Zach Kehs: there is no limit to how bad code can get
On 6 September 2026, Simon Willison's blog carried a short quote from Zach Kehs's essay "There's No Limit to How Bad Code Can Get." Kehs's argument is a comparison: a building that keeps having floors and rooms added to it will eventually collapse under its own weight, so physical construction has a natural ceiling on how bad it can get before the structure fails outright. Software, he argues, faces no equivalent constraint. There is always room for the code to get worse: another layer of indirection can always be added, and performance can always be allowed to degrade further, without anything physically giving way to stop it. The post is a brief blockquote rather than a full write-up, and Willison adds no commentary of his own beyond citing the piece and its author.
Key facts
- Zach Kehs's essay "There's No Limit to How Bad Code Can Get" was quoted on Simon Willison's blog on 6 September 2026.
- Kehs's central comparison: buildings collapse if floors and rooms are added indefinitely, so physical construction has a natural limit on decay.
- Software has no equivalent limit, according to Kehs: there is always room for another layer of indirection or a further drop in performance.
- The blog post consists of the quote itself, with no additional commentary or examples from Willison.
Why it matters
The comparison names something engineers feel but rarely state outright: nothing physical forces software to stop degrading. A building's own weight eventually caps how much can be piled on; a codebase has no such mechanical feedback. Left unchecked, added indirection and slipping performance can accumulate indefinitely, because the code will keep running, just worse, rather than falling down.
Who it affects
Anyone responsible for a codebase that outlives its first few sprints: engineers doing day-to-day maintenance, tech leads deciding where to spend refactoring time, and managers weighing feature work against paying down complexity.
How to use it
The piece is an argument, not a tool or a technique. Its practical takeaway is that the limit on code quality has to be imposed deliberately, through review standards, refactoring time and architectural discipline, since physics will not impose one on its own the way it does for a building.
How solid is it
The claim is presented as an assertion rather than a demonstrated result. Neither Kehs's quoted passage nor Willison's post names a specific codebase, gives an example of the described decay, or explains the underlying mechanism beyond the building analogy.
Risks and caveats
No data, case study or concrete example accompanies the argument in the material quoted here, and no context is given for what prompted Kehs to write the piece or where the original essay was published. The idea should be read as a framing device for a familiar problem, not as an empirically supported finding.
“If you continue to add floors and rooms to a building forever, it will collapse. Software faces no such constraint. The code can always get worse. There can always be a new layer of indirection or a reduction in performance.”
— Zach Kehs, "There's No Limit to How Bad Code Can Get"