Actually we did start with Godot - but it was too much too fast and he was overwhelmed.
He mentions this in his blog post. So we ultimately stepped back and I wanted him to learn to be resourceful. I give him a series of small changes (eg: write a function in any language that displays "hello world" on the screen) and he delivered those results in JS (he already knew HTML/CSS at this point). The rest comes as a natural progression.
I guess in this case, the best and pragmatic option was to use what you know.
Godot is designed for beginners, but also those that don't like to get their hands too dirty with code. Also the web export is likely inferior to what he can do with JS directly.
With that in mind I'd recommend he sticks with JS. If he's interested in doing more advanced games he should look into Three.js, or Pixi, Babylon, Phaser and other engines (and eventually even just raw WebGL and using shaders), all within the browser. There's also the web audio API which can be used to synthesise sounds and music from scratch and then there's WASM which can be used as a compile target for languages like C and C++
JS itself can be used as a procedural, functional or OOP language so while it's easy to learn it has lots to master.
Finally, if he does want to move outside the browser, there might be better choices than Godot, like Raylib or Defold or even more professional engines like Unity or Unreal.
Not saying it's a bad choice, just that where he is now has a lot to offer, and there might be better alternatives depending on his skills and aims.
I was originally interested in Godot because it had a web export. I feel like building on the web helps with shareability (eg: this post!)
But the "open web" seems to give limitless potential. Canvas, network APIs, lots of interesting libraries to build on, etc. I wasn't familiar with the ability to synthesize sounds - that would be interesting to look into.
I'm curious to know why you think Godot's web exports are too poor to be useful even to a 9 year old? I've published multiple HTML5 games on itch.io using Godot, including some with P2P multiplayer functionality. Apart from complete inoperability on Safari, I haven't run into many issues?
Because he's building native JS games already. And they likely work in Safari.
It would be a step backwards, unless he wants to publish for desktop/mobile as well.
I even found Godot's desktop export to be less performant than things I've built natively for the browser when lots of onscreen objects are involved. The only way I found to get around that was to use Godot's C++ extensions or modules, which defeats the object of an engine like Godot.
Unity, Defold and Unreal proved unsuitable for other reasons, and the most viable platform for building games for desktop and mobile after working with JS and WebGL turned out to be Raylib. I can only share my experiences, everyone should do what I did and shop around, but if you're happy working in the browser, there is no reason to move away from native technologies.
Hey, keep on bein' an awesome dad, dude! Hobbies that can double as highly profitable careers later on are the best kinda hobbies (speaking from experience). Nothin' quite beats the feeling of getting paid to do somethin' you actually enjoy doing, or the feeling of being able to create things you want or need using your own skills and hands. ;)
Software is pretty cool from that perspective. Creative, intellectually stimulating & pays well. It helps that software powers everything now.
If he does choose to pursue this in the future, he'll have a strong start. There's still a question of what changes will come to the industry as a result of AI.
Actually we did start with Godot - but it was too much too fast and he was overwhelmed.
He mentions this in his blog post. So we ultimately stepped back and I wanted him to learn to be resourceful. I give him a series of small changes (eg: write a function in any language that displays "hello world" on the screen) and he delivered those results in JS (he already knew HTML/CSS at this point). The rest comes as a natural progression.
I guess in this case, the best and pragmatic option was to use what you know.
In the future, we'll try Godot again :)