More and more I find people I have met before coming back from completely different directions. This has lead to a curious and no doubt quite spurious calculation.
For example, I keep a file of the letters of reference I write for ex-students, to save time in case they want a repeat reference, and to make sure my lies are consistent. Looking over these I recently, I found that not one but two (out of 15 or 20) are from people who have subsequently come to Merlin with business propositions. One of our scientific advisory board members recently referred me to Igor Aleksander on a technical matter. Not only had I quite independently contacted him when writing 'Artificial Intelligence from A to Z' with Jenny Raggett some 10 years ago (on the advice of Piers Burnett), but I found that one of his neural net group had written to me to ask for a reference for one of my final year project students. A member of one of our company’s scientific board correctly identified the rather odd things on my college tie as heraldic pelicans – he was at the same college as me, although many years before.
In August one of the founders of a company called QT genetics contacted Merlin asking for funds with the immortal line ‘You may not remember me, but ...’. But I did – he was the only guy to consistently come top (ahead of me, chiz chiz) in our A-level biology class at school (I beat him in chemistry, though.) Bear in mind that there are alleged to be about 600,000 births a year in the UK – the chances of this are not large.
Is this solely because I move in a restricted world of biotech meeting rooms? No, because on November I was waiting for a train on Kings Cross tube station (Piccadilly line, noted for its crowds) and a guy I worked with at PA when we were both working in Cambridge said ‘Aren’t you ...’. If London really did contain 6 million people, what are the chances of meeting one on the mass transit system who I had met before?
The jackpot example, though, was in July when Eddie at Rothschilds arranged a dinner with Jonathan Weber, one of the Rothschilds (and now Merlin) scientific advisory board. As soon as he came in he said 'Bains, eh? I knew a Bains at school'. Turns out he went to the same prep school as I did, 35 years ago. Bloody hell! Either he did some pretty damn impressive homework, or he has an amazing memory to discover this at all. I could not remember a single thing about him.
From this and similar data, one may readily calculate the population of the world. This is exactly akin to the 'release and recapture' methods used by ecologists. You capture an collection of animals (voles, rats, professors, whatever), tag them, and release them. Some time later you capture some more. How many of them are tagged? Statistically this tells you the fraction of the population represented by your original sample. It is exactly analagous to meeting someone at (say) a New Year's party and then meeting them three months later in a business meeting (and frantically trying to remember why they look familiar). If the world contains 6 billion people the chances of this are fantastically small, but if the world actually contains three people then it is nearly a certainty that you will bump into them more than once.
This does not work if you keep track of the collected animals, of course, by radio transponders or regular letters. If I re-contact Professor X (or Vole Y) because I met them before, then that is not an independent sampling.
Given that, we can calculate the population of my universe. My complete letters file, address book, and contact database, etc lists about 1000 people who are more than casual contacts (ie excluding the people with whom I exchange a business card in a meeting and then forget all about). These are people I have exchanged letters with, or long e-mail exchanges, and in nearly all cases met. 26 of them I have met through two quite independent routes at least twice, from which we can calculate the population of the world, and the answer is (rather surprisingly) that the population of the world is 17,083 people. Sampling bias means that this probably does not include children and the retired, although I could well believe that we have had in excess of eighteen thousand children in our house some weekends.
This also predicts that, given an average family size of 2.2 children, I should have also met slightly more than 1 person who is the brother or sister of someone else who I have met for a completely different reason. In fact this has happened twice: stand up Andy and Rod, and Mike and Adrian. So, strong statistical validation there on a sample of two.
(I confess to being slightly unsure whether this should be 1.1097 people or 59.4 people. Well, this was never going to be published in Nature anyway. No doubt numerate friends will now all write in and tell me all the other places my sums are wrong.)
There are a whole lot of other co-incidences, which I am not sure how I should include in the calculations. For example, I have met professionally two people called David Stewart, two people called Dr. Phillip Stanley, and two called R. van der Meer. Also one called Paul Schaap and one called Paul Sharpe. And several called John Taylor (or Tailor), but that is the most common name in England (more common than John Smith, despite common belief) so that is probably not statistically meaningful.
This either proves a) the world really does contain only 17,000-odd people, and the rest are holograms (smelly, hard-elbowed holograms on the underground, but no more real for that) or b) I am become a boring old fart, so set in my ways that I never meet anyone new any more. I leave this choice as an exercise for the reader.