Well in Game of the Month we actually test these theories out.
Our most outstanding 100k result was Kuningas's 850ad win in the Indian COTM of August 2005. Analysis of his culture graph (using CivReplay on his final save file) shows that he builds his first temple in 1830bc. He has built only 4 temples (in 22 towns) by 610bc, at which point he captures Artemis. He does not then engage in any significant culture building until his switch to Feudalism in 280ad. By this point he has 224 towns; he will not build any more towns (ending the game with 221) or gain significant extra territory (he has 1230 tiles at this point, and the dom limit is somewhere between 1250 and 1300).
The best 100k result in PtW during my time in the GOTM community was Bremp's 1180ad result playing Persia in March 2005. Without TOA or Feudalism available, his building program is much more evenly spread out. He starts with a library in 1075bc, and has 5 built before his one and only revolt (surely to Republic) in 875bc. By 210bc he is making 83cpt in his 30 towns, which works out to an average of about one lib or temple in each town. This is the last turn in which he does not complete a cultural building; from here on he adds an average of 14.2cpt per turn, at a fairly uniform rate. He ends with 160 towns and 1170 tiles, against a similar dom limit to Kuningas's game. He cruises up to the limit throughout the game, again in a fairly linear fashion.
So my conclusion? Jazzmail's method is appropriate for PtW, and MAS's method is better in C3C. Certainly the way to play 100k is more obvious in C3C, where Feudalism and TOA are available. And it is interesting to note that going all the way to Communism isn't worth the effort in a PtW 100k run.