(This is very much a work in progress, looking for comments on this first stab at it.) It's an ongoing question as to when it's better to appoint a specialist, and when it's better to work a tile. The direct yield of specialists and tiles is easily counted, save for one question... what's a GPP worth? This is an attempt to put a rough value on GPPs. First, the easy part. The minimum value of a GPP is zero; if you accumulate them in a city that doesn't produce another Great Person before the game is over, they did you no good at all. Now, the hard part. We're really trying to deduce the value of a GPP per turn - because we are comparing with tile yields, which are also accumulated per turn. But if we consider settled GPs, it's hard to figure this out; the longer things go on, the more GPs will be produced, so the value of a GPP per turn (used for settling GPs) depends critically on how many turns are left in the game. This gives us some reassurance we're taking the right approach relative to gamespeed; on Quick, each GPP produces a higher fraction of a GP - but with fewer turns left to pass, it will produce the same number of GPs overall. So let's instead consider bulbing. The value of bulbing peaks when a GS's 1500+3*population beakers "fits into" a technology; somewhere around Education. But by then several GP have been born already, so the cost of one will be high. Let's guess that, in fact, the cost of a GP is around 1500 GPP; then conveniently each GPP is pretty well interchangeable with a beaker-per-turn. But... if the city worked tiles, it might get commerce or hammer multipliers from buildings; but if the civ was Pacifist or Philosophical, or had the Parthenon or the National Epic, the GPP might also be multiplied up. So approximately, the value of scientist GPPs (that will actually produce GS) is (GP birth rate multiplier) * (maximum value of bulbable technology, or 1500+3*pop) / (current GPPs per GP), to be compared with the actual post-multiplier yield of the tile in question.