9 tiles is actually the 8 tiles surrounding the city hub, at least from what I can see of the screens above.  That's 4 subtiles per tile, or a total of 36 possible build points.  That may still be too many to drive city specialization, I don't think I've built 36 independent improvements yet.

How about more maintenance when buildings are further away from the hub? maintenance = distance * base_maintenance or something like that. Snake away if you wish, but it is going to cost you.

Don't do this. This is a gamey limitation that worsens the overall experience. Snaking is a shitty mechanic, but so is being arbitrarily limited to a certain size. It makes no sense logically that a city could only expand so much, and from a gameplay perspective it tackles the symptoms rather than the problems. Snaking is symptomatic of city design being irrelevant with regards to the city's contents, but not with regards to the worldmap.  The "best" solution to the second problem would be to model building productivity in such a way that positioning matters for buildings, encouraging people to think about building positioning in logical ways within the city. Of course, this is of limited feasibility - it would doubtless take a lot of code work to implement, and conveying the requisite information effectively would require reworking the city-building interface. A lazier but still acceptable solution would be to just take control away from the player - the player doesn't decide which tile his new building goes on, so snaking doesn't happen. And since the players can't effect it (and nothing of value is lost), most players won't care.
Another good solution is to look at why players are snaking in the first place - the irrelevance of building placement is why players don't refrain from snaking, this is why they do it in the first place. Players want to be able to block off chokepoints. Make it possible to do that  through a mechanic dedicated to that purpose, and snaking won't happen. If players can just build a wall extending from their city to the coast on one side and the mountains on the other, snaking is more or less irrelevant (except with regards to resources, and we have outposts for that) so the problem is more or less gone.

You also say that limited city footprint would encourage specialization by prohibiting taking all good options. I don't think there's a big problem with generalization in the current build, but to the extent that it exists, city generalization is a symptom of insufficient link between buildings and local resources. Although it's possible that no one solution will suffice, a good way to fix the problem is with buildings that effect resource production, such as (for example) doubling a city's crystal or iron or horse production - or make buildings that are advantaged, such as one that gives +1 gold production for every diplomatic capital point that the city generates in a season.

The valid criticism of the 9 tile limit is why I dropped it as an idea a few months ago. One tile limit seems to be a better choice. 

I don't like the idea of making building position affect its usefulness either. Too much management for most users and it opens up a whole new can of worms as far as exploiting the system goes. There are many directions you can go with cities, but most of them see at least as problematic as the current system. At least the 1 tile city reduces memory constraints. Most people agree that this is a very important factor.

I like limitations that force a choice. 9 tiles (that is really more, depending on how big the buildings are) works for me. Building EVERYTHING was a terrible mechanic.

1 Tile for level 1-2 cities, 2 tiles for level 3-4, and 3 tiles for level 5 cities I say.

This rewards you for city growth (feeling rewarded is good), as well as helping contain both snaking and memory usage (these things are also good). What do you think Frog?

Great points regarding walls and why people build the way they do, Cruxador.  I am all in favor of a separate city build screen if it gives us the ability to build fortifications that matter during tactical battles and strategic map movement. 

 Let a city wall off a peninsula if it is no more than 4 tiles across.  I mean you can almost do it with the destroy land spell now if you want.

Let an outpost build a ballista or catapult on top of the keep that exerts a ZOC that prevents neutrals or hostiles from entering w/o proper treaty/tribute or the airspeed velocity of an unladen swallow.  (Or attacking the outpost and taking the keep in a special tactical map with walls, moats and vats of boiling oil... {THOSE would be fun techs to research... scaling ladders, flaming arrows, battering rams, training War Trolls or Ogres to destroy walls, a swarm of wall-climbing spiders.}) 

Great Wall could even be a one per civ or one per world achievement or an Earth spell obtainable via quest or levelling up.  Different civs and spell schools could have a myriad of different ways to siege or defend against one:

Earth to Mud (destroys wall)

Water Guardian (takes water from moat and turns it into a powerful elemental) 

Flying troops or magic carpets for air mages

Burning walls or moat for fire mages (they already have fireballs offensively) etc.

Thanks for the many updates this weekend!

I'd be down with 9 tile limit if you reduced the size of buildings to 1/4. Limiting tiles makes sense but limiting buildings does not. If you want to build on a river/coast you only get 5 building tiles minus any adjacent resources.

You may not be able to pick out buildings from normal scrolling view but you can always zoom in.

Supporter of one tile city w/ limits on number of improvements.  So long as improvements are further specialized.

To address the need for the occasional "super-city", there can be improvements that grant their city additional room for improvement.  Each faction is permitted a certain number of these improvements (perhaps contingent on a Civilization tech).

I am okay with all 3 options (snakes, 9 tile, or 1 tile). One way to fix snaking if the teleportation is the only problem is to make it so that you have to walk all the way across your city to get to the other side (it counts as a road). You can still defend the city from anywhere for the sake of convenience though.

Whatever you end up deciding on, I hope I still get to physically place the buildings, even if its purely aesthetic due to the 1 tile system.  

The current plan is 1 tile cities. But we also are aware that a lot of people like having the cities in the maP in their full glory. So before we make a radical change it's a good idea to make sure we leave no stone unturned and let those who feel one way or the other have a place to express their views.

I wrote a long post a ways back about how I would do a 9 tile limit system. It would need to include upgradable tiles. Those tiles would need multiple upgrading paths. I don't see a good 9 tile system coming from using the current improvement mechanics with less space to build. People would be destroying things to build late game structures. Cities would still be able to creatively snake by settling next to a mountain or river. Use of terrain like hills and forests would pretty much go out the window. And it would still allow 3 tiles worth of teleportation. The only teleport a city should have is one that was built by the Ancients and then used for slave trafficking by the Goa'uld.