"I'll look it up, and then I'll know"

The search starts from a simple idea. You have a symptom, somewhere there's a page that says what causes it, and once you've read that page you'll know what you have.

A search can't deliver that, and the reason is simple. It has one piece of information about you, which is the words you typed. It doesn't know your age or your family history. It can't tell how much coffee you drank, and it can't look at your calf, test your reflexes, or watch you walk. It doesn't know that you've slept badly for two weeks. It doesn't know the twitch started the day after you moved a wardrobe. All it can give you is a list of every condition in which those words appear.

That list is long for almost any symptom, because bodies have a limited number of ways to signal trouble. Tiredness, headache, twitching, dizziness and a change in bowel habit each appear on the symptom lists of dozens of conditions, from trivial to grave. The search gives you the list. It can't tell you where on the list you are, and that was what you wanted to know.

There's a study that shows the size of the gap. Hannah Semigran and colleagues at Harvard tested twenty-three online symptom checkers. These are tools designed for this exact job, which ask you questions in the way a search engine doesn't. The researchers fed them forty-five standard patient cases with known diagnoses. The checkers put the correct diagnosis first about a third of the time, in 34 per cent of attempts. They listed it anywhere in their top twenty in 58 per cent. So tools purpose-built to narrow things down still fail to name the right condition, even among twenty guesses, for roughly four cases in ten. An open search followed by an hour of reading isn't likely to do better.

Compare what happens in a consulting room. The doctor asks when it started, what makes it better or worse, and what else you've noticed. They look at the leg. They may test its strength and reflexes, which takes a minute and tells them more than you could get from a month of reading. They know how common each possibility is in people like you. And they have a tool that reading doesn't offer, which is time. "Come back if it's still there in two weeks" sounds like being fobbed off. It's a real test, because most harmless things will have gone by then.

I'm not claiming doctors are always right. But what settles the question is an examination, some history and the passage of time, and a search has none of those. What you get from an hour of searching is a longer list than you had after ten minutes. You don't know more about your own body. You know more about diseases, and that usually leaves you more worried.