While I acknowledge that Axident is a far better civ-player than I am, from a strictly mathematical standpoint this argument is incorrect. The amount of great people received is given by the total number of GPP you produced so far, minus the amount "stored" in cities. The amount stored is upper bounded by 100(n+1)*c where n is the number of GP so far, and c is the number of cities contributing GP points. Because of this, the only reason to have more than one city contributing GPP is if this increases the total amount of GPP created. The second scenario here creates less GPP overall, and it does not create less waste than the first. The only reason that it may seem as if the second scenario wastes fewer GPP is because it will produce fewer GP overall, thereby making the maximum amount of GPP stored in one city smaller.
So, while the second method "works" in the sense that it also creates GPs, and in a given game may produce enough of them to win, it is always worse than the first method, both in terms of how many GPs are produced, and when they are produced.
The only reason I can see to use the second method is if the two cities produce diffrent types of GP, and you feel that the correct mix is more important than the total amount of GPs.
(Been lurking for some time, first post
)