Thanks to Trystan for reminding me of the blog. Here is a screenshot from yesterday's work. This is just test code, to make sure that I can navigate between shapes reliably, it is not the finished game! But I post it to prove that I was working on it last night, before seeing Trystan's message.
I don't post much about the game, because after 27 years I think endless promises don't carry much weight it's better to actually finish the damn thing. :) But this is a very big week for TEDAgame, so this is a good time for an update.
On Monday (11th March 2024) I began to spend all my free time on the game. This is a big milestone. Before now, I kept getting sidetracked on other projects. But recent developments in AI research mean my other projects are pointless unless I can get the game done. So now, no matter how urgent another project is, the game has to come first.
To clarify, in 2022 I published Jack Kirby's History of the Future. Back in 1976, Jack Kirby predicted Judgement Day for 2026. A few months after I published the book, Chat-GPT burst onto the scene. Kirby wrote about people with spaceships and gigantic robotic suits, gathering all the world's data, and said that 2026 is when they decide whether to (a) turn us all into gods, or (b) destroy the world. I think AGI will fulfil that prediction, in the precise year that Kirby predicted. So I was busy writing a sequel, provisionally titled "I Told You So." However, last week I had an epiphany. I was on a social media site, where an intelligent person was making a long argument for something. Hundreds of people replies, all agreeing that he made an excellent point. But here is the thing: he was one hundred per cent wrong in everything he said. So I was about to reply, but I realised, "what is the point?" I could spend all day writing a careful analysis of every claim he made, and my reply would then be number 932 on an old thread and nobody would see it. And even if they did, nobody has the time to think through such long and complicated stuff. Even worse, a large proportion of the replies were either paid trolls or bots. This was a hot political topic where struggling people in "developing nations" such as India were paid a few cents per post to reply with approved talking points. And increasingly, some of the replies are by automated bots. This is an example of the "Dark Forest" theory of the Internet. The public space is broken, and being open there can be dangerous. A few weeks ago I posted on the same topic, and the next day a bot found my Facebook account, trawled back through ten years of posts, and got me auto-banned for something somebody else posted. (Ten years ago, somebody posted a stupid fart joke on my Facebook wall. It included a drawing of a bare bottom. I ignored it. So ten years later - this year - I was auto-banned for posting nudity!) The public Internet is now the wild west. Only combat matters. I am mildly autistic: I cannot win in that space by posting careful arguments. For me, such posts are not just hugely time-consuming, but either irrelevant or dangerous. The only way I can make a difference is to get a following FIRST, and THEN leverage my social position. So the more I want to write on other topics, the more I must focus on getting the game done, and making it count.
OK, back to TEDAgame. This is where I am (all dates from memory):
1992: I first played Zak McKracken. Amazing!!!! explore the whole world, and outer space, and all the most amazing ideas, in a fun box. I eagerly awaited the next game like it.
1997: after 5 years it was clear that nobody would make that kind of game, so I decided to do it. Started making plans on paper. My goal: all of time and space, unlimited stories.
1997-2000: Delays due to having a young family, full-time job, and running my local church.
2000: took a course in Pascal and C++, started to code the game in Borland C++
2001: realised that this would take me 50 years, so looked for a ready-made game engine that was flexible enough to make the unlimited game I wanted. Settled on SLUDGE (a fan version of the Lucasarts SCUMM engine)
2001-2003: sometime around here, people told me that I needed better graphics. I had a sketchy style for fast creation, but they said the graphics had to be beautiful instead. I listened to them and scrapped all my quick art and created slow art instead. BIG MISTAKE. In hindsight I should have stuck with the quick art. Gameplay is more important than art style. By improving the art I was sacrificing gameplay.
2003: Nobody had made a game that big using SLUDGE, so I discovered a memory leak that nobody else had found. (Memory leak = the program does not release memory perfectly after use, so if the game gets bigger and bigger then it uses more and more RAM memory, becoming slower and slower until it is unusable.) So I switched to the safest, most tested engine I could find: AGS (Adventure Game Studio). It could launch games from inside other games, so I planned to have an unlimited game in that way. I released my first (and biggest) test game: Les Miserables. People liked it.
2003-2007: Released 4 sequels. The second game was too experimental, and people who loved the first one hated that one. Also, I discovered that, just as with SLUDGE, nobody had used AGS for such a big game that launched other big games, and I found another memory leak: it was not possible to chain as many games as I wanted.
2008: After the second memory leak debacle, I began to create a Javascript game engine. It worked fine with Chrome, Firefox, Opera, etc., but - you guessed it - Internet Explorer had its own memory leaks. Internet Explorer also forces you to write different code, doubling the work. Most of people still used Internet Explorer, so I had to abandon the Javascript game. This was very demoralising. I cannot describe how much I hate Internet Explorer.
2012-2016: By this point I was so far from my dream of an endless game that I lost all enthusiasm for the project. I gave it up for a couple of years. I am a bit vague about what I did game-wise in this period, I think I only gave it up completely for 2 years. I probably tinkered with different ideas for the other 2 years.
2016: I next created a much simpler game engine: basically "choose your own adventure:" style online picture books. I created a dozen different simple games this way, Those are the ones that you can still see (I think) on TEDAgame.com
2017: Microsoft stopped developing Internet Explorer in 2016. HOORAY! Other browsers were much better by this point. I realised that I could make the game I originally planned, using Javascript.
2017-2022: the new engine got bigger and more complicated. IIRC I created two different game engine (i.e. the first one had problems). I also spent much of this time creating over 5,000 graphics. From late 2020 I got sidetracked writing two books about Jack Kirby (see above).
2022: AI art burst onto the scene in August. I realised that I could create much better art, much faster, this way: e.g. spend a day training it to create a nice background scene, then automatically create 50-100 variations. And it compressed very nicely (typically 60kb per background). So I spent almost a year creating 20,000 backgrounds, plus objects, animated people, etc.
2023: abandoned using AI art. It was great for stuff you have seen a thousand times before (e.g. forests, streets, alien worlds), but it was useless for anything creative. E.g. it took me days to get a half-decent picture of ancient Babylon from the air. I needed thousands of pictures like that, as well as all the coding, so that was impossible. I was also becoming nervous about bandwidth: even at just 60k per background, the typical user would cost me hundreds of megs of bandwidth.
Good news: by this time, CSS was so advanced that I could create my original plan of an automatically generated world, without any pre-rendered art. So I began a new game engine. It went very well at first. it was becoming a struggle by September or so. I hoped that the next generation of AI would help me with the coding, so I took a few months off to work on my updated book about the future. This book is turning into a big project: lots of original research, fully illustrated.
2024: that brings us to March: this very week. As noted, I realised that the game has to come first. I also realised that AI will not help with coding until we basically have AGI. AI coding is like AI art: it can help with familiar stuff, but is no good for anything creative. But I had a good think about the problems that we slowing down my coding and found a better way to do it. This is yet another new start, but it should be much simpler than before. (It also allows for better navigation.) Simpler is always better.
The bottom line: everything (income, books, and things I am passionate about) now depends on getting this game done. So it WILL be finished. I should have a playable game by the end of this year. But I will not release it into the wild until I am happy with it: probably next year some time.
Finally, just out of curiosity, I had another go at telling AI to create Babylon from the air. This is a definite improvement on before. But it still has multiple problems (everything is wrong when you look closely), so even without bandwidth issues, my new game engine will be better, at least from a gameplay point of view.
No comments:
Post a Comment