I had an idea for a movement limit a while back that is a little different than the one that M57 suggested and seems simpler to me (might be easier to code too).

Each border/territory has an additional property called "Attacks Used" or something like that.

When attacking through that border (or maybe just when capturing) that many attacks are used up instead of the standard 1 attack (only for a map with limited number of attacks of course). This could be 0, a fraction, or a whole number.

For instance:

railroad territories using 0 attacks

roads using 1/2 attack

mountains using 2 attacks

everything else defaulting to the standard 1 attack used.

Thoughts?

I don't see this as being any different than my proposal.  The bottom line is that movement ability/disability is a variable territory attribute. My understanding is that though coding the functionality may or may not be a lot of work, making the changing condition of the territory visible to the player during their turn was somewhat of a sticking point ..and a valid one.  A number of ideas that addressed that issue were discussed.  BTW, as a border feature the aforementioned issue becomes quite a bit more daunting. Besides, the feature works perfectly and cleanly as a territory attribute.

There is no question in my mind that such a feature would open up quite a few possibilities for designers.  As you suggested, the effects of terrain on movement could be mimicked. This would mostly eliminate the need for neutral walls, and put controls on the blitz problem that many existing maps suffer from.  I for one have a number of maps waiting in the wings that are all but unplayable without it.

If I were to honestly prioritize the features I would like to see Tom implement, this would be #1, even above the three I voted for in the currently active feature thread, but I'm pretty sure it would entail a bit more work than the popular proposals, and I understand that Tom's work-load is overflowing these days, so..

Anyway, there's no harm in continuing the discussion looking long-term. Assuming we have no difference of opinion regarding mechanics, visibility of the current state/value of the attribute needs to be solved.