It is not just hating on OOP, it is where the ideas came from, what we could have had, and what he thinks we should have.
Strap in, it’s a long one, but it was really enjoyable:
1 Like
I watched this video on stream and… I’m kind of embarrassed to admit it, but it took me about 8.5 hours. You can see me yap about it in the VOD. Also, I took some notes.
After taking some time to reflect on it, I found a few issues with the talk:
- It’s purportedly about the history of OOP, but Casey Muratori spends quite a bit of time opining on some of the negative aspects of OOP. I largely agree with him, but I felt that this detracted from the talk by making it less focused.
- He makes a claim that I think is false (I am unsure) where he says “And we could have had [discriminated unions] from day one if someone had just believed that this was good”, implying that the idea of discriminated unions was not developed for a very long time afterwards because no one liked it. However, ML was introduced a few years later, I believe it had such a feature, and it went on to influence functional programming languages like OCaml and Haskell. To be fair, I could also read this as him exaggerating for effect, since much of the talk is about what could have been, if OOP had developed a bit differently.
1 Like
My impression is that he is talking about himself and the industries he has worked, aka the places where C and C-like languages were taught and used; that also likely applies to most of the people in the room and listening to the talk. Just lamenting that they could have had it since it was there to be adopted but it didn’t happen.
1 Like
Watching your VOD was a lot of fun too! The constant tangents and dives into ideas is just great.
One that really caught my interest was the non-text-buffer code… I feel like if I can find some time I’d love to prototype something like that. Why have IDEs have to process code and find commonalities between variables and such when it could just be encoded in the code file itself and the editor just represents it, no need to scan the code all the time. People run LSPs all the time just to get this kind of AST info, but it could just be how the code is represented from the get-go.
Kinda the same theme when talking about playing to computer’s strengths with stuff like graphite and such. We are using suboptimal interaction and storage formats that are like traditional methods instead of doing more. Code is just text, vectors and images are just end-result maths and bitmaps. We could be doing so much more! Some CAD programs figured it out to some extent and they have parameterized and parametric modeling, time travel, constraints, etc.
We gotta build a cooler and more powerful future!
1 Like
Yeah! I agree. Dion Systems built such a prototype a few years ago that has lived rent free in my head ever since I first witnessed it: Gallery | Dion Systems
1 Like
It is beautiful 
I also poked around on the site a bit more and the team page shows Ryan Fleury is/was a member! He also wrote one of the blog posts about them releasing metadesk which is some kind of metaprogramming.
He did a talk at this same BSC2025 conference about the pains of code and debugging and the tool he’s working on: Rad Debugger (though it only runs in Windows right now)
I found it to be a drier watch, but still had some interesting parts.
Yeah, I found the talk to be really dry. I didn’t really enjoy it honestly 
It did, however, remind me of the Tomorrow Corporation Tech Demo, which I love very much and it also makes me so mad because it lives rent free in my head:
I would love to use a debugging tool like this. I could make it I suppose, but I want to be making games hahaha
1 Like