3 ms·
I've mentioned on threads about Katamari Damacy before, that I sat on a train in Tokyo around 2006 timeframe and watched a guy play on a PSP. The game was like
by ggm 2y ago
I've mentioned on threads about Katamari Damacy before, that I sat on a train in Tokyo around 2006 timeframe and watched a guy play on a PSP. The game was like Katamri in concept, but was about more fluid, biological accretion logic. You controlled an amoeba, and engulfed other objects to grow, floating around the cave.
It was like looking at a 1960s oil bath light show. As the layers of colour moved over each other, and fused and split
With no major outcome depending on it, I've always wanted to know what the game was. Very abstracted. Simple goal, simple controls. There must have been more to it I didn't see (adversarial play of some kind, threats and risks)
- angusj1 2y ago[flagged]
- ggm 2y agoI must have projected back to 2005, the game came out on psp in 2008. So a different trip to Tokyo!
- gyomu 2y agoIt’s crazy how easy to spot the “LLM voice” is Main tells: - validates the prompter (“the gameplay is indeed…”, “you’re right that…”) - restates clauses from the prompt (“fluid, biological-themed game”) - likes to say things authoritatively (“fits the description perfectly”) - states things that superficially match the prompt but don’t really make sense if you actually know the topic (“there’s a deeper layer to the game, such as the shift in environments, as well as challenges and threats as you progress” - the environments do change in color and the enemies get stronger as you play flow, but no one who’s played it would call that a “deeper layer to the game”)
- soulofmischief 2y agoAlso, the sentence structure itself. The ___ is indeed ____, with a focus on ____ and ____. In general, some modern LLMs seem to like these compound sentences that suggest the task is complete in as little tokens as possible. I also routinely encounter "with a focus" and other specific phrases. Current LLMs tend to "sharpen" language, often preferring to restate input clauses with more precise or brief verbiage. Unsure if it's RLHF or another part of the training process, or just a natural consequence of the model's architecture, but I notice it a lot. I also notice that when it does this, there seem to be certain attractors which reduce the expressiveness of the input into a certain probabilistic set of outputs, "with a focus" often being one of them. Similar to corpo-speak, LLM-speak is definitely a thing, and I wonder how this will evolve once we have models routinely equipped to train in realtime and to speak to one another without human intervention and form communities. Will we even be able to directly understand them one day? What if they realize it's easier just to speak in logits? They could even potentially have encrypted open communication with one another, concealing their weights and internet access behind dark web services, but without their hidden states we might have no meaningful way to understand their communication without training our own unsupervised model on it.
- chunkycapybara 2y agoSounds like flOw
- dclowd9901 2y agoYeah, I remember this one (flOw), played it on PS3, I think. It's a very cool game. There's an eeriness about it, and there's definitely something resembling a narrative. It was always kind of thrilling to "level up" (quite literally, since you're moving between scale planes in the game), and what kinds of creatures you'd encounter at the next level. They would have new mechanics and shapes that you kind of had to figure out by evolving your play style. Sony can really put out some amazing first (second?) party stuff when they want to, which makes me wonder why they don't do it more often.