This is the video transcript. Read the original article with all the details →
Good morning and welcome back to the ValorosoIT channel, the channel dedicated to retrocomputers and vintage electronics. Today we see an interesting conference held by Carlo Santagostino on 17 September 2023 in Tradate, at Varese Retrocomputing. If you like these topics, that is: historical computing, video games, vintage computers, I already invite you to the subsequent editions of Varese Retrocomputing which will obviously be held in the next few years on dates to be defined, but also to subscribe to my ValorosoIT channel. You can find it on YouTube, on Instagram, on Facebook, on Tiktok... In short, on all the social network platforms. And you can also visit the blog www.valoroso.it, where I publish summaries of the articles and videos which you can then also find on the various social network platforms.
The topic of artificial intelligence is very fashionable in this period and Carlo Santagostino talks about it by addressing its historical evolution in video games. Carlo Santagostino is a computer science expert, also an expert in the history of computer science. He is a computer science professor in the television series Il Collegio, and is also the founder of RetroCampus, an association that deals with the dissemination of topics relating to retrocomputers and vintage video games and historical preservation. I thank Fabio Massa, known as BioMassa, and Mariangela Sapia for the audio and video recording. And now let's see this interesting conference. HI!
Thank you all, welcome! Welcome and what are we talking about? I am Carlo Santagostino. For those who don't know me, I have been involved in disseminating the history of information technology and the history of video games for several years. I am a computer scientist and have been working in this field for years now. Also because if I look old, I'm actually young. No, it's actually the opposite. I'm 53 years old and I started programming when I was 12, and I've always been involved in IT. It has always been my passion, my study and the work I have done.
My first job was as a video game programmer in the 80s, in the early 80s I published video games on the Sinclair ZX Spectrum, like many of those up here, who show off their vintage creations upstairs. However, in 2010 I founded an association called RetroCampus, and I am one of the founding members. Me and other people. Above you will find some members of the association who are having an exhibition on the upper floor. The mission of RetroCampus is precisely that of disseminating and preserving the history of computing and the history of video games.
I give many conferences on different and varied topics. In particular, this year there is this buzzword: AI, artificial intelligence, which is very, very fashionable at the moment. 'Artificial intelligence', or AI. Today, like today, saying 'artificial intelligence', you say it and everyone...
In this conference I will talk about the use of artificial intelligence algorithms in video games. However, now we will not stay here to discuss artificial intelligence in general, the various approaches, etc... But I have to give you a little introduction. More than anything I have to explain that, nowadays, when you hear the word 'artificial intelligence', it is obviously, in general, at this moment you often hear it linked more as a marketing term.
It is used to describe computer programs that use simple algorithms, even ordinary ones, such as sorting and matching, to create the illusion of intelligent behavior. They therefore give software this aura of complexity or scientific and technological progress and say: 'made with artificial intelligence.'
Well, in the majority of cases this is not the case, in the majority of cases. Why is this? The real field of research of artificial intelligence algorithms is to try to imitate the intelligent behavior of us human beings for solving problems, artificially recreating characteristics such as machine learning or decision making. So this, and only this, is the field of artificial intelligence and algorithms that deal with artificial intelligence. Why do I say imitate intelligent human behavior? Because now we won't go into detail, this is not a conference on artificial intelligence. But I have to give you a little introduction. Currently, all the algorithms from the history of artificial intelligence, which date back to the 1950s, there are studies and research in this field, in reality they try to imitate what we know how to do, that is, look for a solution to problems. So we try to recreate in an artificial way, using mainly information technology, which was the science that made this possible, even if it was also theorized before with mechanics with other possible solutions, to recreate a kind of coherent logical thought. With information technology we have been able to recreate this somewhat. But it's about imitation. Imitate and let's not talk about reaching our intelligence. This is a topic that is being discussed extensively here these days, but we won't go into it because here it gets philosophical. There are ways of thinking, there is nothing real yet, there are ways of thinking of a certain type and ways of thinking of another type, which say that one we will get there, others we won't get there. But, to give you an idea, I use an example that I use often, that I do often, to make you understand exactly what I mean by the fact that currently any artificial intelligence algorithm imitates our intelligence, but it is not our intelligence. The example is this: you will surely have seen, currently, with computer graphics, with CGI it is possible to recreate real situations in an impressive, perfect way. Very simple, for example there are some rain scenes. We see the rain. Ok, there are now some scenes recreated in computer graphics of the rain water falling: it's perfect. It really feels like it's raining, okay. So did we recreate the rain artificially? No, we simulated the appearance of the rain, that is, what we see as the rain. But if we really had to artificially recreate the effects of rain with software, what should we do? It is not just the water that falls, the rain is part of a very complex cycle, which is that of the water cycle, from the evaporation of the sea to condensation, to the formation of clouds, the formation of clouds with all the physics of the clash of clouds, of the creation of these drops of water in the atmosphere which then fall onto the earth, which drag material that is suspended in the atmosphere and irrigate the soil, which nourish only in a certain way and then create the possibility for plants to be able to absorb nutrients from their roots.
So, think about how complex the rain cycle is. So I recreated the rain by using computer graphics to see the rain falling? I made it sound like it's real rain, but it's nothing, but it's not even a hundredth millionth of what rain actually is. Well, currently this comparison that I made to you is what it is now, what artificial intelligence algorithms are now. We talk to ChatGPT, we see MidJourney drawing, it feels like we're talking to a person, it feels like we see a person drawing, but it's nowhere near a millionth of what our coherent thought is, okay?
This is the current state, where will it take us? Here there are theories, then there are those who say that by increasing complexity we will arrive at general intelligence, etc. There are those who say that we won't get there because the path is another, but there we enter the philosophical and therefore I won't go there. The topic of the conference is different. Here we will talk about video games, today, and the use of artificial intelligence, artificial intelligence algorithms in video games, because, contrary to what I said before, that is where now it is enough to stick - I started by saying that today as a marketing label artificial intelligence is put in accounting and payroll software: made with artificial intelligence. Why? Because it's marketing right now, it's not simple sorting algorithms, but it becomes a thing, no, by putting in... - instead in video games there is a particular connection, because what we could define video game artificial intelligence exists.
Because in video games there has always been a tendency to create what is an artificial person. What is the artificial person? The one we play against, that is, that the algorithms that underlie the actions of non-player characters are focused on the appearance of their intelligence, to give good playability within the restrictions of the game environment. So the approach is similar to traditional AI research, what I said before. What is the approach of artificial intelligence algorithms? That of creating an appearance of intelligence similar to ours, that is, that of recreating an artificial person. When, in 1950, Turing published the one on computer science, computer machine intelligence, his treatise, which defined the famous Turing test, was precisely to say, let's make a program that looks like a person, okay? So, in reality, and in fact if you study the history of artificial intelligence, you find video games right away. Because the first examples of recreating the human thought process, of creating something that feels like a person I'm interacting with, was just playing against someone else. In fact, right from the start, until 1950, that is when Turing wrote his treatise to define the Turing test, things like this happened in the computer science environment of the time: the famous Bertie the Brain. No?
Computer companies used games to prove that their computer worked, then wrote game algorithms to play against people. These are among the very first studies of artificial intelligence algorithms, that is, trying to understand how to create an algorithm that plays against a person. Bertie and the Nimrod are two examples, one from '50, one from '51. Bertie the Brain was made - sorry, but I use notes because the topics are vast - then, it was built in Toronto by Josef Kates and it was presented at the 1950 Canadian National Exhibition and it demonstrated a new type of tubes because in those days computers were tube-based, there was no transistor yet, computers were tube-based. It was an advertisement to show the valve function of Additron, which was invented by this Kates, it was a more compact, smaller valve, so it managed to reduce the enormous dimensions that calculators had at the time. And how did he do it? He demonstrated by playing tic-tac-toe, the famous tic-tac-toe, against a human opponent. So he had written an algorithm that played tic-tac-toe against an opponent. This is already a proto AI algorithm, in fact. The same thing happens to Ferranti. Ferranti, in England, in 1951, to advertise at the Festival of Britain, played the Nim game, which is another classic game, no, the Nim, the one with the moving tablets, etc. The Nim is done with the Nimrod, no, on the Ferranti Mark 1, which was the first commercial computer in history, that is, one that could be bought. Because the previous ones, however, were all for university or military use and study, while the Ferranti Mark 1 could be bought, so companies could, large companies, we are talking about multinationals, could buy a computer. In fact, the Ferranti Mark 1 was made, where, in fact, there was the presentation of the Nimrod, again in '51, at the University of Manchester, where the hardware that was the basis of the Mark 1 was designed. This gentleman here, no, that is the Ferranti advertisement, no, Electronic Digital Computer, and this, this gentleman who was Christopher Strachey, was the first to write software for playing games. First, he wrote it for checkers and then Dietrich Prinz, also at the University of Manchester, the first computer program that played chess, therefore a complex game, a very complex game like chess.
For many, replicating the human reasoning that underlies chess demonstrated that it was possible to recreate logical reasoning.
In fact, these two games written at the Manchester unit, which are among the first computer programs ever written in history, from '51, are in all artificial intelligence texts as the first examples of artificial intelligence algorithms. Precisely because they recreate what human reasoning is. So, they are recreating what was the first time we played against an artificial machine. This attempt that had been made, you remember the famous mechanical chess player, who was fake, there was the dwarf inside who had created everything... but for centuries people had been trying to build a machine that could play against a man and, finally, with computers it can be done. A machine is created that plays against a man, therefore a machine that reproduces human reasoning and allows you to play against a man for the first time.
Consider that from then on, the evolution of the game of chess was incredible. It took him until '97 to get there, from '51, you count the years, from '51 to '97. In 1997, IBM's Deep Blue managed to beat Garry Kasparov at chess. So, even, to beat the world chess champion. And this was in 1997. Why the Mark 1 matters, et cetera. The Mark 1 is also important for another thing, because we also have the first example of the game, always in the research field, this happens in '52, A. S. Douglas, who is at the University of Manchester, also on the Marc 1. The Mark 1 was the first computer to use a cathode ray tube. In reality, he didn't use it, they hadn't put it there, the cathode ray tube, in what you see is the control center, was not there to interact with man, but it was there because the cathode ray tube was used as data storage. In practice, since the cathode ray tube has a latency, the memory cells were written with the electron beam which could then be measured, so there was like the delay lines in Mercury, the same thing. Except he used the tube to do this.
Alexander Shafto Douglas had the intuition, as you could see these dots coming out, which then gave the status of registers, memory locations, etc. He said: 'But I can use these dots to also interact with the operator in front', and in fact this is the first example in the context of his PhD thesis, which was Human-computer, Human-machine interaction. He used a cathode ray tube for the first time to visualize something that the person interacted with, and he did this, which is remembered today as 'OXO', which is considered practically the first computer game ever, because it uses a cathode ray tube, a digital computer and it was made in '52.
So, in '52 we also have this. Again, it's like Bertie the Brain, the algorithm is very simple, the game of Tic Tac Toe is very simple, compared to chess. In reality, they had already played the game of chess with the computer in '51. Only, you saw the photo from before, he went with the punched tape that gave him the positions, and he responded by moving. Instead, with this, already on the screen, I interact and see what happens on the screen. So from here, to this point, oh well, now there is the explosion, because we have that information technology, starting from those 50s, during the 60s, explodes all over the world. After Ferranti, there will be the birth of the Digital Equipment Corporation, IBM computers, there will be all the big companies that will start selling computers in the world. Computers will begin to spread mainly in universities, large computer centers, research centers and large companies. But to spread, right? And above all, many, including new researchers, new people, will therefore begin to be able to get their hands on these enormous computers, this innovation that is emerging. And from here, obviously, the kids at university study, research, but above all play. So they started thinking they could play too. It is also true that the first examples of video games, which are often cited, after the one that was 'OXO' and these that I have just mentioned, where there was an interaction, therefore there was a search for the creation of the artificial intelligence character, the player who does not exist, of the artificial person to play against. Those who, in reality, everyone remembers, so Space War, to name one. In the books you will often also find 'Tennis for Two' indicated, but 'Tennis for Two', in reality, I willingly remove because in reality it is on an analogue computer made to advertise a ballistic launch computer. Two people saw it...
Then, for various reasons, it became famous in the books because it is one of the examples that has survived, everyone cites it, but both 'Tennis for Two', however, and 'Space War', above all, born on the PDP-1, is what is considered by all standards a video game, with the two spaceships. Up to Atari's first commercial video game, and the first commercial video game, inspired by 'Space War', and then the classic 'Pong', which spread the video game all over the world. All these examples I gave, they're all two-player video games. I mean, you can't play 'Space War', 'Pong', 'Tennis for Two' alone. You can't play alone, you have to be against each other, which is very fun, because it allows you to challenge each other and there is fun between people, but there is no artificial person creation algorithm, so they don't fall into what are the field of the so-called artificial intelligence research, the creation of the artificial person.
To find other examples of this type, we have to go to the field of university research, where many video games began to be developed with a language born in the 1960s, which is BASIC.
BASIC was the first multipurpose language. That is, without being born for a specific function. It is very simple to learn and therefore to master, and IT has spread at all levels since the 1960s. It was the most used programming language for everyone, for the second half of the 60s, for all of the 70s and for most of the 80s. At any level, any computer, from university supercomputers up to our home computers, at least until the birth of Dennis Ritchie's C which came later. From there, we started programming in C, at a higher level, and so on. But BASIC was widely used, because the other languages that were born before had specific tasks. They were created for mathematical functions, for administrative functions, therefore they did not have this general purpose, this generic use.
The first one we're going to remember is a program from '71, here I found a small piece of listing of the '73 version, which is obviously inspired by Star Trek, because all the nerds of the time like, as many now like Star Trek, and strategy based and inspired by the famous television series of '66, it was originally published in '71 and was widespread within the university and was a strategy game which was the basis, then, of many video games that came later
In reality, this game was designed to be able to be played even with computers that didn't have a screen, because to have a screen, you had to have the terminal, and many university computers in those years, in the 60s, in the 70s, often didn't have the terminal. You wrote the program on punched cards and sent it, and then the result came to you from the teleprinter, so it was printed, or you got back the, always the, punched sheet of paper, or you got back, at most, a typescript from a typewriter connected to a computer. And in fact this game was designed to be able to be played on paper, practically. I mean, you gave commands, and then he printed out the map for you, of space, gave you the location of the spaceship, of the Klingons, etc. And it was a strategy video game, so there were these algorithms that simulated the intelligence of the enemy ships, therefore of the Klingons, where they moved, what they had to do, etc., and you had to play with strategy, positioning, etc.
In short, a video game that was the basis of many video games that came later. It was also one of the most popular video games in the world, when it was released on a text that was published in the early 70s, in '73, the first version which was '101 Basic Computer Games', which was basically a book by David H. Ahl, who published this book with the collection of 100 computer games written by students during the 60s and 70s. And 'Star Trek', which was the most complex one inside, was also what made people buy this book who then with the first hobby computers, so the Altair, etc., first computers that are becoming widespread in the mid-'70s, between '75 and '76. And then it was released for everyone, 1000 versions, it came out for Apple II, Commodore PET, TR-80, etc., etc., etc. But what always happens in the 80s, a big thing happens, actually, a thing that not many people know about.
This happens: a system called PLATO is invented, some of you will have already heard it, some not, but I'll basically tell you in two words: PLATO (Programmed Logic for Automatic Teaching Operation) was the first computer-assisted education system. It started in 1960, on the ILLIAC computer at the University of Illinois, and in the 1970s it was used in practically all the universities in the world, which were connected to this PLATO network, which is practically a proto-internet, because everything we do on the internet now, in reality, was born on this PLATO network, which was the first place where chats were born, where forums were born, and where video games were born, even online, multi-games, video games with graphics were born above because they were developed on the Plato system. It used plasma graphics terminals, with a plasma matrix of 1024x1024 points, therefore very high, and allowed you to draw or use graphics with these PLATO systems.
Above all, the first examples of role-playing video games, Dungeon Crawlers, were born on PLATO, obviously directly derived from Dungeons and Dragons, the very popular board game. 'DnD' is a video game written in 1975, as you can see it already had graphics and everything. And the other is Pedit. One is 'DnD' and 'Pedit'. They were born at the same time, between '74 and '75. Gary Whisenhunt and Ray Wood at Southern Illinois University wrote 'DnD', and 'Pedit 5' was written by Rusty Rutherford in '75. They are known for being practically the first video games ever in these years that had, for example, 'DnD' had the so-called final boss, that is, you went around the levels, at a certain point you found a very strong enemy that you had to fight against. First you found simple enemies, and then came this more enemy... wizard, dragon, in short everything set in the world of Dungeons and Dragons. And they are seminal video games because in reality from the development of these video games which had an enormous diffusion in university environments, they will then also be brought to computers with terminals, therefore made with characters, but after having been born graphically on the Plato system.
From these experiments which circulated more or less freely, came the first commercial productions for personal computers in the early 1980s. And the famous 'Akalabeth: World of Doom', from which Richard Garriott's 'Ultima' was born, is directly inspired by these video games, even as a command interface, etc. That is, Garriott didn't invent it, he took it from these games that were freely circulating in university school environments in those years. And he, in 1980-81, made the first version of 'Akalabeth' on the Apple 2, then made 'Ultima', which was then made for the Commodore 64, then spread all over the world, arrived, became a success, even in Japan, etc., etc., etc. Then from there the whole history of role-playing games was born.
And here there is a lot of research in what is the creation of the character of the artificial enemy, because they must be endowed in a certain type of intelligence, that is, they must have some kind of interaction. Here it was easier because we are talking about computers which in those years were the most powerful computers you could find on the face of the earth. In reality, however, the video game was also spreading to another environment. I mentioned 'Pong' earlier. Do you remember it? Here it is, I'll open a brief parenthesis: often when you search for artificial intelligence and video games on the internet, 'Pacman' comes up. As if to say, 'Pacman' was the first. As you've seen, it was a long road before we got to 'Pacman'. But, in reality, the video game was becoming famous in what were the arcades, the coin-ops, that is, the video game was spreading all over the world in what was the arcade video game market, therefore in the arcades, in the bars, at the time, etc.
But the problem is that while here to develop these, you had computers with excessive CPUs. CPUs, not even microprocessors, CPUs that took up rooms. In Arcade video games, at the beginning, you didn't even have your own microprocessor, they were just TTL logic, they were logic wired into electronics. So you had to program with TTLs, you wanted to try to recreate even minimal intelligence of something. In fact, all the first video games are dual player, or at most, if they are single player, they have enemies that follow fixed logic, predetermined logic, they don't have a minimum of intelligence. In fact, why do they mention 'Pacman'? Because if we look at the most famous video games, we remember 'Space Invader' (78), 'Galaxian' (79), in 1980 we arrive at 'Pacman'. 'Pacman' had enemies who had some kind of minimum intelligence. That is, at least the problem of giving a different character to each of the characters had been addressed. And so, while before, in Space Invaders, the movement was fixed, they always made the same movement, 'Galaxian' adds variables but it's always fixed. They are predetermined patterns, that is, the enemies always go the same way, they always make the same turns, they don't change. With 'Pacman', however, you have the impression that they are following you. So, 'Pacman' is the first Arcade example? No, obviously not, because obviously, if you go looking, you discover that it's not what you might expect. What is the first video game where you start to insert a minimum of intelligence into the arcade? Back then, as I told you, video games were exclusively TTL, until 1975, where Williams introduced a microprocessor into video games for the first time. And he does it with a video game called 'Western Gun', it was a Japanese video game developed by Taito. Taito develops in TTL, is sold to Williams. Williams, to improve production cycles and save money, decides to use microprocessors instead of TTLs. He takes the Intel 8080, then remakes 'Western Gun' with the Intel 8080. But he remakes a game that was TTL for two players. 'Western Gun' was one gunslinger against another, but you had to play against each other, the gunslinger didn't automatically move. Okay? And the same logic is also reported in the microprocessor. So, even if it is the first game with a microprocessor, which theoretically has the potential of a home computer, a personal computer, in reality it doesn't exploit it, because it plays exactly the same game.
There was another very famous game at the time, the most famous video game after 'Pong' in the early 70s, was a video game called 'Gran Trak 10' by Atari. They were the famous toy cars seen from above that were spinning. The first ones were in TTL. 'Gran Trak 10', 'Gran Trak 20', are in TTL, and you can play with one to four players. But even if you play alone, the cars and opponents always move along the course, without the slightest intelligence, nothing, zero. However, since this is a very successful video game, Atari continues to make sequels to this video game, and in 1976, it produces 'Sprint 2'. I mean, was there 'Sprint 1'? No. 'Sprint 2' was called that because it had two steering wheels, 'Sprint 2': two steering wheels. Okay. Which is this. 'Sprint 2' uses a microprocessor for the first time in Atari, the 6510 from MOS Technology. It is put in this video game, 'Sprint 2'. And, thanks to the fact that they put the microprocessor in it, they manage to make sure that even if you play as a single player, first of all there are always four toy cars, even if there are two players: one is the white car, the other is the black car. Then there are always two gray cars. The two gray cars are always driven by the computer and have a certain intelligence. That is, for the first time they insert algorithms such that these two toy cars seem to be driven by another player. That is, they react to the fact that you are driving, so they approach, move, try to overtake you. That is, they interact, they are not a fixed track that I just go, okay? They look like two players. It was difficult to find why it wasn't written anywhere, I had to dig, in the end I figured it out using the flyer of the original video game, of the advertisement, where it explained that you can... 'One Player Racing', as you can see, it also says 'One Player Racing', so only by yourself can you play against the computer. And then he explains that the two gray cars are always driven by the computer. Even if you play two players, one black and one white. If you play alone, black is also controlled by the computer. And he specifies in this that, thanks to the use of the microprocessor... I can't find the place where it is written, but it is written, it said: 'we use a microprocessor'.
So, he gave this emphasis to the fact that it uses a microprocessor, therefore a program written with a minimum of intelligence algorithm. This happens in '76. It is currently the oldest I have found in the Arcade video game. I haven't found any others, for now, and then maybe it will turn up, but for now I've only found this one. However, I also asked myself the problem in the home video game, because in those years, at the same time as the Arcade video game, the home video game was also born. Magnavox had introduced the famous Odyssey with Baer, which was the first console in '72, at the same time as Atari's 'Pong' in the Arcade. And then the RCAs were being born, various consoles with cartridge microprocessors. Then in '77 there was Atari which presented its Atari Video Computer System, the one that would later trigger the craze for domestic video games at home. Atari VCS Video Computer System, later called 2600, many of us know it as 2600 now, 2600. But actually, the original name was VCS, Video Computer System. And he comes out with some games planned… Ah, here he is, perfect. It's him. Combat? Actually, yes, but it had, because it comes with pre-set video cassettes, no, cartridges, a series of cartridges. This series of cartridges were almost all simple games, for one or two players, or for one player, but without a minimum of intelligence. However, there were some who had a modicum of intelligence. What was the problem? That the cartridges, do you know how much memory those cartridges had there? 2 KiloBytes. 2KB. That is, you have to fit all the game code in 2 KB. 2 KB ROM, 128 bytes RAM. This was the Atari 2600 hardware, okay? So, in 2 KB you struggle a bit. In fact, 'Air-Sea Battle' and 'Combat', when you played… if there were two of you playing, it was a lot of fun. If you played against the computer, it shot at random, it didn't have a minimum of intelligence, that is, it was programmed to shoot at random. In fact, he kept shooting in the air, he shot from all sides, and then every now and then he hit you by chance, but he really didn't have any intelligence algorithm. They were preset, like in 'Space Invaders'. They had to do that thing all the time, they didn't react in the slightest.
What is the first of these games where there is a minimum of this type of code? I found it in this cartridge, 'Video Olympics'. 'Video Olympics' compared to simplicity, 'Video Olympics' were the 'Pong' games, even for the VCS. 'Pong' games were very popular at that time. So, let's also put the one with the 'Pong' games in the basic VCS cartridges. The incredible thing, 'Video Olympics' was programmed by Joe Decuir, French name, so I pronounce it very badly, spelled 'Joseph Decuir', so. This guy, who is he? He is one of the two designers of the VCS, he is practically Jay Miner's deputy, who will accompany him for all the projects up to and including the Amiga. Joseph Decuir wrote many... he also designed many parts of the Amiga together with Jay Miner. He writes this 'Video Olympics' in 2 KB. And what does it put in there, in 2 KB? 'Video Olympics' features 50 different games, with variations. You see some of them. In addition to 'Pong', it provides variations with goals to play Soccer with more players, with fewer players, with goals and without goals, with basketball, turning the paddles horizontally. So… I did it just to show you these are the 2K of the game. Everything, the whole cartridge is in here, okay. This 2 KB, 2048 Bytes. And what does he put in 2048 Byte? He also puts in something that, for the first time in history, you can play 'Pong' against the computer, and he calls it 'Robo Pong', that is, you can play 'Pong' not against another, but against the Atari. And I don't plan... easy, if you make the bar that follows the ball, he always wins. I mean, the end. You synchronize the bar with the vertical axis of the ball, and he always gets it. So, what does he invent? An algorithm is invented which, every so many times, loses synchronization with the vertical axis of the ball, and furthermore, every bounce of the ball against, this counter is reset. So it gives the illusion that it's not perfect, but it puts human error in and it's a lot of fun to play with. In fact, it has a lot of positive reviews. Do you think that this very simple algorithm, inserted here, that is, together with all the game variants of 'Pong', all in 2 KB... all in 2 KB... succeeds, this simple routine, but it works so well that the developers of the 'Arcade Atari Arcade Hits' collection of 1999, who made Atari's 'Arcade Hits', used this algorithm by Joe Decuir to copy it exactly as it is and the same for the '99 remake of 'Pong'. They used this algorithm. So, he invented it, it fit in 2K, and it worked, and it worked well too. So it's impressive to think about what Joe Decuir did in 2KB.
But it doesn't end there, because obviously home consoles amplify this desire to play against the computer. Because, while if you went to the arcade, you always found someone to play with, you were at home, maybe you were alone, you had to play against the computer. So making video games from home that necessarily required a second player wasn't convenient, we needed to find the possibility to play at home. Therefore, even video games that were not single player before now include the possibility of playing single player at home. Among these, there is one in particular that I want to point out, because earlier I mentioned 'Gun Fight'. 'Gun Fight' was the first video game, I told you, with a microprocessor. The version is made, then it's not official, because at the time it wasn't common to take the rights, we're talking about the early 70s, so all this wasn't there yet. 'Pong' came out, everyone played 'Pong', the gunslingers game came out, everyone played the gunslingers game, the cars game came out, everyone played the cars game, they all copied each other. What happens with 'Gun Fight', which is released as the home version 'Outlaw', from Atari, which is the home version of 'Gun Fight', but where the possibility of single player is added. So, the opposing player has an intelligence to fight. How many kilobytes does this cartridge have? Two. It came out the following year, but it was still 2 KB. Who was this 'Outlaw' written by? From a very young programmer called David Crane, the one who later made 'Pitfall' a few years later.
Pitfall and not just Pitfall, later I will also mention another one. This is done by David Crane. In '78, then Alan Miller comes out, this comes out, 'Basketball' for Atari, which comes out in '78 and you can play basketball against an opponent. It was so successful that an Arcade version was also made. Contrary to what happened, that is, it wasn't the Arcade version that then made the domestic version, the domestic version that then made Atari create the Arcade version. The Arcade version comes out in black and white, because at the time Arcade video games were still all in black and white. So it wasn't in color, unlike the Atari which was already in color at home. Instead it comes out in black and white. But it has another important thing, in addition to the fact of inserting an artificial player, therefore a one player, to play against, for the first time a player is inserted... a black player. We have a white man and a black man. This happens in 1979, because the domestic version comes out in '78. The Arcade version was released in '79.
The character I mentioned before, that is, Mr. David Crane, as many of you know, or perhaps don't know, leaves Atari, because Atari didn't give much credit to good programmers, and therefore he leaves and founds Activision, the most famous and main software house of the time, which still exists today. He founded Activision, many good Atari programmers left Activision. And, in 1981, 'Tennis' was published which, for the first time, apart from the graphics we finally see a tennis court that looks like a tennis court, and then we play against an opponent who has a certain intelligence, so he responds, it seems like we are playing against another person. This happens in '81, in '81, we are no longer in '77, so the cartridge finally had... 2K, always 2K! We are still at this game with these graphics, with the player playing, all in 2K: 2048 bytes. Now, do it yourself! Joke!
And this search for the artificial player, for the insertion of artificial intelligence into the players, continues, it also continues in arcades, because arcades now all use microprocessors. The first Arcade video game remembered, very seminal, where an artificial player is inserted, who is an artificial person, who therefore has a certain importance, is this 'Karate Champ'.
It is also a fundamental video game because it is the one that gives birth to all the fighting video games from then on, which will be born from that moment. In 'Karate Champ,' you fought another character who had his basic intelligence. Then the algorithms weren't that complex. Where do we start to see an increase in this complexity, i.e. finally see something that starts to amaze us a little? So, not just algorithms that simulate players, apart from chess, which as I said, was moving forward, there the chess playing algorithms became very complex.
That Mr. David Crane that I mentioned before, who is rightly praised as a game designer of video game programming and video game invention, after the huge success of Pitfall, Pitfall 2... planetarium, made Ghostbusters for the Commodore 64, also by Activision. That was also an incredible success, at the time, all over the world. Still within Activision, he begins to think about something that could revolutionize this world of gaming. He invents a video game where he finally creates the artificial person with whom we speak and communicate, taking from Eliza's algorithms that already existed. Eliza (chatterbot) was written in the 1960s, it simulated speeches, the algorithms that are still the basis of today's chatbots, of Natural language translators, many things that we use, of the ChatGPTs that we use today, obviously not with these databases, not with this complexity.
He creates this game, 'Little Computer People'. We discovered that a person lives in your computer, and you have to play with this person. He created this video game, that is, a video game, and we are talking about '85, the first version was made for the Commodore 64, then it was produced for all the computers of the time. In '85 'Little Computer People' was released, that is, a small person who is inside your computer. You turned on the game and a figure of a person was created randomly, according to certain algorithms, which was different for each copy of 'Little Computer People'. A seed was inserted inside the floppy which created a single person made in a certain way who had a series of characteristics different from floppy to floppy.
And when you launched it the first time, the first time it was written on the floppy disk it no longer changed, and it interacted with you. You had to… it was like… like a Tamagotchi and you played with it, you interacted with it. He had to eat, you played with him, he had to play the piano, he could play the piano, he could go, you could play checkers, you played different games. He went around his little house and you interacted with this artificial character, with these Little Computer People.
Even this video game, at the time, those who lived through that period remember it, because it was a huge success, even if we in Italy less so, because we in Italy, in that period, were in the period of piracy and therefore this only worked well if you got the original. Nobody here had the original. So it was very successful in America, in Japan, in England, but not here, because here the pirated copy arrived and it didn't work, and so we didn't understand why it didn't work.
Oh well, it was our dark ages. I have already told you this in another conference. Unfortunately we missed a few things in the 80s, around that time. However, Little Computer People by David Crane is from '85. Probably the first time we really see the realization of the artificial person inside the computer, that is, create an algorithm actually tries to emulate. Because, in reality, in the evolution of video gaming, I mention someone like John Madden Football or strategy games like Herzog Zwei or Dune II, they were very successful.
They are managerial, strategic games, where behind the scenes they tried to emulate a strategy, for example, even in John Madden Football, and in other sports video games, it was often... in that case, John Madden himself had collaborated to create algorithms to ensure that opposing players had strategies, like the ones he used as a football coach.
In Herzog Zwei and Dune II, however, there were algorithms that complexly simulated the interaction of other players when you played single player. The other players had to have a strategy, they had to have algorithms that were not simple like those of the games I mentioned before, something more complex. These algorithms begin to use formal tools of artificial intelligence algorithms. Herzog Zwei it is the first time, for example, that finite state automata are used to define military strategies within the programming, which had to compete against the main player. It was published in '89 by Techno Soft and came out on the Sega Mega Drive. While Dune II, Westwood Studios, always finite state automata, was released in 1992 for PC, Amiga, Atari ST, etc...
However, the problem is that these intelligences did not actually behave. The problem with these intelligences is that, being encoded within the game, they cheat! That is, they know things that you as a player cannot know, like the location of the whole map, things, etc.
They see the cards, they see the cards, basically there is this, so they are intelligences, but algorithms that don't really imitate the player as if he were a player, but they imitate what the player does, but in reality they cheat because they know things that you don't know. So then there were the difficulty levels, you could turn up, where the AI cheated more or cheated less than the things it knew. The first time an attempt was made to recreate the intelligence of players who don't know things, therefore who act independently of what is, is in another famous seminal video game, Doom.
In Doom, however, since this intelligence structure of the enemies was still basic, something happens that those who have played Doom often remember: it happens that sometimes the enemies start fighting each other and kill themselves, because they mistake the other self as the player. So they kill each other. This is an error introduced by the fact that we were trying to create enemies, internally, who didn't cheat, but who were enemies like you were inside the video game itself. Okay?
Oh well, I had arrived here at two in the morning, so I didn't go further to do the slides, even if in reality we are still going forward. But, for example, I can mention 'Age of Empires' from '97, 'Civilization 5' from 2010, which actually use the algorithmic structures of video games. And then currently they have them in contemporary video games, the first 'Half-Life' in '98 uses a pathfinding algorithm to find, in short, the paths to take. Each enemy entity had its own pathfinding algorithms to go around inside the video game, as if you were… as if it were a player actually playing inside the video game.
Oh well, there's a lot to go on, but it's already gone too far. Oh well, in '96 there was 'Creatures', and, well... in 'Halo: Combat Evolved' in 2001, it was the first time that a technology called 'behavior tree' was used, which is an algorithm invented precisely for this reason, following Halo in 2001, Halo: Combat Evolved which then made history in the video game world. This algorithm is indeed an AI algorithm, it is precisely an artificial intelligence algorithm which was then replicated, improved and used in many video games, in the shooter 'F.E.A.R.' from 2005, is really busy. It is also used to generate behavior that is sensitive to the context of the video game itself. So, for the first time in a game, the behaviors of enemy players behave depending on what is happening in the game environment. If I shoot from there, he reacts. The algorithm reacts because it sees the shot, but not because it knows that I had shot, but because it saw it, okay? As if I were the real player who is seeing the shot himself.
Alright, come on, I've told you everything. However, we have more or less arrived at the contemporary age. I hope you were interested, that I said some things you liked. Don't forget to follow the Computer Archeology channel, where I say these things and many other stupid things. So, both the audio podcast and the video channel on YouTube, where there is also Davide Gatti who you see back there and many other people.
And see you next time! Thank you all.