Home · The site · Cookie policy
The site
Cookie policy
Which cookies this site sets, when, for how long, and how to change the choice.
What is set before a reader has chosen anything
The honest answer is nothing at all, and it belongs at the top of this page because most cookie policies bury it three sections down. On a first visit the pages load, the banner appears, and not one "statistical" cookie has yet been written (the banner first, the answer second), because the only thing the site wants at that moment is an answer to the question it has just asked.
That answer is what the whole arrangement turns on, and the sequence is worth stating precisely. Until "consent" arrives the statistics are not loaded at all ("Google Tag Manager" is never called and the measurement library is never fetched), so there is nothing to delete afterwards for the simple reason that nothing was ever created.
The one cookie that records the choice
Accepting or refusing writes a single cookie named "fbc_consent", in the "first-party" domain of this site, and it lasts six months from the moment of the choice (six months from the click, counted by the browser itself). What it holds is the choice and not an identity: a short value saying that statistics were accepted or refused, with no address in it, no number for the reader and nothing that could be matched to a person.
It is also the cookie that makes a refusal stick, which is the part readers tend to find counter-intuitive when it is explained to them. Without it the banner would return on every page and a reader who had said no would be asked again at every click (every page, every visit), and that is the commonest reason people end up accepting something they never wanted.
Six months is the point at which the question comes back by itself, because consent given once is not consent given for ever; a reader's view in April is no evidence whatever of the same reader's view in October, and we would rather ask twice a year than quietly assume.
What acceptance adds, and what it does not
Acceptance loads Google Analytics, and it loads it through "Google Tag Manager" rather than directly, which is why a reader watching the network traffic sees the tag manager arrive first and the measurement code a moment behind it (the manager, then the library, then the first call). From that point the cookies Google Analytics sets are written in this domain, their names begin with "_ga", and their durations are the ones Google publishes for them rather than anything invented at this end.
What those cookies do is counting (pages, "sessions", the route somebody takes through the site), and what they produce is an aggregate rather than a file: totals of visits to a page, with no history attached to a visitor. We read them to see which pages are used and which are ignored, and for nothing else whatever.
Refusal costs a reader nothing that he can see, and we have taken some trouble to keep it that way. Every page, every table and every record behaves identically with the statistics switched off (the tables, the records, the four boxes, all unchanged); we would rather hold an accurate picture of a smaller sample than run a "banner" that nags!
The region service, which writes no cookie at all
The mechanism that works out whether a reader sits in a "protected market" deserves a line here, because readers reasonably assume that it must be storing something. It is not: the cookie lifetime of that service is set to zero and no frequency cap is applied to it at all (zero, and no cap), so it keeps no record of having seen anybody and writes nothing whatever to the browser.
The cost of that choice is a small inefficiency, and we accept it deliberately rather than by oversight. The determination is made again on every page instead of being remembered, which is slightly wasteful and leaves nothing at all behind, and a reader who blocks scripts receives the international version of the page without being asked anything.
What the service does receive, for the length of that single determination, is set out item by item in our privacy policy. It is worth reading if the question matters to you, because the list is short and the caveat at the end of it is the honest one: a "region" deduced from an address is an estimate, not a fact about anybody.
Changing the choice, or taking it back altogether
The control that reopens the banner sits at the foot of every page, beside the links to the privacy and cookie policies, under the plainest label we could find.
Opening that control shows the choice currently recorded and lets you replace it with the other one. Refusing after having accepted stops the measurement at once and leaves the old "_ga" cookies to expire on their own schedule, and anybody unwilling to wait for that can delete them from the browser inside the same minute (a minute's work, in any browser).
A browser can also act on its own account, and against a site like this one it wins every time it tries. Deleting the cookies for this domain removes "fbc_consent" along with the rest, after which the banner returns as though the visit were a first visit (a first visit, as far as the site can tell), and blocking third-party scripts outright stops the tag manager before it loads.
Two last points, both dull and both true. Nothing on these pages is "personalised" on the strength of a cookie and no advertising network is present to read one; the affiliate links work from a plain address and need no stored identifier of any kind in order to function.