Very rarely is one strategy always more effective than another. Unless you're talking about rushing AI's with workers or something.
If you get ENOUGH STUPID LANDLESS ISLANDS, slavery will be better for a long, long time. A capitol with 3 tiles to work besides its city square?! Sign me up?
. It had 4 seafood though. You use what you get.
You only make so many new cities though. You DO want them to become core productive cities ASAP, but if you're looking at 1-2 of these vs 7-8 "core" cities, and you see an opportunity or wish to strike an AI, it's probably time to leave slavery for the workshop hammers. At that point you'll have more workers which means these newer cities will be working improved tiles and should grow quickly anyway. You'll be wanting the hammers in core cites, either to build stuff like university/observatory most efficiently or to make units.
As you pointed out, this also depends HEAVILY on the tiles the map has available to you. If you have 5 grassland hills in your production city along with iron and enough food, what's the need for workshops? Minimal. If you're in one of those (ZOMG MASS JUNGLE RIVER) types of deals, you can't rely on the whip forever
.
It's one of the reasons I like civ so much. The map variance is so great that it's improbable that 2 games will be alike, and the fact that the optimal strategy will vary adds a lot of depth to the game.