Ok, thanks for the additional info - and I do have some comments on those points.
>. I build temples mainly to get that first expansion, so I can work
> all my tiles. This makes city placement more flexible, I can pull in
> valuable tiles from a distance. I somehow must be overrating this argument.
Ah, would I guess right to think you enjoy civ1 / civ2 style "optimal city placement" where each gets their full 21 tiles to work? The more one tends to far apart placement, the more important a temple in each city gets by FAR. I've come to VERY MUCH like a 3-step placement between cities. The benefit to defense and war are tremendous. Also, when cities are 3 away that row of border gap you would expect between them will disappear - so you get to work any tile you want without needing a temple. So if I might generalize your thought... build a temple when you want to work more tiles. But as to the when, build a temple WHEN you NEED to work more tiles ASAP. See I don't disagree with a temple per city, and I very frequently eventually get a temple per core or second ring city, it's definitely NOT a high priority unless there is a lux or wheat sitting outside the border. In other words if the tile you pull in won't even be worked until it's size 6 or 7, build the temple when you're size 5 or 6.
> 2. This is the crucial one. Since my early civ1, civ2 days, I like to
> develop the population of my cities as evenly as possible.
Ok, I did sense that civ1-2 philosophy :P It's not true that even *builds* lead to an avoidance of waste of commerce. Even happy faces lead to avoiding a waste due to luxury %. High commerce low shield cities will see greater benefit initially, and even more after marketplace, than low commerce high shield cities. So high commerce favors holding off on the temple until after marketplace.
> So I actually like to build granaries (which I seldom do anyway) in
> low-food cities. For example, if my capital has 4-5 extra food, I do
> not even build a granary. If it only has 2-3, I build it. In other
> cities I almost never build granary.
As you point out, yes there are several critical mistakes here. (Not mistakes as in weed, mistakes as in barriers to mastery of the game :P )
1. Worker farms spitting out a worker per 1-3 turns basically forever have a massive impact on overall city productivity and economy.
2. High food secdonary sites with decent shields too are the best settler producers there are, with a granary.
3. Subtle but very powerful item to learn. Civ3 is a "quantum" game, not "classical". Building a swordsman, there is no difference between 10, 11, 12, 13 or even 14 shields per turn. There's a huge difference betwen 14spt and 15spt. One sword per three turns, vs one sword per two. That's a whopping 50% productivity gain. That's why you hear some players rejoice over reaching "the mark" of 10spt, 15spt or 20spt depending on what the key unit being built is at the time. There is likewise a key quantization of food. +1 stinks, +2 is twice as good, +3 a fair improvement, +4 no biggie, but +5, ah that's magic. With a granary +5 food means it grows every two turns. At +4 food it's every three turns. a 50% difference. Get +5 food in a city build a granary and with just 5 shields, presto!! You have one worker coming out every 2 turns for the next two millenium. It's better than that - if the extra tile that is brought in the instant the city grows brings in extra shields, you can actually get one worker per two turns with what 'seems' to be only 4 shields per turn (plus a mined hill that is chosen by the governer next)
Bottom line: a city that can reach 10spt is of superb value for cranking troops and should specialize in doing so. Often this is the capital or the first built city (and/or the future FP). A city that can reach +5 food and 4-5 shields *BEGS* for a granary and to become a worker farm. A city with +5 food and 7-8 shields BEGS for a granary to become a settler farm, spitting out one settler per four turns like clockwork, which is awesome. Some people throw out the term 'farm' simple to mean "it's got a lot of food" but never set the city up properly to get this kind of 'super-effective' production.