A small magazine about things that claim to work
Issue 3 · 27 pieces
We buy what we write about. Every figure is labelled as ours, attributed to a named third party, or the maker's own claim — because most of the numbers in product writing are the third kind, presented as the second.
Apparatus
Every app in this category publishes an accuracy figure. Almost all of them measured it themselves. We ranked seven by a single criterion — whether anybody outside the company has verified the number — and one app clears it twice.
Apparatus
Every camera in this category fails the same way, for the same reason, and it is physics rather than software. What separates them is what happens next.
Field Notes
Barcode scanning is presented as a solved problem. It is solved for recognising the code and not solved for what happens next.
Apparatus
Putting meals and glucose readings on the same timeline is genuinely useful. Everything past that sentence is where consumer apps start saying things they should not.
Apparatus
Interfaces are what you notice and databases are what you get. Three models are in use, they produce measurably different results, and almost nobody names which one they are on.
Money
A plan that never expires, a trial with the expiry buried, and a reduced-database version that is quietly worse rather than visibly smaller. Only the first is free, and only two apps here offer it.
Apparatus
If you are tracking macros seriously you are asking two separate questions — what should my targets be, and what did I actually eat. Different apps answer them.
Field Notes
Log one food, open the full breakdown, and count the numbers. What you find tells you more about the app than any review will.
Field Notes
From directly above, a bowl with two inches of rice looks identical to a bowl with four. This is not a software limitation and it will not be fixed.
Apparatus
A scale is the only instrument in this magazine we could actually calibrate ourselves, so we did — and the results argue against spending money.
Apparatus
Eleven apps advertise an accuracy figure. Nine of those figures were produced by the company selling the app. Here is what happens when you sort the category by who did the measuring.
Money
We tried to establish the renewal price for nine subscriptions. For four of them we could not, so we have said so rather than guessing.
Money
The same arithmetic, performed by seven companies, priced from $34.99 to around $209. Over three years that is a difference worth a moment's thought.
Money
We subscribed to nine things and then cancelled all of them. The distance between the easiest and the hardest exit was about forty minutes and one phone call.
Field Notes
Ours, attributed, or the maker's own claim. Most product writing collapses the second and third, and that collapse is where readers get misled.
Field Notes
Imprecise labels, crowd-sourced entries, logging that turns into another chore — the complaints against calorie tracking are real. The research on dietary self-monitoring draws a narrower, more useful conclusion than either side of that argument usually admits.
Field Notes
Screens delaying sleep is real physiology. Whether a lens coating corrects it is a separate question, and a systematic review of seventeen trials found the evidence too thin to answer it either way.
Household
Every flask in this test kept coffee hot enough to drink after twelve hours. What separated them was the lid, and the lid is not what the marketing measures.
Money
Eleven services raised prices on us in twelve months. Not one of them called it a price rise in the subject line, and three raised prices twice.
Household
The rated life on the box is 25,000 hours. We have had bulbs fail in under two years of normal use, and the pattern is not random — it is about where they were installed.
Money
An extended warranty is an insurance product sold at the till by somebody on commission. Here is how to work out, in about a minute, whether the one in front of you is the ninth.
Household
Edge retention is the thing knife marketing sells and the thing that mattered least, because a knife you will actually sharpen never gets blunt enough for it to matter.
Household
Every nonstick coating fails. What varies is whether it takes eight months or thirty, and the thing that predicted it was not what I paid.
Field Notes
We searched twenty common foods across four apps and checked where the correct entry ranked. On crowd-sourced databases, position told us nothing at all.
Field Notes
±1.1% does not mean your dinner will be within 1.1%. Here is what these numbers actually describe, and the three questions to ask of any of them.
Household
Every set holds food. What separates them is whether the lid still seals after two hundred dishwasher cycles, and only two of these did.
Field Notes
Two of us weighed and recorded every item for seven days, then logged the same food normally. The gap between the two was larger than the gap between the best and worst app.