The world of video games, like any other, is full of seemingly inconsequential moments which cause ripples that become shockwaves that can determine not just the course of a single company, but the entire industry.
With Nintendo, there are numerous moments like this; Gunpei Yokoi’s prototype grabbing toy becoming the company’s first million-selling gadget in the form of the ‘Ultra Hand’ and turning Nintendo from humble playing-card maker to wildly successful toy firm virtually overnight is one example. Another is the last-gasp success of Donkey Kong, a game created by the then-unknown Shigeru Miyamoto with the intention of clearing stock of unsold Radar Scope arcade cabinets in North America. Had neither of these events transpired, then Nintendo’s future would have been very, very different.
Nintendo’s number two guy, Hiroshi Imanishi, came out and told everybody, ‘No, you guys do not know anything about how to make Nintendo games, and you will not make Nintendo games.’ I mean Square and Enix, are you kidding me?
However, for every significant historical event, there are perhaps just as many which go largely unnoticed, as do the people who were instrumental in making them happen. Allan Scarff isn’t a name you’ll see mentioned in the same breath as Miyamoto, Toru Iwatani or Satoru Iwata, but this unassuming Englishman nonetheless had a significant role to play in the world of video games – and in the history of Nintendo in particular.
Scarff may not be a household name, but he was unwittingly instrumental in opening up the third-party licencing business for Nintendo’s Famicom in Japan at a time when the company had locked down access to just a handful of external developers and publishers. The burgeoning Japanese personal computer game market was home to some significant companies, but they weren’t making games for consoles like the Famicom, which represented a huge potential audience thanks to its incredible success in Japan following its launch in 1983.
“The five biggest publishers of [personal] computer games went to Nintendo,” Bullet-Proof Software founder Henk Rogers explained to John Szczepaniak for his seminal Untold History of Japanese Game Developers series. “That would be [us], Square, Falcom, Enix, and I think… T&E Software. So five of us, the presidents of these companies, all went to Nintendo to ask to become Nintendo publishers. And Nintendo’s number two guy, Hiroshi Imanishi, came out and told everybody, ‘No, you guys do not know anything about how to make Nintendo games, and you will not make Nintendo games.’ I mean Square and Enix, are you kidding me? These end up becoming the biggest publishers in the Nintendo business!”
While Nintendo had signed licensing agreements with arcade companies like Capcom, Konami and Namco which permitted them to create games for its Famicom home system, it was sceptical that home computer companies would understand how to make successful console titles, which resulted in the Famicom getting a selection of games that arguably didn’t accurately reflect the trends in the Japanese games industry – RPGs, for example, were pretty much the sole preserve of home computers like the PC-88 and Sharp X1 at this time, and it wasn’t until Enix was permitted to bring Dragon Quest to the system in 1986 that Japanese gamers became seriously hooked on them.
Rogers would later become famous for being a major player in the effort to secure Tetris for the Game Boy – Bullet-Proof Software would handle the development of what is arguably the most famous version of the Russian puzzler – but at this time, he was mainly focused on creating games for Japanese computers, having built his business on Black Oynx, an RPG he created which would influence many Japanese developers. Rogers knew that he had to get his games onto the Famicom, and discovered a means of obtaining an audience with Nintendo’s infamously tetchy president, Hiroshi Yamauchi. “I found an article in a magazine that said Hiroshi Yamauchi played Go. And I’d actually gotten a hold of a copy of an igo [a computerised version of Go played against an AI opponent] game written by Allan Scarff in England, for the Commodore 64.”
I remember playing all of Allan’s Go game versions. I’m proud to say I was the chief tester for new versions of programs he made. He used to say I could break almost anything
For those who aren’t aware, Go was invented in China more than 2,500 years ago and takes place on a 19×19 grid. The aim is simple – to surround a larger total area of the board with your own stones than your opponent – but despite the relatively basic setup, Go is an incredibly complex game in terms of strategy and potential moves. Creating a virtual version of Go that can react and behave like a human player would be a stern challenge even to a modern-day programmer working on their own, but back in the mid-’80s, it would have been a Herculean task – yet Scarff had achieved it on 8-bit hardware.
Scarff’s game was called Microgo1, and was released for both the C64 and BBC Micro home computer systems. “I remember playing all of Allan’s Go game versions,” Scarff’s son, Christian, tells Nintendo Life. “I’m proud to say I was the chief tester for new versions of programs he made. He used to say I could break almost anything, so I would test the user interface to destruction.” A huge fan of Go and a keen amateur Mycologist, Scarff made sacrifices for his growing family, as Christian recalls. “Growing up, Allan could be a little distant, but you have to remember he had foregone secure employment to concentrate on the business. With three kids to feed, this was a major preoccupation.” Even so, he found time to share his intense affection for Go with his offspring – although Christian freely admits that sibling rivalry was just as much to blame for him taking up the game. “My guiding light in Go was very simple: beat my brother. I’m not sure any advice was taken without this aim in mind.”
Scarff’s Microgo1 – which was followed by Microgo2 – proved to Rogers that it was possible to create a competent version of Go on an 8-bit system. Rogers – a sharp Go player himself, having been introduced to the game by his father, just like Christian Scarff – took a solo trip to Nintendo’s HQ in Kyoto following Imanishi’s rebuttal. He knew that Yamauchi’s affection for Go could be the key to getting permission to create games for the Famicom, as he recounted to John Szczepaniak: “I went to Kyoto and met Mr. Yamauchi, and said, ‘Look, I can make an igo game for your Famicom.’ And he looked at me and said, ‘I can’t give you any programmers.’ And I replied ‘I don’t need programmers, I need money.’ So he says, ‘How much?’ I thought of the biggest number I could think of at the time and said ‘$300,000’, and he shook my hand. That was it! I was making our first Famicom game.”
With funding secured, Rogers had to then track down Scarff himself in order to make his Famicom Go title a reality. “After I had made the deal I got to tracking down Allan. I said, ‘You gotta come to Japan!’ He lived with my family for the nine months that it took to port his stuff over. This was a miracle of programming. Allan had written an igo game based on cellular automata, and it was just brilliant what he did! There was no memory, so he couldn’t do any pattern recognition. So the game’s code was all like a ‘trying to survive’ kind of thing.”
He played it, or tried to, because he didn’t actually know how to use the controller on the Famicom. I could not believe it, like he’d never touched it before! This was the first game which he took any interest in
Despite Scarff’s efforts and nine month’s of work, Yamauchi was not impressed when Rogers returned to Kyoto with the finished product. “He played it, or tried to, because he didn’t actually know how to use the controller on the Famicom,” Rogers told Szczepaniak. “I could not believe it, like he’d never touched it before! This was the first game which he took any interest in. So he tried to play, and he gave up, and handed the controller to his underling sitting next to him. Then he said, pointing to the screen, I want to go there, I want to go there, and there. He played one game and said, ‘No, it’s not strong enough for Nintendo.’ I said, ‘Mr Yamauchi, this is the strongest igo game that is ever going to happen on this machine. This is an 8-bit machine, and it’s a miracle it can play the game at all!'” While Go was usually played on a 19×19 grid, Bullet-Proof Software’s game was limited to a 9×9 grid as the hardware simply wasn’t up to the immense task of computing all of the possible permutations of each move (according to Wikipedia, the number of legal board positions in Go has been calculated to be vastly greater even than the number of atoms in the known universe).
Ironically, the simple fact that Yamauchi was such a tough customer to please is what opened up the Famicom to third-party publishers. Nintendo’s president was adamant that the game was not up to the standards expected of first-party Nintendo software and the company would therefore not be publishing it, despite the investment of $300,000. “I said, ‘Let me publish it. It’s strong enough for my company.'” says Rogers. “And I could see him calculating, and he said, ‘What about the money?’ I said, ‘I will pay you back 100 yen, or a dollar, for every copy that I sell. Until I pay you back.’ And then he shook my hand again, ‘Deal!’ That’s how I became a Nintendo publisher.” The title was eventually released on both the Famicom Disk System and Famicom as Igo: Kyuu Roban Taikyoku in 1987, with Bullet-Proof Software handling the publishing duties.
While Igo: Kyuu Roban Taikyoku only sold 150,000 units – and therefore didn’t earn enough to recoup Nintendo’s initial investment – it transformed Rogers’ business dramatically. The numbers of games sold on the Famicom dwarfed what the company had been used to on home computers; Bullet-Proof’s version of Tetris for the Famicom, for example, sold 2 million units in Japan alone. “That eclipsed anything that I ever did on personal computers in Japan,” Rogers later told Szczepaniak. Other personal computer firms, such as Enix and Square, would experience even more good fortune when they were accepted as Famicom developers.
Speaking of Tetris on the Famicom, this title was just one part of a much wider story involving the domestic rights to Alexey Pajitnov’s famous block-dropping puzzle title. When Elektronorgtechnica – the Soviet organization in charge of the import and export of computer software – got wind that UK publisher Mirrorsoft and its American subsidiary Spectrum HoloByte had claimed to own the licence for the game and were sub-licensing it to other companies without permission, it triggered a chain of events which would ultimately lead Rogers to the negotiating table in Moscow. He presented Elektronorgtechnica – or Elorg, for short – with his company’s version of Tetris for the Famicom, and then offered to pay Elorg an advance on sales generated by royalties (something it had yet to receive from either Mirrorsoft, Spectrum HoloByte or Andromeda Software’s Robert Stein, the businessman who had ultimately brokered the contested deal in the first place).
I think it was a very important time in his life. He mentioned the Rogers family a lot, and I think was grateful for their help while he was over there
Rogers worked with Nintendo in order to secure the console rights to Tetris, and the game would launch as a Game Boy pack-in title in North America and Europe. The rest, as they say, is history – but it’s a history in which Scarff’s name is often unfairly omitted or ignored. Had Rogers never seen Scarff’s igo game on the C64, it’s unlikely he would have possessed the confidence to march into Nintendo’s office and assure Yamauchi that he could create a Go title for the Famicom – and without that event, he would never had created a version of Tetris for the Famicom, and we may never have had Tetris on the Game Boy. In fact, without Scarff’s igo on the Famicom, third-party publishing on the system could well have remained out of reach for companies like Square and Enix for several more years. No Square and Enix on the Famicom means no Dragon Quest or Final Fantasy – at least not when they originally launched, anyway – and while it’s possible to make (or break) hypothetical guesses until the cows come home, it’s not unreasonable to suggest that Bullet-Proof’s Famicom title triggered a domino effect which has had numerous unforeseen consequences over the decades.
Following his work on Igo: Kyuu Roban Taikyoku in Japan, Scarff returned to his native England. “He disappeared from the world after that,” Rogers told Szczepaniak. “I kept in touch with igo programmers, but he never resurfaced.” Christian Scarff admits that he was too young to recall this period in detail, but his father later opened up about his experiences in Japan. “I think it was a very important time in his life,” Christian tells us. “He mentioned the Rogers family a lot, and I think was grateful for their help while he was over there.” While Scarff retreated from the world of video game development, he wasn’t totally idle and instead committed himself to solving the conundrum which had driven him his entire life: the topic of artificial intelligence.
His wife Elizabeth, in conversation with John Szczepaniak, proudly labelled him as “a pioneer in the field of artificial intelligence” who, when asked which fields AI could be applied to, would reply “Everything”. In an interview with a local newspaper in July 1992 – two years after Scarff and his family had relocated to Newcastle upon Tyne in the UK – he said: “What I’m trying to do is work out how humans play Go. It is a very simple game to play, but we don’t know exactly how we do the things we take for granted.”
“Allan was so enthusiastic about the potential of AI,” his son Christian tells us. “He would bend anyone’s ear who would listen. Of course, he would grab and talk at length to anyone who was in any field related to AI; it turns out for Allan, almost every field was touched by AI. However, I found myself in a lucky position when visiting between term times at University that Allan used me as a sounding board for his ideas. I guess it was not so much my response that was of interest, more that he could practice making it sound intelligible to the layperson, so to speak.”
Scarff’s desire to unlock the mysteries of AI almost went hand-in-hand with his love of Go, a game he was first exposed to in 1969 by a student friend by the name of Phil Bristow. During his life he actively participated in several UK-based Go Clubs, eventually reaching third ‘dan’ level. After moving to Newcastle 1992, Scarff continued to research Go theory, eventually publishing his ‘Global Connectivity Strategy’ in 2000. He continued to research and investigate AI and how it could be used to create a program which could teach itself how to play Go, as well as working the specification for an ‘Acolyte Neural Net System’. These documents are Scarff’s lasting contribution to the fields of AI and Go.
One particularly memorable dream he recalled was when a man stood in front of him, holding out a closed hand. The man then opened his hand and revealed a precious object of dazzling beauty which illuminated the whole scene and made Allan gasp
Christian Scarff, who has achieved UK dan level at Go and regularly participates in Go tournaments, feels that the ancient game was just one of the many ways his father tried to unravel the mysteries of true AI. “Go is deceptively simple, yet we complicate it all the time,” he says. “Often mistakes are due to over-complicating things. The answer normally is the simple one. I’m still philosophically working Go out – plus, it’s fun. The beauty about Go is that you can’t just out-calculate the human with a massive tree search algorithm. What are humans doing that allowed them to ‘guess’ moves through ‘intuition’? So yes, I think Allan saw Go as a way into what the human brain was doing.”
“Allan often dreamt about the research he was doing on artificial intelligence,” Elizabeth Scarff told John Szczepaniak in 2018. “One particularly memorable dream he recalled was when a man stood in front of him, holding out a closed hand. The man then opened his hand and revealed a precious object of dazzling beauty which illuminated the whole scene and made Allan gasp.”
Tragically, Scarff died from cancer on 9th December 2011 at the age of 65, and despite the incredible achievements of his life, he never shook the feeling that he hadn’t quite done everything he set out to. “I think he was aware of being a pioneer in the field,” Christian Scarff tells us. “But underneath, I sensed a sadness that the full potential of AI was not being realised. Of course, after his death, some of his milder predictions have come to fruition.”
Scarff’s wife Elizabeth, speaking to John Szczepaniak in 2018, perhaps sums it up best. “In his last years, Allan spoke of trying to overcome an obstacle which blocked his progress to artificial intelligence. He described it as like a mountain which he was trying to climb or skirt around. In the end, his illness terminated his efforts. Like Moses, he could see the Promised Land, but wasn’t allowed to enter it.”