EMan,
If I understand you correctly then, you do the calculations for the full 21 squares in the fat X first. Then, if any tiles overlap with another city you keep the overlapping tiles for the higher scoring city, and re-do the calculations for the lower scoring city without those overlapping tiles.
For example, were we to evaluate New Salonika and 251 in the 1st screenshot of post 94, we would first notice New Salonika as having 1 center, 6 coasts, 2 seas, 1 plains cow, 4 grass, and 6 plains. From what you've written above, it seems you would assign a value of +0.50 to a plains cow, since it has 5 food fully railroaded... and your numbers seem to have a base of 4 food when fully railroaded and increase or decrease by .5 for each food over or under 4 food respectively (except for sea squares, and city centers). So, New Salonika has a value of
((2+.5-3-6-1.5)/20)=-.40
With a plains sugar having the same value as a grassland, 190 keeper actually, if I did the calculations correctly the same value as New Salonika, but I digress.
251 has 1 center, 4 grass, 1 plains cow, and 15 plains. So, it has a value of approximately
((.5-1.5-7.5)/21)=-.40476... so it doesn't make the cut. But, supposing for a moment, just to try and understand your system, we had the threshold value at -0.50. Then, since 251 comes in weaker than New Salonika the overlapping tiles will get assigned New Salonika implying that 251 loses 2 grass, 1 plains cow, and 4 plains. Thus, supposing it didn't overlap with any other cities. So, it has 1 center, and 11 plains giving a value of approximately
((-1.5-5.5)/14)=-.42857...
Would you do the calculations that way? What did I miss?