Aren't Farms usually better than Workshops or Watermills?

[Elandal]

For production, you want maximum hammers. This means you want maximum FP-total with F=[pop*2] exactly when possible.

Pre-biology farms are FP=3 (grass F3, plains F2P1). Post-bio farms are FP=4.
Watermills are FP=3 when introduced, FP=4 post rep-parts, FP=5 post-SP.
Workshops are FP=2 when introduced, FP=3 post-guilds, FP=4 post-chemistry, FP=5 post-SP.
Mines are FP=4, FP=5 post-railroad.

So what you want is: enough food to work tiles, then maximum number of hammers out of those tiles. Pre-biology farms aren't as good as post-repparts watermills or post-chem workshops, and are never as good as post-SP watermills/workshops. However, you may need them to provide food required to work eg. mines, workshops, or plains watermills (pre-SP - post-SP grassland workshops and plains watermills are food-neutral, grassland watermills food-positive).

Basically non-resource (non-floodplain) farm is useful only in case you need food to support working food-negative tile or a specialist. Other than that, it's probably the worst tile you can be working.

 

[The-Hawk]

Typically, the only time I build workshops is late in a space race attempt. In the early-middle part of spacerace, the goal is to maximize science, hence cottages. Late in spacerace, some cities need to build parts, so I might build workshops over cottages in a couple of cities. At that point, you have state property, so workshops are a good option.

Culture victory - no point. Cottages, cottages, cottages.
Time victory - No point. Farms, farms, farms (maximize population)
Domination, Conquest - goal is to be finished long before state property makes workshops a good option.
Diplo - also about fast science, therefore cottages. Maybe a few workshops in the city that will build UN.