this to the 0:16:44.680 --> 0:16:48.520 story of John Henry and the and the uh steam 0:16:48.840 --> 0:16:51.760 engine that was laying down tracks, like the man versus 0:16:51.840 --> 0:16:54.840 machine story. But this was the man versus machine story 0:16:54.840 --> 0:16:58.880 for the twentieth century. So in u there was actually 0:16:58.880 --> 0:17:03.840 a chess rematch between Gary Kasparov and IBM's Deep Blue computer. 0:17:04.280 --> 0:17:07.320 And Kasparov and Deep Blue had met in ninety six 0:17:07.400 --> 0:17:10.280 for a series of six games, which Kasparov won four 0:17:10.320 --> 0:17:14.840 to two. But in the nineties seven match, Deep Blue 0:17:14.880 --> 0:17:18.280 one it beat Kasparov. I think it's like three and 0:17:18.320 --> 0:17:20.639 a half games to two and a half games, the 0:17:20.720 --> 0:17:25.000 haves being I believe draws. So they you know, now 0:17:25.040 --> 0:17:28.520 we saw a machine beat a chess master for the 0:17:28.560 --> 0:17:31.760 first time. By the way, Kasparov actually demanded a rematch, 0:17:31.800 --> 0:17:34.480 he had match. Yeah, he had claimed that there was 0:17:34.520 --> 0:17:37.640 some hanky panky going on, and IBM declined his request 0:17:38.359 --> 0:17:40.960 because they said they essentially had proven, they had proven 0:17:40.960 --> 0:17:43.520 what they set out to prove. And since then these 0:17:43.600 --> 0:17:49.480 chess programs have become more powerful and sophisticated. But what 0:17:49.520 --> 0:17:54.159 they're actually doing isn't evaluating strategy so much as running 0:17:54.200 --> 0:18:00.119 through every single possible option at that time. Give then 0:18:00.200 --> 0:18:03.760 the board and the peace position. Yeah, like in encryption 0:18:03.880 --> 0:18:06.760 or decryption, I should say, this is called brute force. Yeah, 0:18:06.840 --> 0:18:10.119 you're essentially throwing everything you can at the system to 0:18:10.160 --> 0:18:13.439 find out what works right. So that's that's really what 0:18:13.480 --> 0:18:15.919 brute force is all about. So it might look at 0:18:15.960 --> 0:18:18.480 the chess board and say, okay, if I move my 0:18:18.600 --> 0:18:22.240 queen here, what would be my opponent's best next move 0:18:22.880 --> 0:18:25.800 or what would be all of his or her possible 0:18:25.840 --> 0:18:28.560 next ye yeah, yeah. So let's say let's say that 0:18:28.680 --> 0:18:32.320 you say, all right, let's move my queen. Uh up, 0:18:32.720 --> 0:18:35.560 you know, one square we're moving it, moving the queen 0:18:35.560 --> 0:18:38.679 forward one square. What are all the possible responses to 0:18:38.720 --> 0:18:41.240 that move? And then what are my responses to those? 0:18:41.640 --> 0:18:43.880 And then after figuring all that out, all right, well, 0:18:43.880 --> 0:18:46.080 what if we move the queen two squares? What are 0:18:46.160 --> 0:18:49.120 all the possible responses and what are my And that's 0:18:49.119 --> 0:18:51.880 for every single piece that's in a position that can 0:18:51.920 --> 0:18:55.480 be moved. And remember chess has certain rules that also 0:18:55.640 --> 0:18:59.600 complicate things like there's the rule about castling. Castling adds 0:18:59.640 --> 0:19:03.720 another variable to that sort of stuff. There's also timing. 0:19:03.880 --> 0:19:07.479 I mean, you can't have the computer takes seventeen hours 0:19:07.520 --> 0:19:10.359 to decide what moved moved to make yea. So now 0:19:10.480 --> 0:19:13.879 those roote force with computers that are sufficiently they're sufficiently fast, 0:19:14.400 --> 0:19:17.960 they can make these decisions relatively quickly. If we humans 0:19:17.960 --> 0:19:20.680 played like this, a chess game would last year entire life. 0:19:21.160 --> 0:19:23.080 You would never finish the game because you would be 0:19:23.400 --> 0:19:27.600 constantly evaluating all these potential moves before finally settling settling 0:19:27.600 --> 0:19:30.919 on the one that's waited to be the strongest. But 0:19:31.040 --> 0:19:35.440 that's essentially what chess games and other computer games are doing. 0:19:35.480 --> 0:19:39.159 They are looking at all the different possibilities and picking 0:19:39.160 --> 0:19:42.080 the one that's the most advantageous or at least advantageous 0:19:42.080 --> 0:19:44.920 to a certain degree. Because if you set your difficulty, 0:19:45.280 --> 0:19:47.360 because a lot of chess programs allow you to set 0:19:47.400 --> 0:19:49.919 the difficulty of the thing, what they'll do is they'll say, 0:19:49.960 --> 0:19:53.520 all right, well, we'll just look for x amount of 0:19:53.560 --> 0:19:56.840 time and use the best one out of all of that, 0:19:56.960 --> 0:20:00.399 rather than evaluate the entire board and all the pieces. Yeah, 0:20:00.440 --> 0:20:04.440 and so in this case, the strength of the program 0:20:04.440 --> 0:20:09.120 would probably be based on something like how many moves 0:20:09.160 --> 0:20:13.439 in advance can it look right? And and how quickly 0:20:13.520 --> 0:20:16.679 can it execute that, because if it can't execute it 0:20:16.720 --> 0:20:19.080 within enough time, it may have to, you know, cut 0:20:19.119 --> 0:20:22.720 back on that. Um So, one thing I thought was 0:20:22.760 --> 0:20:27.000 really interesting is that there's a new approach to creating 0:20:27.080 --> 0:20:30.000 a chess playing computer that does not rely on brute 0:20:30.040 --> 0:20:33.520 force and It comes from Matthew Lay of the Imperial 0:20:33.560 --> 0:20:37.160 College of London, and what he did was he developed 0:20:37.480 --> 0:20:40.480 a learning algorithm and a neural network, which we've talked 0:20:40.480 --> 0:20:43.040 about so much on this show, right, neural networks and 0:20:43.080 --> 0:20:45.720 learning algorithms to teach a computer rather than to program 0:20:45.720 --> 0:20:49.680 a computer. So he taught a computer chess. He uh. 0:20:49.840 --> 0:20:54.080 He created a computer program called it Giraffe and started 0:20:54.119 --> 0:20:56.800 to teach it how to play chess. Before he even 0:20:56.920 --> 0:21:01.200 got started teaching at chess, he tested it against a 0:21:01.200 --> 0:21:05.040 a standardized system that is used to evaluate how well 0:21:05.119 --> 0:21:08.920 a computer program does in chess, as a top score 0:21:09.000 --> 0:21:12.720 fifteen thousand, so before we had even schooled it, where 0:21:12.720 --> 0:21:14.359 it knew the it knew the less. It knew the 0:21:14.440 --> 0:21:17.920 rules of chess, but didn't hadn't really learned the value 0:21:17.960 --> 0:21:22.480 of various gambits. It scored six thousand, which is not bad. 0:21:23.520 --> 0:21:25.800 After he schooled it, and by that I mean he 0:21:25.880 --> 0:21:30.560 fed on five million different chess positions into the database, 0:21:30.880 --> 0:21:33.880 because you need huge data sets for learning algorithms to work. 0:21:34.640 --> 0:21:38.320 Then he set Giraffe against itself. It played itself in 0:21:38.359 --> 0:21:41.240 a series of games to start learning which positions were 0:21:41.280 --> 0:21:44.080 the most valuable, which one's made you vulnerable, which one's 0:21:44.119 --> 0:21:47.879 led to victory or defeat. He then tested it again 0:21:47.960 --> 0:21:52.200 and this time it scored uh, like nine thousand, seven hundred, 0:21:52.480 --> 0:21:55.399 so much higher. Yeah, yeah, And that still means that 0:21:55.480 --> 0:21:58.920 it's not perfect, right, but it's it's at a level 0:21:59.040 --> 0:22:03.119 that is consider stint with an international master of chess. 0:22:03.160 --> 0:22:05.560 So this is a computer program that learned chess, and 0:22:05.600 --> 0:22:09.760 it looks at the entire board and the piece positions 0:22:09.840 --> 0:22:12.880 on the board and then evaluates that based upon their 0:22:12.880 --> 0:22:15.560 attacks and defense capabilities. So it's kind of the way 0:22:15.600 --> 0:22:18.879 we play chess. Yeah, exactly. I think statistically speaking, it 0:22:18.920 --> 0:22:22.040 only picks the quote unquote best move about half of 0:22:22.040 --> 0:22:23.960 the time. Yeah, a little less than half, like forty 0:22:24.040 --> 0:22:26.359 six percent of the time it picks the the quote 0:22:26.440 --> 0:22:28.680 unquote best move, but the best move is it tends 0:22:28.720 --> 0:22:31.879 to be in the top three choices around seventy percent 0:22:31.920 --> 0:22:34.399 of the time. And that's pretty good when you consider 0:22:34.480 --> 0:22:37.280 it's not doing brute force, it's not planning all of 0:22:37.280 --> 0:22:42.199 these out. It's it's essentially taking a look and then intuitively, 0:22:43.040 --> 0:22:44.880 which is a weird thing to say, but that's as 0:22:44.920 --> 0:22:47.639 close as I can get feeling out which one is 0:22:47.680 --> 0:22:50.920 the best approach, So about half the time it gets 0:22:51.000 --> 0:22:54.080 the absolute best choice, and seventy percent of the time 0:22:54.400 --> 0:22:57.199 the best choice is within its top three options. But 0:22:57.320 --> 0:23:00.159 this doesn't mean that you cannot beat the pants right 0:23:00.200 --> 0:23:02.880 off of a computer in some games. Yeah, you can 0:23:02.920 --> 0:23:05.520 totally do that in some games. There are some games 0:23:05.520 --> 0:23:09.560 that computers still cannot beat the best human players, And 0:23:09.600 --> 0:23:12.640 then there are other games where computers are just horrible, 0:23:12.720 --> 0:23:15.480 like they can't even beat average players. Yeah, if you're 0:23:15.640 --> 0:23:21.280 really stronge, well, yeah, when you when you play a 0:23:21.320 --> 0:23:23.280