[woooooo]

In the linked article, the situation was that they absolutely needed to change some hard-coded MONTHS value from 3 to 4 in order to keep the factory running.

That change should be shipped in isolation. With manual testing to cope for the lack of coverage, presumably. Refactoring doesn't have the same urgency as keeping the factory running, no matter how much we all believe in keeping the campground clean. It can wait until Monday in this particular case.

[cratermoon]

OK, but that's not the question I asked. What I want to know is how we can make refactoring as much a non-negotiable part of the process as code review, tests, CI/CD, or whatever you consider essential, non-skippable, even under a short timeframe?

[woooooo]

In the context they were in, the answer to all of those questions is "shut the fuck up, we can talk about it later".

In the normal course of business, it's a different conversation. Even then, if you're making some poor dev refactor a bunch of code because they had the misfortune to touch it, maybe you should have done the work yourself a long time ago. Or written a linter and ticketed owners.

You don't want to make feature development some sort of pain lottery.

[cratermoon]

Am I reading you right, that you consider refactoring "pain"?

[woooooo]

Unplanned work being arbitrarily scoped into sprint is pain. Doesn't matter the source.

In practice, your approach often turns into "junior engineer abuse" where they have to clean up a bunch of unrelated pre-existing stuff to make the seniors happy as a condition of shipping their feature.