What the record cannot see about stopping

1. When

No date, anywhere. Readers comparing this small inventory with broader advice can also see PCWorld.

An object that stopped in 2012 and one that stopped last month are identical entries. The record knows when each object arrived and that it is now out of use, and nothing about the interval.

Every duration figure on this site rests on an interval the record never observed.

What it costs: every statement about how long things last before stopping. The years section works entirely from arrival dates and the present state, and cannot produce a single figure for time-to-stop. This is the largest gap in the method.

2. More than one reason

The field takes one value. Four objects in the record had two stopping events — replaced, then given away — and all four are coded by the last one.

What it costs: the replacement figure is understated by four and the given-away figure overstated by four. Known, small, and reported in both places.

A multi-valued field would have fixed this and would have made every percentage in this section ambiguous, because a hundred and thirty-five objects could then produce more than a hundred and thirty-five reasons. The single value was a deliberate trade.

3. What replaced it

Thirty-eight objects were replaced and none of them points to its successor.

What it costs: the record cannot follow a succession. In at least four cases two generations of the same kind of object sit in the same cupboard, and the inventory holds both without any link between them.

A field recording what replaced what is the second most useful thing this method is missing, after a date.

4. Whether it was gradual

Twenty-four objects are coded as having stopped for no stateable reason, and that category was assembled from people's inability to supply a cause.

What it costs: the record cannot distinguish a genuine drift from a forgotten event. Eighteen per cent of the stops rest on a distinction the instrument cannot make, and the article on those twenty-four states both readings without choosing.

5. Condition at the moment of stopping

An object is recorded as working or broken now. Whether it was in good condition, marginal, or limping along when it stopped is not recorded.

What it costs: the line between it broke and it was replaced because it was getting worse is drawn by whoever answered, and different workplaces may have drawn it differently. Some share of the thirty-eight replacements were probably objects on their way to failing.

This is the one place where the breakage count of nineteen might be materially wrong in either direction.

6. Anything about the object itself

No brand, no model, no price, no age at purchase, no specification.

What it costs: every comparison. Nothing in this record can say that one sort of thing outlasts another, because the record does not know what sort of thing anything is beyond its kind — screen, cable, chair.

This is not an omission to be fixed in a later version. It is the rule the method was built on, for the reason set out on the method page, and a record that recorded brands would be a different record making a different kind of claim.

7. The things that were never counted

An object that broke years ago and was thrown out the same week leaves nothing behind to be remembered by.

What it costs: the breakage count is the one most likely to be too low, because breakage is the ending most likely to remove the evidence of itself. A replaced object usually goes in a cupboard and gets counted; a dead one often does not.

The record cannot estimate the size of this. It can only name the direction, which is upward.

8. Which kind an object is

Kind is in the record, but it was assigned by whoever took the inventory rather than asked, using ten categories settled at the pilot.

What it costs. Every reason-by-kind figure in this section is a count of somebody's classification, not of somebody's answer. The line between storage and organisers and other was drawn once, by one person, and nineteen objects ended up on the far side of it.

That matters most for the two cells this section leans on hardest: the fifteen cables that stopped for no stated reason, and the seven storage items retired by a change of work. Both are robust to small reclassifications — a cable is a cable — and both would move if the categories had been cut differently.

What a reader should do with the eight

Treat the counts as counts and the explanations as description.

The counts are firm: a hundred and thirty-five stops, nineteen breakages, twenty-four unexplained, and the kept-against-gone split by reason. Those come from closed fields present for every object.

The explanations are not. Why cables drift out of use, why employer equipment ends up stored, why a change of work retires a printer and keeps a tray — all of those are accounts people gave, or inferences drawn afterwards from the shape of the table, and every one of them is offered as a reading of eleven places rather than as a mechanism.

The distinction is the whole of what this page is for. A record of this size can count cleanly and cannot explain anything, and the honest version of it says which sentences are which.

What remains after all seven

Four statements this section will stand behind.

A hundred and thirty-five objects are out of use, and nineteen of them broke. A count, from the entries, with the caveat in point 7.

Fifty-one per cent of the stops carry no information about the object. Replacement and a change of work together, both external to the thing that stopped.

Event-type reasons produce departures and drift-type reasons produce cupboards. Twelve kept against seventeen gone for breakage and handovers; seventy-four against thirty-two for the other four.

Eighteen per cent of the stops have no stated cause. Reported at full size rather than folded into the other five.

Everything else in this section is description of eleven places, offered as description and not as a finding about desks in general.