rossdm

premise

"let AI handle low-stakes work, humans handle critical work" isn't always an accurate mental model for framing suitability or ROI of coding tasks from non-engineers.

hypothesis on factors affecting non-engineer outcomes

• Contributor comprehension
  • Explains why; reasons about call sites and completeness; owns mistakes
• Organizational capture
  • Corrections become ADRs / agent rules / reusable components
• Process & scope discipline
• Handed off as a described PR whose claim matches the diff; assumptions validated; re-implementations re-check inherited scope

rossdm

background

the (known) origins of the term "comprehension debt"

Jeremy Twei coined the perfect term for this: comprehension debt. It’s certainly tempting to just move on when the LLM one-shotted something that seems to work. This is the insidious part. The agent doesn’t get tired. It will sprint through implementation after implementation with unwavering confidence. The code looks plausible. The tests pass (or seem to). You’re under pressure to ship. You move on.

symptoms

• volume & verbosity inflation; PRs grow in size, making diffs harder to review & reason about.

rossdm

Keybase proof

I hereby claim:

• I am rossdm on github.
• I am rossdm (https://keybase.io/rossdm) on keybase.
• I have a public key whose fingerprint is 8D79 38E0 A969 D0FF 2E26 2685 85DC E351 D63C EF7D

To claim this, I am signing this object:

rossdm

/**
 * column layout
 */

background: #f06;
background: linear-gradient(45deg, #f06, yellow);
min-height: 100%;


  .container { display:block;width:100%; /* or whatever you are using */ }