Metin learned hot metal typesetting in 1968 and practised it professionally until 1994, by which point he was one of perhaps a dozen people in the country still doing it for money. He is asked about this often, and he has become impatient with the way the question is usually framed. The assumption behind it is that he watched a craft die and must therefore be mourning. He is not, particularly. Hot metal was hot, loud and toxic; the lead was lead. He remembers the noise more than the beauty and does not miss the varicose veins. What he objects to is not the loss of the craft but the confidence of the account that replaced it. The story told about the transition is that computers made typesetting accessible to everyone. In one sense this is simply true. In another it conceals a substitution. The typesetter did not only place metal. He made several hundred decisions per page that nobody had ever articulated: about spacing, about where a line should break, about which of two technically acceptable arrangements would be easier to read. He made them at a speed that only came from twenty years of not needing to think. Those decisions did not migrate into the software. Some were encoded, badly at first and quite well by about 2005. Many were simply dropped, and the reading public absorbed the deterioration without complaint, because a slight increase in effort per page is not a thing anyone notices. He is careful here. He does not claim the old way was better overall. He claims something narrower and harder to dismiss: that a skill embedded in a person, unarticulated, disappears when the person does, and that nobody can measure what was lost, because the measurement would have required knowing what to measure while the person was still there. This has happened, he points out, roughly continuously since the eighteenth century, and will happen to the current generation of designers within his grandchildren's lifetime. He teaches a two-day workshop now, twice a year, mostly to art students. They photograph everything. He lets them set one line each, badly, and keeps the results. "They think they are learning printing," he says. "They are learning that the machine had opinions."