If they do that, will they not put Adobe at direct conflict with Anthropic? If a court rules that the resulting project is in violation of Adobe’s IP, will that not set a terrible precedent that makes LLMs unusable for software development across the board?

score 1 · canonical

There’s a lot I can relate to in this post, but I think there’s a crucial point that often gets missed in analyses like this. As software developers, we analyze this situation as a case of us vs them (i.e. management), but in reality, the management also has as much to worry about in all this as we do.

If LLMs can completely replace software developers, then many of the companies we work for are also at an existential risk. In particular, companies that provide software services and products. Perhaps, before long, a lucky vibe coder with the right setup will pull the arm and the slop-machine-3000-ultra will spit out a great competitor. Software shops are equally afraid of losing their moats, just like software developers do as individuals.

And even if management in a particular software shop isn’t afraid of losing their moat themselves, they will have a hard time finding investment, since the investors have to worry whether they’d be burning money on something that next year’s LLMs can just be made to write from a single prompt. That single prompt can just be “make a competitor to company X”. I actually think that as of 2026, this freezing effect on investment (combined with the brutal macroeconomic environment) might be the main factor in determining the bargaining power and the everyday experience of a software developer.

score 1 · canonical

Fair enough, I actually had the sentiment in mind when I was writing down the time range. I heard that argument quite often in relation to the rise of bytecode and interpreted languages as alternatives to C/C++ for application development in the 90s. Though perhaps people had been saying this before that too.

score 1 · canonical

This reminds me of how in the era of exponential CPU speed ups (1995-2015?), people thought it wasn’t worth optimizing your programs, because the next generation of CPUs would close the gap for you. Maybe people now think that we don’t need innovations anymore, because the next generation of LLMs will either do it for us anyway, or they will make progressively better use of the existing technologies.

score 1 · canonical

Obligatory watch on the subject: https://www.youtube.com/watch?v=hIZxTQP1ifo (Type classes vs. the World, Edward Kmett)

score 1 · canonical

Query planner optimizations aren’t just convenient, but they also allow you to write very reusable code around your database. Either in the database or in your own code, it allows you to assemble subqueries that mean something and then you can consume them in a multitude of ways.

Say you have 4 ways to assemble expressions like this and then 4 ways to extract results from them. In your approach, you’d have to write 16 separate programs to operate over your tables and indexes explicitly, and in all likelihood there would be very little reuse between these programs, but with the query planner you can nicely decompose your domain with subqueries, you can see the contents of those subqueries by running them against your DB and then reuse them to let the query planner write non-trivial programs for you. It’s all correct by construction modulo performance.

With your 16 programs that are conceptually related but implementation-wise unrelated, you’d also have to write a lot of tests to make sure they’re doing something coherent among themselves.

score 6 · canonical