Home Creators Posts Import Register Favorites Logout

Content

i hadn't written a patron-only dev diary this week because i was finishing up the new WWT build at the time, but i have since written TWO blog posts, and wanted to share them here.

first: the great ren'py schism of 2026. the lead dev of ren'py has been using LLMs when working on the engine, which has alarmed a lot of game developers who use ren'py and don't want AI tainting the things they create. i don't like AI either, but there is nuance to pytom's decision, and i choose to respect it and just stay on an older version of ren'py that i feel like i can trust. npckc posted about this too, and we had been chatting about it a lot as we were seeing people get up in arms about it on social media (specifically bluesky, a site/app run by a very pro-AI crowd, ironically!)

i love ren'py, and i hope people will read our posts and not harass pytom over this! can we please assume good faith and politely disagree?! i know it's a controversial topic, but i felt very strongly about needing to defend him, even as an anti-AI guy myself... i just don't want people to forget how much pytom has done for us up to now, i don't like seeing people get thrown in the garbage over something so relatively minor, yknow??? it harms the community for zero gains. ugh!!!

ok putting that aside... the second thing is a more lighthearted post that i would normally write as a dev diary, about timeline games! since it's only tangentially related to my own stuff i decided to just make it a public post right away. i think i want to do this more often, though i still need to carry over some of my now-public posts here to the blog on my site... that's a problem for future nami tho... (we're all counting on you, future nami.........)

i'm planning on finally opening asks this weekend, and looking forward to drawing more timber-related content. other than that, starry flowers steam page will be coming soon.

that's all for today's updates~ thanks for reading!

Comments

TotemSP2

This... is Nami Bloggin'! *Goblin Bloggin' theme* ...anyone?

Yajoovya

Interesting stuff.

The Sunroses

i would never ever harass someone over this, but i do strongly disagree with pytom on LLMs being a "net positive" for coding. like i struggle to understand how it'd actually save time if you have to comb through all the code it generates to make sure it does what you want anyway. it feels like a placebo. and even if you know what you're doing it feels like it carries a lot of risks. like we do agree that clever autocomplete is a legitimate use of this, but there are plenty of tools to automate the tasks he's talking about that don't use LLMs or the like. most decent code editors and IDEs have find-and-replace, multiple selection, etc.

The Sunroses

i just don't think the state of the "industry" is one where we can even know the full impact of using these things. as far as i'm concerned, it's mostly snake oil.

NomnomNami

YEAH definitely not a net positive in my eyes either. quite detrimental in fact! what if companies just paid humans to do the work instead! i don't trust its output so i'd never use it myself... the placebo thing is so real, it just displaces the work you're actually doing to doublechecking a robot's work, which sucks...

The Sunroses

yeahh. like i don't think people are evil for falling for it but it IS a scam, and it makes me sad

Emull

the combination of timeline, choice selection and endings is exhilarating.

Tournesolia

The hundred line! Loved this game.

dhfshfdskfjdhs

Yeah i havent liked social media either. (*damaged)

Dream Fractal

GenAI is a bane for art - the text and images it creates are worse than garbage, for many many reasons. But for code generation, specifically, I think there's an argument to be made for a use case. LLMs are half-way decent at programming. It can get working software off the ground quick. The real issues with using it for say core Ren'Py is that with a community piece of software you want it to be (a) reliable, and (b) readable by other users. And LLMs are (a) unreliable and (b) sometimes generate gibberish looking code that even if it works it's hard to read, making it difficult for other users to work with that code going forward.