It's been one of those weeks where it was difficult to remember what I was up to at the beginning of it without reading my journal and my mood seemed to dip. Still there were a fair amount of things that I did.
Read
I read through the original Dungeons and Dragons Players Manual and found it to be a neat historical artifact. It was clear that tabletop roleplaying was a novel idea at the time and the rulebook slowly introduced players to it in the form of a Choose Your Own Adventure format in the book itself.
The Choose Your Own Adventure in itself was pretty interesting in what it introduced in terms of mechanics and monster types. It was pretty challenging for a single character and you would have had to know what you were doing to get through all of it. As it was I took 3 goes at it and only died in one of them. It was apparent that there was a lot of conditional logic that had to be squeezed into the numbered sections for which one you turn to and that combat with only one player character is kind of dull.
The rest of the book had the character classes which were limited and demi-human races like elves and dwarves. The demi-human races didn't even get to choose classes being all the same. Elves seemed to be very overpowered having good combat and spellcasting ability.
Speaking of spellcasters, magic users seemed to be at their weakest with tiny hitpoint amounts, only one spell per day and hardly any offensive magic to choose from. The Players Manual being quite short only had first and second level spells listed in it. The version of the rules I was reading had clearly been written after the panic of DnD being called occult as it was very clear that these weren't real spells needing some kind of ritual.
Lastly there was some gentle introductions to the dynamics of playing in a group, sharing loot and having a dungeon master to organise the adventure. Seeing how simple the book was it was thus intriguing the minutia it went into concerning carry weight of loot with individual coins being tracked as how encumbered you were. More than 100 coins or any kind of armour would result in you being slowed down.
The tone of the rules were much more conversational and you got a feeling that this had come from a group of players who had decided to make their own rules up for the games they were playing. It kind of put me in mind of the articles that you would get in early (circa 1999) Warhammer White Dwarfs. Very much hobbyists.
I also read quite a bit about AI this week getting through several months worth of weekly reviews of AI research in Import AI by Jack Clark, one of the cofounders of Anthropic. It is neat that among the practical reporting on research papers there is also some speculative fiction.
Watched
I watched quite a few AI related things this week. One being the interview with Jacob Coxon, the AI researcher who recently resigned from Anthropic, on the Daily Show. I don't normally watch the Daily Show so it took a bit to get used to its blend of humour but the interview was good with Coxon explaining the problems with the labs in terms of safety and that actually he does believe in the potential of AI to bring about an utopian state.
Indeed it was very interesting to see the difference in perspective, perhaps brought on by experience, between Jon Stewart's reaction to the idea of there not being jobs in the future and Coxon arguing that there would still be meaning. I expect a future where you aren't forced to work to survive would be rather popular but going off of the comments people don't trust AI corporations to deliver that.
There were also interviews with another researcher who resigned from Open AI, David Robinson, on the Ezra Klein show and Yuval Noah Harari on The Rest is Politics. The latter was quite sobering in terms of the choices facing Britain about how dependent we could be on future, American, AI models.
Lastly there were a few AI generated music videos being shared, AI is a normal technology? being one of the better ones. What is significant here is that it was made by a language model calling other tools rather than a generative video model, all those animations were written in code. It's a sign of how good AI models are getting at at being given a complex task and just going away and doing it.
Played
I completed the rest of the RealIndustry demo, getting into more advanced processing facilities like electrolysis and chemical plants. The game was just getting into the oil age and there was a wealth of new ore types in industrial scale mines. Understandably the content in the demo cuts off before many of the later intermediate resources get used.
There were some interesting new mechanics added like automation done without colonists, pumps as renewable resource inputs to recipes that don't get consumed and less powerful but consistent energy sources like wind. I have to say the complexity was kind of getting to me as aluminium and stainless steel were being added as I found it difficult to keep track of which part of the production process took place in which building. Also the difference between the likes of tin, nickel and zinc in terms of which gets used in which recipe.
The end of the demo had some underwhelming real world puzzles which amounted to plugging questions into google to get the answers. That the company is actually controlled by amoral AIs feels very played out and honestly your AI advisor's lines could have been cut completely as they are so generic.
Still the game idea is solid and just as a show of the support to the sole developer, who is a computer science student in Ukraine, I bought the full version. I haven't booted it up and I'm a bit averse to the idea of having to go through all the tech tiers again. I will be looking forward to learning about all the new resources though. Also the soundtrack is rather nice and freely available here on youtube.
Programmed
Playing RealIndustry really got me thinking about how I might make a game like it myself and what I'd get if I took my simple project and asked an AI to improve it. To that end I've been brainstorming about what I'd want to have in my own realistic factory building game and wrote some basic code for an incremental clicker version of it.
Of course RealIndustry is already kind of an incremental clicker game but I thought it could hew more closely to the likes of Cookie Clicker by having a very simplified model of automation where you build multiple buildings to help you gather resources more quickly. I also thought it would be nice to compress everything onto one screen instead of having separate screens for each building.
I didn't get very far in it other than making a list of 108 resources and programming an interface to go through the production loop of collecting resources for and then with wooden and stone tools. Still it was nice to do a bit of Javascript programming again and do my usual Program/Control/Simulation/Display structure with proper classes for once.
Other
Speaking of programming, the Ludum Dare programming competition is going to be happening the weekend after next. I was a bit concerned that it wouldn't happen due to the website not being functional for a long time and it has taken a while for the theme voting to commence. it's up now though and I've thrown my own theme ideas into the ring. Either 'big numbers' or 'the factory must grow'. I certainly know what kind of game I'd make with such themes.
I'll be away in Newcastle from tomorrow so I won't end up writing about the usual kind of things next week. Still there will be places to visit, shows to see and books to read.
No comments:
Post a Comment