GARRISON would leave a group of fighters in a province.
There must not be any existing garrison to install your own.
Garrison , lord Feasel the Great , with 10 soldiers
The garrisons would be on guard. Their hostile/defend attitudes
would be those of the noble that created them.
Garrisons would collect some amount of gold from the local tax base.
Gold not used to pay the maintenance of the fighters would be forwarded
to the lord (the unit that created the garrison). 
A noble's status can be displayed based on how many provinces he owns:
king entire region
Initially this would just be for show. But actual uses sometimes
present themselves once a rating like this is available.
John Sloan suggests binding the land not to the noble, but instead to a
castle. Now this is very interesting, but unleashes some extra complexity.
1. This makes a castle the focus for a group of owned provinces.
Instead of tracking down the noble who owns the land, who might
be very far away, or attacking every single garrison, one can
simply take the castle.
1. One has to specify which castle a garrison is owned by, since
it might not be obvious. Another parameter to GARRISON.
2. We need an adjacent rule so that you don't bind a province to
a castle in a continent half way around the world. Something
like: there must be a path of owned provinces from the garrison
to the castle. Otherwise, the gold is not forwarded to the
3. Since the garrisons point to a castle, we can't directly say how
many provinces a noble owns (and hence, the noble's status).
If the noble stays in his castle, we can do this, but it seems
that an important noble would want to walk around showing off
his importance, not stay cooped up in a castle.
This brings us to:
4. Since GARRISON has been gunked up with this extra argument,
it's harder to use it as a general "leave some of my men here"
For instance, we would like to garrison the castle, and indicate
that the noble is the owner. The gold may or may not auto-forward.
But then we could determine how many provinces a noble owned.
The pro point seems essential. Thus we have
GARRISON [castle] <item> <qty> [...]
Garrison a province or structure with a group of men.
There must not be any existing garrison in the location,
and the character must be the first unit in the location
in order to garrison it.
Provinces must be owned by a castle. This castle must
be specified when installing a garrison in a province.
Other structures (including castles) may be garrisoned
with men to protect them. The noble installing the
garrison becomes the owner. Structures may not be owned
by other structures.
Garrisons may not execute any orders. Extra soldiers may
be given to them, or taken from them. Extra gold given
to a garrison does not forward to the castle. The men
Garrison units in a province are "on guard". They have
the HOSTILE and DEFEND attitudes of the noble that is
their ultimate owner.
Garrisoned provinces collect 60 gold/month from the local
tax base, from which they pay maintenance. The remaining
gold is forwarded to the owning castle, provided there is
an unbroken path of the provinces sharing the same owner to
Province garrisons provide a location report to the owner.
Province garrisons function so long as their is an unbroken
path of provinces sharing the same owner between them and
their castle. If the path is broken, the garrison continues
to exist, but does not forward gold or location reports to
the owner, and loses knowledge of what its DEFEND and HOSTILE
attitudes should be.
Garrison units could keep men maintenance free, but this would quickly be
abused to simply avoid paying maintenance for your men. So they have to
be paid out of their tax stream. Choosing the right amount of gold is
tricky. Typical profit should be 10-50 gold/province/month. To keep a
reasonable number of fighters, then, each garrison should collect 50-100
gold/month. Perhaps a sliding scale, where the revenue a garrison collects
increases with the number of fighters? Ick.
-- Rich Skrenta <email@example.com>