Ask HN: Why I rarely see game dev startup here? | Hacker News Reader

Ask HN: Why I rarely see game dev startup here?

Do investors despise game development companies, and do you have to go solo? If so, I would have expected to see at least one on Show HN. Or maybe I missed it—definitely not as popular as other apps.

Also, why haven’t LLM wrappers penetrated this sector? Is it because OpenAI and other “parasites” haven’t had the opportunity to access source code to build their wholesale theft, so they don’t have the code to instruct you to build games? Or maybe there are a lot, and I just missed them.

View on HN

Games are idiosyncratic, creative, and funded mostly by knowledgeable insiders funding devs with track records. The failure rate is mitigated this way. Plus the revenue doesn’t favor the VC system. First investors aren’t given such extreme advantages, and seed money for proofs isn’t necessary. A demo costs much less than an app. I’m in game dev and VC is the last place I’d seek capital. AI is the last place I’d look for creative help (though AI is involved in the intricacies of baking and texture refinement).

Can you tell me about how you use AI for baking and texture refinement?

I’m a beginner making small 3d webxr experiences, so tooling is a bit scarce outside of blender. Recently started making my own models and doing stuff like ao mapping and lightmapping. Editing texture maps using AI sounds helpful.

I’m not sure how these work, but AI accelerates the refinement of realistic textures. As an example, we just switched to Metahuman for human rig building.

https://www.metahuman.com/en-US

The devs building the rigs use real people, and Metahuman uses AI to actually increase their imperfections.

We then shift into Blender where we construct hair and clothing. The hair uses internal AI that accelerates the process of individuating hair behavior we then export into unreal where the hair responds to physics.

"why haven’t LLM wrappers penetrated this sector"

The vast majority of games are sold over Steam. It's so overwhelmingly dominant that it eclipses almost every other platform put together, with some exceptions for consoles and mobile. Even when it's targeted at these, Steam is such a powerful marketing/discovery channel for games.

When Steam says no AI, nobody does AI. It's not just no AI generated content, some argue it means no brainstorming dialogue via AI and so on. The game dev community selects for this. This means that alternative platforms like Itch tend to also lean towards anti-AI even if they could differentiate by allowing AI. Consoles tend to also build for PC and adopting AI would cost them too much.

Pro-AI people tend to move to other platforms like Play Store and App Store, and lately we're seeing more quality control ramp up on these.

Game developers are extremely rare because they can actually program. Most people writing software for a living cannot actually write software. It’s the difference between struggling at copy/pasting other people’s words into your own plagiarized essay versus writing an original novel or textbook.

Minigames are posted quite often https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que... I'm not sure if you are looking for game companies.

That’s a really good observation- I’ve noticed the same. I think it’s because game dev and startup culture overlap less than we expect.

Game dev usually leans more art and creativity than problem-solving or scalable business models.

Different mindset, funding path, and timelines. Though lately, game dev tools and AI pipelines are starting to bridge that gap.

How applicable are AI pipelines to game dev? A few months ago, I worked on an iPad game (Unity) and found that AI tools were pretty bad and unhelpful. The other dev didn't use AI at all

The new companies you see posted about there are not an even representation of all companies; it is skewed towards VC-funded, San-Francisco based SAAS and (now) LLM-wrapping companies in particular.

There is more to the business world than this! Game dev is one example.

For most other ideas you can build a quick MVP to validate or perhaps even just a landing page. Games take much longer to get off the ground.

I think the reason we don't see many LLM wrappers is because Games are technically complex, multi-disciplinary and ever changing.

One interesting company in this space I've seen is CreativeMode - generating Minecraft mods based on a text description. Pretty brilliant use of LLM coding actually. Found out about it when my kid asked me to pay for it. Doesn't look like they ran a Show HN / Launch HN (not one I can find) despite being a YC company.

i think the stuff posted here is mostly designed for devs as the end consumers

Reply on news.ycombinator.com