📎 Webclip
Bad ideas that aren't bad anymore
The author says many ideas that used to seem risky are working better now because AI has changed the costs behind them. The old instincts still matter, but they need to be tested against current conditions instead of accepted automatically.
Reading notes#
- Code review can be replaced by a workflow where Claude drafts the PR, the ticket owner reviews it, and CI plus Greptile provide extra checks.
- Splitting a team across separate streams of work no longer seems as risky when a collaborator can quickly bring someone up to speed.
- Deep expertise in one technology matters less when a collaborator can answer framework questions quickly; judgment and fundamentals matter more.
- Rewriting from scratch, building instead of buying, building the same feature three ways, and introducing an unfamiliar language or tool all look less obviously bad when AI cuts the time and cost involved.
- The main question now is what cost made a practice a bad idea, and whether that cost still exists.
- The author thinks leaders will need to hold their experience more loosely and let teams prove them wrong more often.
