The ORMs-versus-SQL debate never dies
A 2014 post arguing that ORMs teach bad habits and developers should just learn SQL resurfaced and drew fresh discussion. The comment that landed hardest drew a direct parallel to today: in 2014, people were indignant that they should have to learn SQL when ORMs existed; in 2026, people are indignant that they should have to learn anything at all when LLMs exist. The through-line is a recurring cultural fight about abstraction versus understanding.
The technical discussion underneath is more specific. Commenters note that Postgres has moved far beyond what the 2014 post described, with JSONB and other features making the landscape substantially different. Others advocate going even further down the stack, arguing for learning LMDB or RocksDB directly.
The meta-pattern: every generation of tooling provokes the same argument. The people who insist on understanding the layer below are consistently right about correctness and performance, and consistently lose the argument about adoption.
So what?
For founders building data-intensive products, this thread is a reminder that abstractions are fine until they are not, and the teams that understand the layer beneath their ORM will debug faster and scale further. Hiring for SQL fluency still matters.