If the items in the backlog aren't related to the OKRs, then it's possible that they should be thrown our with the new cycle like the author says, but I wonder if it could mean that the OKRs need to be changed in certain situations as well. If there are things the teams value in the backlog that aren't part of any OKRs, and the dev teams are in some sense closer to the product than upper management, then that input should be filtered back up to be considered by upper management since the teams "on the ground" might be seeing information that the management doesn't have.
In many companies, engineering puts out fires, prevents disasters, gets rid of tech debt, etc. In short, engineering does a lot of maintenance that is needed to keep the entire company afloat. If that work stops, the company will go bust.
It's such an obvious key objective, it should be in every top-level OKR, yet it almost never is, which means that there's nowhere the engineering department can slot in that work. And how do you measure it?
