It should also be clear that this article is about "deleting" code from an active project, not about "deleting" it entirely from the version control system. Thus, any code "deleted" through the described process could still easily be restored if necessary.
From reading the comments, I get the impression that there is a much deeper misunderstanding going on. This seems to be not so much (or not at all?) about identifying rarely run code paths, but about build units that have been abandoned on the dependant side.
So if there's a fancy onLocusSwarmAttack() that has never been run outside a test in a module to help applications deal with various kinds of datacenter outages called TenPlagues, it won't be in the crosshairs of Sensenmann. But if one day that module gets a successor (perhaps BeiWeltuntergangScheibeEinschlagen if it's made by the Zurich team?) and all code is supposed to eventually migrate to the new one, Sensenmann will tell you that TenPlagues has finally left the building when (if) it has happened.
Again, just my impression from the article and the discussions here (many seem to think it's about onLocusSwarmAttack()), probably resulting from a linguistic barrier between the language used at Google to talk about their monorepo and more conventional English. I guess they have done some quite deliberate shifts to help people "think monorepo"?
