Obserfessor

Space enthusiast & reply-reply guy: pedantic is my brand. Go SpaceX! Retired HW Engineer. Often follow back. Please limit DMs to post followups

[Obserfessor @Obserfessor]

I am not even sure Cybertruck is a failure. Pragmatically, the CT is brilliant, akin to iMac. In an entrenched market (American trucks) it attempts to differentiate with a strong love/hate view, giving a higher probability of converting lovers—EV trucks were always going to be a tough sell, whose primary customers are rural communities.

[Obserfessor @Obserfessor]

@darrellprograms @wingod Yeah, that’s stupid. We are in a cycle of improvement and past low performing cells are an expected part of the process. Eventually, it will become stabilized. Most of this will probably end up going space longer-term anyway (ok, long term, but I think it’s real).

[Obserfessor @Obserfessor]

@AndrewatL5 @BenjaminDEKR @Robotbeat And in this context, I don’t think it’s wrong to be skeptical, but it is almost putting him on the hater page because he refuses acknowledge both sides of the skepticism.

[Obserfessor @Obserfessor]

@Robotbeat @BenjaminDEKR This guy is starting to sound like a rage bait troll.

[Obserfessor @Obserfessor]

@wingod @MollySOShea @lrocket I always admired SpaceX’s ethos of putting cost as king. I’m an engineer, but I understand that cost drives everything, and lowered costs allow the technology to push further and further with feedback effects.

[Obserfessor @Obserfessor]

This seems related to the observation booster10 made of a recent Falcon9 landing, but in the final approach the grid fins have no control authority and it’s very likely that it was at a peak angle when the system arbitrarily shut off. And if that’s what’s going on well then that’s not a big deal.

[Obserfessor @Obserfessor]

@BenjaminDEKR It was hard to predict the social pushback and the power requirements. Nobody knew that AI was going to turn into this monster that needed massive training to scale.

[Obserfessor @Obserfessor]

@aakznd @VladBastion Couldn’t pay you to invest…. I mean, isn’t getting money from your investments, one of the major reasons that people invest?

[Obserfessor @Obserfessor]

@booster_10 The grid fins have almost no impact on the final descent approach due to the low velocity.

[Obserfessor @Obserfessor]

@StevePratico @ChadBianco I can only imagine that he did it intentionally.

[Obserfessor @Obserfessor]

@starsailing11 I have been using JSON data structures in an SQL lite database, and am migrating to a relationship graph, to dramatically improve what narration can provide.

[Obserfessor @Obserfessor]

A major feature of the program is it’s world architect. At the beginning, you have the option of either importing from a bounded scenario seed file that you can create inside a chat LLM or have it generate in game. At the moment, I’m trying to use open source LLM’s on my Mac mini for gameplay, but really haven’t tested that much. Yes this is a very complex aspect of the game, but Iv’e spent more time troubleshooting the closure logic than the sophistication of the scenarios. There’s also a second mode of the world architect to create edge content during game play on the fly.

[Obserfessor @Obserfessor]

@starsailing11 I think without a major architecture change to how LLMs manage thinking context, that this problem is going to last quite a while longer.

[Obserfessor @Obserfessor]

@starsailing11 I looked closer at your link here. I’m not sure if this is going to be an issue, but my game will be able to tap into frontier models for the scenario creation portion of the game— at this stage, I have not really tried to create a truly intricate game scenario.

[Obserfessor @Obserfessor]

The one thing I would say about this whole experience it’s been a lot of fun. I don’t know if I’ll be successful with this game or not but for me it was mostly about learning to use agentic coding.

As far as the game itself goes, I was never planning on selling it, and just giving it away, but who knows. I guess if it’s super successful, I might tap into that.

[Obserfessor @Obserfessor]

@starsailing11 This has been my problem. I am more interested in and fairly good at acquiring skills, but I often don’t have anything I want to do with them. That being said I did finally have an idea and now am knee deep in, but honestly I am not sure it will actually work.

[Obserfessor @Obserfessor]

@starsailing11 Prose

[Obserfessor @Obserfessor]

Yeah, I figured I could use exactly that when I got to that point. But really I’m I’m only in the validating the concept part. It only works with one player doesn’t have much graphics, really a primitive chat interface. But yeah, it uses the Mac voice input and output for now. I know now that the concept will work, but the real problem is will it ever be smooth enough to feel like you’re really talking to a real human being, or will it really always be kind of brittle? Certainly directly talking to an LLM can achieve the real human being feel or close enough to it. But because all the pros has to go through a deterministic state controller, it’s not that simple.

[Obserfessor @Obserfessor]

@starsailing11 Honestly, I’ve never even used the X voice platform to talk with people before— not even sure how.

[Obserfessor @Obserfessor]

@starsailing11 If you wanna chat about it, let me know. I like talking to people about it, but there’s a lot more to it than I can put really into text. I thought about posting my experience, but that’s a big job in itself.