What we got wrong

What goes in here

Anything published on this site that was not true, and anything that was technically true and left a false impression. That includes our own pages about ourselves — two of the three entries below are exactly that. Rewordings, design changes and typographical fixes do not go here; there would be no room for anything else.

If you find something, tell us. Say which page and which figure. If the paper disagrees with us, the paper is right.

The log

6 September 2026 — five drug guides carried no MHRA safety warning

What was wrong. On 11 May 2026 the UK's Medicines and Healthcare products Regulatory Agency strengthened its safety warnings for finasteride and dutasteride, and set out what someone taking the 1 mg hair-loss dose should do about it. Five guides on this site described those drugs at length and said nothing about it. A guide that omits a regulator's current warning is not neutral about it; on a prescription medicine it is the omission that does the damage.

What it says now. Each of the five carries the MHRA's position in its own words, its advice to stop the 1 mg dose on developing depression or suicidal thoughts, the Yellow Card reporting behind the change, and the two things that belong beside those counts: that spontaneous reports are suspicions rather than established causes, and that a regulator reviewed the same evidence and strengthened its warnings anyway.

Where. Finasteride, dutasteride vs finasteride, low-dose finasteride, topical finasteride and stopping treatment.

6 September 2026 — the privacy page understated how long an IP address is kept

What was wrong. Privacy said the enquiry form's rate-limiting rows "are deleted after an hour". The code that does the deleting keeps them for up to two, and only deletes when a later request happens to touch the table — so an idle period extends it further. The sentence was written from an intention rather than from the code.

What it says now. That the rate-limiting table records your IP address, that rows are deleted once they are more than two hours old by the next request that touches the table, and that they are not joined to anything you sent.

Worth saying plainly: this was the third false sentence found on that one short page in a week, after "nothing counts your visit" and "no IP address is kept beside your message". The page describes our own behaviour, which is the thing we have least excuse for getting wrong. The durable fix is not a proofread — a retention claim should be generated from the code that does the deleting, the way the affiliate disclosure is already generated from the module that would earn the money.

6 September 2026 — the Turkey transplant figures came from clinics selling the procedure

What was wrong. The hair transplant cost guide repeated the two figures that circulate everywhere for the Turkish market — a national procedure count, and a per-graft price against the American one — as though they were measurements. Both trace, through our own reference list, to clinics that sell the procedure. We could find neither in an independent source, and the nearest thing to an industry count publishes a worldwide total with no country breakdown at all.

What it says now. The figures are still on the page, because they are what you will be quoted and removing them would not stop you meeting them elsewhere. They are now named as clinic marketing, with who published them and what could not be found — and with the observation that a price comparison published by the cheaper seller is an advertisement with a decimal point in it, pointing in the direction that sells the flight.