My overarching concept for forts is to acheive two basic ends:
1.) The fort acts as an actual barrier of sorts to enemy movement. You can get around it (if there's open land), but it's better to deal with it first.
2.) The fort -- much like forts in history -- can eventually develop into a settlement over a longer series of turns.
Now, how to balance and accomplish all of this is something I agree needs tinkering with. The actual ZOC concept from the first two games is gone, and unlikely to come back (which isn't such a bad thing). But, expansion of cultural borders accomplishes much the same effect. As has been stated, even with roads, enemy units move as if on unimproved terrain. So, this would accomplish the goal of ZOC.
In terms of eventually becoming a city, this is something which, I think, history bears out. Forts can and do become cities, although not always. They typically develop over time and by virtue of necessity. As the fort grows, its needs grow and a civilian population begins to move into the area to provide those needs. Eventually, this can result in a permanent settlement more akin to a city or town than just the basic fort itself.
Here's how I see this developing.
1.) Worker builds fort. May be within territory or outside of territorial boundaries. (as an alternative, a unit promotion called "combat engineering" could be awarded, which allows the unit to do things like build roads, or construct a fort. No chopping, mining, etc., though. Those are strictly worker functions)
1a.) The fort itself is simply a defensive structure at this point, which takes advantage of the surrounding terrain. The fort provides a base +25% defense in addition to the terrain bonuses on the square.
2.) Upon construction, the fort's own square has its own cultural border (like when you take over an enemy city). This is only the single tile.
3.) In five turns, the cultural borders expand to the nine-tile square on their own. No change in defensive bonuses, but the fort, now well established, has created its "ZOC". (Note: to accomplish this, at least one unit must be in the fort or border expansion halts.)
4.) 10 turns later, the fort's "ZOC" expands to a fat cross. The fort cannot work any of the tiles in the fat cross, however, and the "ZOC" will not expand any further at this point.
5.) From this point forward, the player may opt to send a worker unit to the fort, and establish a settlement (the worker is consumed in the process). The settlement will start only by working its own square the way newly established cities do, but will not have the ability to work a second square (unlike new cities). The fort can, however, build units (assuming it has the necessary resources available in its city box).
6.) In ten turns, a second tile can be worked. From this point forward, the settlement develops like any other city. Incidentally, the defensive bonus of the fort disappears at this point, since the tile converts to a city.
I think this is balanced because of the time involved as compared to the benefits. The actual defensive structure bonuses don't change until the fort develops to a city, at which point they disappear. The "ZOC" is created through the expansion of cultural borders, but it takes a relatively long time for this to happen. Also, while the fort can grow into a city, consider how many turns this would take.
First you'd have the construction time of building the fort itself (which is, what, 12 turns currently? Something like that, I think). Once the fort's built, you have five turns to get the fort to nine-square status. Another ten turns expands it to fat cross, after which you need a worker to get to "settlement" status, followed by ten more turns to actually become a city.
So...
Fort construction time + 5 turns + 10 turns + worker + 10 turns = city (from a fort)
vs.
Settler.
The fort serves its purpose this way, and can eventually develop into a real city, without being "spammable" and unbalancing the game.