Look after your face

The science of projecting people’s faces into the future passed another milestone in June, according to the BBC. For those of you who missed the show, child psychologist Kris Murrin used advanced computer technology to ‘age’ photographs of children in order to see what they might look like 30 years later. Nothing new, you might say – after all, The Actuary did the same to Charles Kennedy on the cover of May’s issue. But the Murrin project was an advancement because it took in eating habits as one of its inputs. Eat junk food, it said, and you see yourself expand. See figure A for an illustration. Now, a cynic would say that the same result could have been achieved by cutting and pasting the child’s eyes and nose onto the face of a fat adult. But I am a member of the actuarial profession’s Working Party into the Projection of People’s Faces (our motto is ‘making physiognomical sense of the future’). This is a very serious subject, and face projection is a field to which, I feel, our profession has much to give.The first point to note is that there is not just one true future face. There is, rather, a range of possible faces. It depends, of course, on the inputs. The portly gentleman depicted in figure A, for example, was the result of a projection based on the assumption of a high junk food consumption. An alternative basis for projection would be to put him on the Dr Gillian McKeith diet, and he might look quite different. A lot more miserable, probably. The maximum period over which you can reasonably project someone’s face is about 30 years. Beyond that period, the range of possible faces becomes too wide. The expanding flannel of doubt becomes like an oversize beach towel.You do, of course, need to be extremely careful when giving advice based on the output of such software. Suppose you tell a client that, given a healthy balanced diet and a reasonably active lifestyle, she can expect to look like figure B when she’s older. Based on your projections, she might take to skipping school and practising her autograph, looking forward to a career in acting. But then you find that your projection model has been incorrectly parameterised, and she actually turns into figure C. In these litigious times, you would stand no chance.The way to protect against this is to thoroughly test your projection software. Pick a test case, project the face into the future, take the median face, and discount it back to the present day using your median assumptions. You should end up with a reasonably similar face to the one you started with. By way of illustration, I did this to Michael Pomery and Harvie Brown using a commercially available piece of face projection software (see figures D, E). If you come upon a discrepancy, as in figure F, then further investigation is needed.But to whom would such a service be of any use? I can think of three markets. Firstly, the vain. Second, would-be actors, as discussed above. Third, those who wish to know what their partners will look like in 30 years. Here our profession could really act in the public interest: this technology could serve to reduce divorce rates, by the very simple process of reducing marriage rates. It might also serve to increase suicide rates (figure A) but that, as they say in the east end of London, is a small price to Dorian Gray.