Why Tiny Programming Tricks Punch Above Their Weight

A large share of day-to-day engineering productivity comes not from deep expertise but from a stash of small, self-contained pieces of knowledge — the kind that require almost no supporting theory to apply. Keleher argues that knowing a language feature exists, recognizing that an unexplained TCP delay likely traces back to Nagle’s algorithm, or remembering the right git incantation to escape a jam quietly compounds into real speed over time. His running example: you don’t need to know Python to get value from python3 -m http.server.

The post is largely a curated list of these nuggets, weighted toward the shell and databases. Highlights include fuzzy history search via fzf (or replacing shell history entirely with atuin’s SQLite-backed store), running SELECT with no FROM to probe database behavior, EXPLAIN ANALYZE for query tuning, and the git “pickaxe” (git log -S) for finding every commit that touched a string. He also nudges readers toward modern defaults — ripgrep over grep, shell globstar over find, reusing a Node https.Agent to cut request latency, and turning on zsh’s autocompletion, which ships off by default.

The broader point is cultural: inside a company, the highest-leverage knowledge is often this same small-nugget form — which data source debugs which problem, who owns which system, what command does a rolling restart. Keleher recommends institutionalizing the sharing of it, citing his own habit of posting one trick a day to an engineering channel. One per day, he notes, was the right cadence to stay useful without overwhelming people, and even a 9-out-of-10 hit rate makes the tenth tip worth it.