Marco Gomez

How I created an OpenSource 3D game with @threejs and MML while live streaming the development of the entire game in a way that it could be watched inside the game itself 🤓 Time stamps: 00:00:00 - Intro 00:00:18 - The game engine 00:02:06 - The MML documents 00:04:06 - multi-player sync 00:07:58 - memory game example 00:09:20 - client-MML communication 00:10:50 - reusability and composability 00:12:06 - MML tags and attributes 00:13:32 - interoperability 00:14:11 - the platformer game 00:16:15 - more MML tags and attributes 00:18:16 - let’s write a game 00:24:52 - 3d objects with primitives 00:33:41 - creating interactive blocks 00:41:29 - creating click event handlers 00:44:30 - creating animation functions 00:51:25 - the game step function 00:59:22 - checking states for victory or draw 01:03:42 - creating display labels 01:15:16 - the reset game function 01:17:29 - the finished game 01:18:50 - final considerations If you liked the project, please also follow these legends: @MarcusLongmuir : the father of MML; @MorgeseSacha and @deej_io : my buddies at the MML Team; @RJFWhite and @HermanNarula : creating the craziest possible shit you can imagine (and some you can't even imagine yet) at @Improbableio

I honestly don't know how much their compute limits can choke inference and negatively affect model results when usage increases massively (like what happened to Sol), but it's genuinely fascinating (quite annoying sometimes, I admit 😅) how much each model's quality fluctuates over time.

@TheCodeTherapy @MrCollison So fable is probably working a lot better now since less people are using it 🤣 at this time

Played a bit with SOL 5.6 ultra with all of these limits removed. Asked it to implement a simple fix, it created a 24,000 line change diff with fully re-written a bunch of systems for no reason 🤯 What is the 5.6 equivalent of the 5.5 xhigh that doesn’t go insane?

Yeah, I experienced results very similar to what you're describing with Fable, too. The entire renderer is mine and was handwritten code, but whenever I ask for changes or a refactor, I feel like I end up investing more time in course-correcting than I would if I executed the changes myself. I believe the models are getting better (Fable was the first one I've seen write somewhat decent WGSL), but I feel we're still a few iterations away from reliable results. I'm very curious to see where we'll be by the end of this year.

It’s great if I give it a detailed, focused spec. I’ve already architected the higher level code at that point If I set it off running without any guidance (no report back before actually doing something + iterate) it’s just a slop cannon. All the things you mentioned + really dumb wgsl code (and im not that good at shader code myself) Fable not so much. It gets the mission and follows the pattern 90% of the time on some quite heavy stuff

I literally gave up using AI to deal with Rust + WebGPU. Too many per-frame allocations, unnecessary heap traffic, avoidable clone() usage, avoidable to_string(), format!, collect(), join(), unnecessary usage of Box, Rc, Arc, Mutex, and RwLock all over the place. Basically zero care for performance. Are you having a good experience with your tech stack using Sol?

That’s super interesting! Do you think it’s anything to do with the code you work on evolving? I find most models, with exception of fable, do better when code is less evolved I literally had this today - Sol ripped out a bunch of features and rewrote a hunk of good rust wgpu stuff I’d hand rolled - it had basically nothing to do with the task at hand A couple days back when the codebase was less mature (this thing came together fast AF) it found and addressed a shed load of little inefficiencies and did a great job. I don’t trust it to do the same today haha. It’s kinda like AI music mastering - it doesn’t know when to stop additive work

"How do you validate if it’s going to work?" By shipping. "How do you know if people will buy it to not?" Shipping. "How do you validate product market fit?" Shipping. "How do you validate if a feature is worth building?" Getting real user feedback (yes, real user, that you can obtain after shipping) "How do you validate a design?" ...

Validation is a mirage. “How do you validate if it’s going to work?” “How do you know if people will buy it to not?” “How do you validate product market fit?” “How do you validate if a feature is worth building?” “How do you validate a design?” You can’t. You can’t. You can’t. You can’t. You can’t. I mean you can, but not in spirit of the questions being asked. What people are asking about is certainty ahead of time. But time doesn’t start when you start working on something, or when you have a piece of the whole ready. It starts when the whole thing hits the market. How do you know if what you’re doing is right while you’re doing it? You can’t be. You can only have a hunch, a feeling, a belief. And if the only way to tell if you’ve completely missed the mark is to ask other people and wait for them to tell you, then you’re likely too far lost from the start. If you make products, you better have a sense of where you’re heading without having to ask for directions. There’s really only one real way to get as close to certain as possible. That’s to build the actual thing and make it actually available for anyone to try, use, and buy. Real usage on real things on real days during the course of real work is the only way to validate anything. And even then, it’s barely validation since there are so many other variables at play. Timing, marketing, pricing, messaging, etc. Truth is, you don’t know, you won’t know, you’ll never know until you know and reflect back on something real. And the best way to find out, is to believe in it, make it, and put it out there. You do your best, you promote it the best you can, you prepare yourself the best way you know how. And then you literally cross your fingers. I’m not kidding. You can’t validate something that doesn’t exist. You can’t validate an idea. You can’t validate someone’s guess. You can’t validate an abstraction. You can’t validate a sketch, or a wireframe, or an MVP that isn’t the actual product. When I hear MVP, I don’t think Minimum Viable Product. I think Minimum Viable Pie. The food kind. A slice of pie is all you need to evaluate the whole pie. It’s homogenous. But that’s not how products work. Products are a collection of interwoven parts, one dependent on another, one leading to another, one integrating with another. You can’t take a slice a product, ask people how they like it, and deduce they’ll like the rest of the product once you’ve completed it. All you learn is that they like or don’t like the slice you gave them. If you want to see if something works, make it. The whole thing. The simplest version of the whole thing – that’s what version 1.0 is supposed to be. But make that, put it out there, and learn. If you want answers, you have to ask the question, and the question is: Market, what do you think of this completed version 1.0 of our product? Don’t mistake an impression of a piece of your product as a proxy for the whole truth. When you give someone a slice of something that isn’t homogenous, you’re asking them to guess. You can’t base certainty on that. That said, there’s one common way to uncertainty: That’s to ask one more person their opinion. It’s easy to think the more opinions you have, the more certain you’ll be, but in practice it’s quite the opposite. If you ever want to be less sure of yourself, less confident in the outcome, just ask someone else what they think. It works every time.

@XorDev God bless you and your family, Jon. This, too, shall pass. I hope life gets back to feeling normal quickly.

It's been a hectic year. My brother's brain surgery, buying a house, moving, wedding and now my sister's condition. I miss coding

Permit me, Sir, to venture an observation which I trust you will forgive. I cannot but remark that the distinction of your bearing, the elegance of your person, and the quiet refinement of your manner bespeak a gentleman of that exceedingly rare and impeccable stamp which England alone has, for generations past, been reputed to produce.

@N8Programs @mrdoob @trq212 Do you guys have any plans in that direction? That would be very interesting. I understand Anthropic's "big fish" is the B2B, but I can see a lot of end-user conversion and ecosystem commitment coming from something like this. I mean, there's a pitch for it.

@TheCodeTherapy @mrdoob You said it better than I did. There is already Claude for Open Source: claude.com/contact-sales/… But I think it would make sense to have a Fable grant for the mrdoobs and such.

This is where I think a special Fable grants program makes sense - there are a class of OSS contributors (@mrdoob being an exemplar) who are so important to the broader community that them having sustained access to Fable at reasonable prices is a worthy investment.

@CharlieDarli @TimSweeneyEpic @theo Who claimed LLMs are crap? I'm just making it very clear that trusting the quality of the output produced by an LLM and trusting the quality of the output produced by a compiler are two completely different processes and two completely different "reasons to trust".

@TheCodeTherapy @TimSweeneyEpic @theo I’m sure assembly programmers thought compilers where crap because they added 3 more calls in some scenarios where the programmer knew how to not need them.

How much better do the models have to get before you'll stop reading the code?

That's not a great reference. LLMs can't be compared to compilers, technically or philosophically. Compilers enforce semantics and remove degrees of freedom, no matter how high-level the language they compile is. LLMs do the opposite. They add degrees of freedom and produce outputs with no enforceable invariants. We can only compare trust in two completely different processes if both are, at least in their nature, designed to be deterministic and reproducible.

@theo Moving from assembly language to compilers, there was a 24 month window where it mattered.

@wookash_podcast First came the XBox. Then XBox 360 (full circle). Then Xbox One X... X-Box-One-X... X-B-O-X... XBOX

Wait. DirectX box. DirectXbox. Direct Xbox. Xbox. whoa

This is a bad analogy. With movie streaming, latency barely matters, and even if overcompression reduces image quality, the experience is still functionally the same. In cloud gaming, latency fundamentally changes the experience because every input depends on a round-trip. They're also different problems in nature. Netflix solves a distribution problem, while cloud gaming solves a real-time computation problem. They have different origins, technical constraints, and trade-offs for the providers.

@Jonathan_Blow do you use netflix/hulu or still buy everything on bluray?

you’ll get mad at me for saying this…but cloud gaming is so obviously more economically efficient than physical hardware I think it’s going to be the default soon. your home console / pc is idle 90%+ of the day. meanwhile, data centers targets what, 5%, maybe at worst 10% idle. every second a cloud gamer isn’t gaming, that hardware is being used for someone else, training, etc. I think there should be a new measurement, something like cost-per effective FLOP hour that takes into account the TCO + effective utilization. If a gamer spends $500 on a GPU, uses it for 3 years, but it’s only fully active ~5% of that period…the cost-per relative FLOP hour is crazy high! Meanwhile, a $50,000 datacenter GPU might have a *LOWER* cost-per FLOP hour just because the effective utilization is 90+%.

The Anti-AI Andy is such a boring animal (I'm not talking about people who don't care about using AI, but about the militant doomers). I don't think they're boring because their takes on the negative aspects of AI are necessarily wrong (I have mine too), but because they show a clear lack of capacity and/or effort to adapt to the world's changes. Stop whining and adapt. Delusional hype guy and Anti-AI Andy are NOT the only paths moving forward. @mitchellh gives a sensible example of how you can keep your care for craftsmanship while establishing good boundaries and a balance between human responsibilities and AI-assisted productivity.

Pixel streaming is doable, and that has been proven for a long time now, but the claim that it will become standard has a very weird overlap with any data analysis over what's happening in the hardware industry. Compute resources are being scaled back for the consumer market because it's more profitable for companies like Micron to serve data centers because of AI. For the exact same reason, it's more profitable for the data centers to invest all their computing power in AI too, rather than in pixel streaming. Your claim that gaming on your own computer may become the least viable option for games is not necessarily wrong (personal computers may become prohibitively expensive in the short to medium term because hardware manufacturers can't catch up), but I wouldn't bet that pixel streaming will grow in the foreseeable future. In times where supply is far behind demand, companies tend to put all eggs in one basket (the most profitable one).

Chickenshit leadership. A live-service game like Destiny 2 requires high effort and balls to maintain. It's high stakes. They bet on Marathon instead. What a dumb decision. Bungie lost its ability to read the room ages ago.

@ggsimm I can't even explain how relatable this sounds to me.

the absolute heartbreak I get when I see a big company develop something I was working on as a novel idea, so defeated rn

@XorDev That's a huge step. Congratulations Jon! I'm happy for you!

Bought a house with shaders today

I wish they didn't worry about articulating it at all. Hate is a waste of time. Verbalizing hate is even worse. Life is really hard by itself. We don't need to help it. Grime's original post doesn't hate on anything. It just describes the world as it is, without imposing opinions on what we should do about it. My comment to it was just "yes, it's really hard for the entry-level folks, and here's how I reckon they're making it worse for themselves".

@TheCodeTherapy @GRIMECRAFT You cant really expect everyone to be articulate about why they hate something.probably for the best posts like OP get talked about more. maybe everyone can get more or less on the same page. I mean I only just recently figured out how to articulate why Im not a fan of AI art.

For Game Art students: Outsourcing replaced your entry level jobs for the past decade. AI has wiped out 1/3 of jobs across the board. Everyone fighting AI is unemployable and on a list for layoffs now. Top tier talent are struggling to find work and they are your competition

That's not what people are doing. They're not "acknowledging what exactly the issue is". From what I can see, most people are repeating "AI bad! AI evil! AI enemy" like a broken record. That won't make AI go away, nor solve anyone's problems. Did I personally like my craft more before AI? Yes, absolutely. Would repeating that to myself and to my friends every day help any of us? Not at all, it would only make things worse. The fight is always the same: asking yourself what the best you can do is in your craft with the tools you have today. This is not a post-AI question.

@TheCodeTherapy @GRIMECRAFT I think its good to acknowledge what exactly is the issue and let others know what they're actually fighting. I see no sense in letting everybody fail by letting them think everything is sweet when its not. That at the very least should be made known to anyone taking this path.

I'm not saying that the world is easy and fair, but the exact opposite. But I recognize two opportunities within everyone's control that can help. 1) When we accelerate (which is what AI does), it gets easier to see that things get harder over time (my initial claim), because now it's noticeable within a lifespan, and not across generations. And constantly reassessing how hard something is gets in the way of focusing on solutions to specific problems. Tunnel vision helps. 2) I believe people don't appreciate how rare the level of dedication I mentioned really is, and how infallible it is. While you say "1000s of people are willing to do it", that's not what my personal experience allowed me to see. I can attribute all the milestones I reached in my life to the level of dedication I invested, while surrounded by others who aren't willing to. I'm not judging cause there isn't a "right" or "wrong" lifestyle. But my lifestyle served me well, and I'm just any other guy, so I believe it can serve others too.

@TheCodeTherapy @GRIMECRAFT And not even because theyre less skilled but the outsourcing and AI legitimately did take their place when they themselves could have done the job just as well as the programs and ones that made it through.