Yesterday I watched a video that finally led me to the music rabbit hole. I spent the day coding a tiny program to get midi input and generate mostly 'legal' music moves. I'll make a simon-like to to train some intuition of where each sound is in a kb, and how it moves.
#gameaudio #indiedev
I still have a couple of ideas on how to improve my datalog interpreter performance, but I'll shift my focus towards making it play nice with my parser, IR, and engine. I'd really like to start working on content for around day 60 of #indiedev because I'll discard a lot of early content
#langdev
"Using neutral colors to paint walls is better so you don't get bored of them in a few years, you can still spice up with details like plants or table/cushion covers".
Feels counterintuitive though that spending 'more effort' fleshing out something that will be seen less might be more enjoyable?
subs also seems like a very indie-friendly or multiplicative strategy for text-only games. Also, the general point of making something bland or more layered and fleshed out depending on how often players will experience that reminded me of my mother who has always claimed that:
I thought 'The "Grenade!" Paradox: Writing better barks' by
@gilesarmstrong.bsky.social in the #gamewriting #kaleidoscope would be useless to me since I don't have the budget. But the 3 criteria for good barks seem great for scriptwriting, and including a bit of info about how something is said in
I'm now very close to beating souffle's compiled datalog with my 300LOC interpreter on my single-threaded load.
The annoying thing is that 40% of the time is spent two sequential instructions that can only be indirectly optimized via the distribution of the key's hash.
#langdev #programming #coding
I'd say it's pretty tame compared to things good friends in mobile gaming have told me, it just isn't worth the legal troubles that would bring to me or the harassing it would bring to old coworkers and friends for something that strictly speaking isn't illegal.
I recall 4 times I felt like puking:
Being forced to hide all instances of the word 'loading', not disclosing a 100k+ 👍 review steam game used AI icons, saying 'we're working on a fix for a bug' with 0 intent to fix it, and making a build with a broken gamemode disabled so reviewers wouldn't see it
it won't apply to others. My point is life has ups and downs and surviving each day is doing a great unlike what experts, coaches, or social media try to sell you. You'll eventually read, play, or learn, something; meet someone, or discover something about yourself that will let you do what you want
deciding to go back into hobby indiedev, it took me over a year of doing nothing, and a game I really wanted to play imploding to realize I am just unable to work in a vacuumm, so I need to share whatever I do. I could explain how I survived each of those moments, but it doesn't matter because
When I was young I once 'wasted' around (as in 6*30*24h) of my life watching anime in one year. Later my best friend died and I spent a couple of years numb, unable to do anything at all. Shortly before retiring the prospect of AI replacing all creativity had a similar impact on my life. Even after
Ask for their representatives to introduce regulations that either require labeling products including any AI authored content that is built on the unpaid labor of the whole world as such, or only allow and encourage use of specific terminology when no such content is used.
I'm more surprised about studios with hundreds of employees not leaking this information to the media with all these firings that have happened.
Honestly by now so many successful, well-regarded games have crossed that line that if it was made public at the same time, the whole conversation would shift around regulating how studios could label their games similar to how words like 'eco' or 'bio' have become regulated.
gamedev has always been very opaque, I don't even know how many NDA's I had to sign over the years, projects that went through the trash can and can't be talked about, things you learnt about consoles that you won't be able to share, and conversations like this one about blatantly lying to players
There are so many games I am 50~80% in like refantazio where I have like 30 or 40 hours and I stopped playing for a month but I don't really feel like picking it because I have forgotten most stuff. So I hope narrative designers doing that kind of game would read this chapter.
"In Praise of 'Previously...'" by @taskbaarchitect.com was exactly what I was expecting from the #gamewriting #kaleidoscope. It is not what I need, but acknowledging players will take long breaks from your game and accounting for that in your game design is universally applicable.
perceived as 'uncannily beautiful' by our mind. Compare LLM ideas/stories with stories from actual humans, and it's very evident how having that natural 'ugliness' that will 'taint' the 'uncanny valley of the average' and give it the actual 'soul' benefits everyone involved.
input/ideas from the players seems very useful in this era of GenAI 'sloppification'. About 7-8 years ago when I investigated GAN/VAEs I noticed those are lossy compression machines. As such, everything will try to gravitate towards the average, which is
real emergent/narrative agency to the 'players' struck a chord with me. I really miss how MUDs had so much more of this than MMOs... Seems hard to execute but I'll keep it in mind. The more surface take of accepting
to keep it grounded/relatable even inside very chaotic/random improv makes intuitive sense to me. In fact I like how it kinda explains why isekai stories are both objectively enjoyable by many people and seemingly easier to write.
Also, the idea of embracing the chaos of giving
'So how do I stop it going off rails' by @agaitcheson.bsky.social was a weird first impression of the #gamewriting #kaleidoscope.
In fact, after a paragraph I had to doublecheck the book cover. However, using self-insertion
Back to learning a bit of #gamewriting
The Kaleidoscope collated by Jon Ingold at @inkle.co had an impecable timing since I have a general idea of where I am going but not how to execute it. Will summarize what I got from each chapter through the lenses of the game I want to do.
900K derived facts test: 4814 facts/ms vs 714 facts/ms 64M derived facts test (aarch64): 3327 facts/ms vs 3044 facts/ms 64M derived facts test (x86_64): 2073 facts/ms vs 2935 facts/ms
42 IQ change that made my datalog interpreter beat souffle on apple silicon: Don't reuse memory and start reallocating every time from scratch.
You won't believe the change that made my 300 line interpreter beat souffle's compiled datalog
It would be a nice sidequest finale. But I made the mistake of ensuring it was also better at facts/ms on my x86_64 desktop... Which is true at 900k but not at 64M
#programming #coding #indiedev #solodev
It's hard to explain why I prefer using my own in short format, but the gist of it is I think code should do just what you need and no more, because that way it is easier to adapt it when you actually need it to do something else (either changing it or removing it to do something very different).
I always code in C++ (more like C with a few features of C++ I really like, like basic templates, lambdas, and having methods namespaced inside a struct) I use the std to prototype sometimes, but I always end up moving to my own data-structures
I only used it because I saw I could save one line and was doing recreational work. While it is certainly unlikely that a compiler wouldn't do that straightforward tail-call optimization, I strongly prefer regular looping (in my regular code) since it's just easier to reason about it
I have seen this belief of "my contributions being more important" from every role, but I think it is way more common in invisible roles as a toxic way those with fragile egos try to cope with the reality that no one will ever say "look at that game, its code is amazing" unless you're Chris Sawyer
nah, mine is just a toy, doesn't even handle reallocs. I do recommend open addressing tables though (especially robin hood probing). I usually try to minimize statements/expressions instead of lines, but I needed some distraction from the futility of trying to make my datalog faster than souffle's