Defects in Knowledge Work

podcasts.apple.com

Defects in Knowledge Work

Podcast Episode · Process Debt · 10/03/2025 · 17m

If you run a factory, you can point to a bin of defects. Bent parts. Off-spec batches. The “whoops” pile. In knowledge work, where does that pile live?

In college I did two internships that couldn’t have been more different—one at a Merrill Lynch branch, one at a lollipop factory in Boise. At the factory, you could literally watch quality happen. Sugar heated, flavors added, the rope rolled and stamped around a stick. When timing slipped or mix ran light, the result was obvious: partial pops. Those went in big clear bags—defects you could see, weigh, and learn from.

The contrast has been stuck in my head as I look at how we work today. We move ideas, not sugar. We write emails, not wrappers. We sit in meetings where the “product” is a decision, a plan, a deck. When that product is off-spec—unclear, misaligned, misleading—there’s no bag by the door to catch it. It just… circulates.

Physical work forces clarity because physics is unforgiving. Knowledge work invites ambiguity because the medium is forgiving. Misspell a slide? It ships. Hold a meeting with no owner and fuzzy outcome? It repeats. In a factory, delay hardens candy; in the office, delay softens intent.

So how do we spot—and reduce—defects in knowledge work?

1. Define the spec before you start.In the factory, “spec” meant weight, shape, print. In your team, “spec” might be: audience, decision needed, and the one-page artifact that proves it. If you don’t know the use of the output, you don’t know what “good” is. No spec, no quality—just activity.

2. Make outputs visible and countable.If a meeting exists, what artifact exits? A decision log entry. A one-pager. A ticket moved. Tie time spent to something you can point at later. “We talked about it” is a defect hiding in polite clothing.

3. Reward accuracy over theater.Pretty decks can be persuasive—and wrong. If the most resourced ideas are the best-looking, not the best-tested, you’re routing talent to the wrong places. Track forecast vs. actual. Celebrate the product manager who predicts impact within 10% more than the one who dazzles in the room.

4. Prefer repeatable over one-off clever.One-time analyses are sometimes necessary; standing one-time work up as a process is how defects multiply. If you’re doing something more than twice, define the template, the owner, the trigger, and the exit condition. Don’t let one-offs become zombie rituals.

5. Shorten the hardening window.Candy hardens fast; that constraint drives flow. Create similar constraints: time-boxed work blocks, decisions due by end-of-meeting, async pre-reads with a commit timestamp. The point isn’t speed for its own sake—it’s giving ambiguity less time to spread.

6. Put the “defect bag” where everyone can see it.Maintain a visible backlog of rework: slides re-done after a review, meetings repeated to decide the same thing, reports generated that no one used. Treat them like scrap. Review the list weekly. Ask, “What spec was missing? What signal did we ignore?”

If this sounds basic, good. Quality always is. What’s tricky is that our culture often rewards the urgent over the important and the polished over the useful. In factories, waste hits the ledger immediately. In offices, waste hides in our calendars, our inboxes, and our shared drives—quietly compounding interest.

We just finished a year of these conversations, and the theme that keeps surfacing is simple: know your output, then build the smallest, clearest path to produce it. Everything else is sticky sugar on the floor.

Process Debt Truth: The cost of bad process isn’t the bad meeting—it’s the unmade decision that keeps everyone working on partial pops.