Field note

How to sit with a worst-cell list without ranking your own engineers

16 September 2025 · Nino Beridze

Two people reviewing papers side by side

A worst-cell list looks like a ranking of people if you let it. It is not. It is a ranking of rows whose counters earned them a place in this window, on this site list, after outage hours were dropped. The sitting exists to write in the margin: already in the swap, needs a hop reading, not our cluster, keep for the freeze.

We print the list with the counters visible. A composite index that nobody can unpack in the room will be challenged, correctly, as a trick. If the vendor file gave us PRB utilisation and retainability, those are the columns. If a row is high because of a known parameter freeze, the planning lead should be allowed to say so without the table becoming a performance review of the engineer who left it there.

Six people is enough. More than that and every complaint from last summer gets a speech. We ask for one planning voice, one operations voice, and transmission if the radio story has started to smell like a pipe. The useful artefact is the marked list, which goes into the one included revision of the pack.

I have watched rooms try to turn this into a vendor argument. We will not rank Ericsson against Huawei in a briefing. The pack is about this cluster’s hours and this list’s names. If the constraint is a hop, the next piece of work is a backhaul review, scoped on paper the next working day — not decided with a show of hands at 16:40.