Well, in the real world nearly all rumors are false. So, say, 90%?
Seriously:
* Each major event (see below) has a 25% chance to
generate a rumor (adjust that to get a good level of
rumors).
* A rumor has a 50% chance each month to die.
* If it doesn't die, it spreads to an adjacent province,
the When field is incremented by a month, and each field
has a 15% chance of becoming "corrupted". (In a 4 field
rumor, this means a rumor has a 50% chance of getting
corrupted each month.)
Example:
(1) Attack: Attacker, Victim, When, Where, Result.
John Garth [1301] was attacked by Felton [1517] and others
in Pesbrand last month, but fought them off!
After 2 corruptions:
John Garth [1301] was attacked by Gang of Thugs [1831] and others
in Pesbrand last month and was killed!
One thing about computer-generated rumors is that they'll have to be
about computer events. That can't be about things like peoples
intentions, because it is obvious that the computer can't know about
those. But when you get around to building a mechanism for people to
generate rumors, you could start mixing in some computer-generated
false rumors about intentions, future plans, etc.
You should also consider a SUPRESS command to suppress rumors, so that
if Dr. Pain goes in and wipes out an entire province (as I did this
turn, and you notice no one has mentioned it - proof somewhat of my
contentions about obits), then he can try to SUPRESS any word of it
getting out.
I've got rumors moving around like entities from province to province.
It gets a bit more complicated if you want them to spread like rings
in a pond. You need to spawn off a new rumor instance each time it
spreads. If you do that, you need to keep the # of rumors down so that
you don't get a big combinatorial explosion and swamp your db with
rumors. I think having 1-4 rumors in each province each turn would
be a good initial goal.
Possible Rumors
(1) Attack: Attacker, Victim, When, Where, Result, Where-Fled.
John Garth [1301] was attacked by Felton [1517] and others
in Pesbrand last month, but fought them off!
(2) Large Purchase/Sale: Buyer, Bought, When, Where.
John Garth [1301] purchased 17 Riding Horses in Pesbrand
two months ago.
(3) High-Level Skill Gain: Who, Skill, Level, When, Where.
John Garth [1301] reached Military Leadership 4 last month
in Pesbrand.
(4) Carrying Money: Who, Amount, When, Where.
John Garth [1301] was seen with over 5000 gold in Pesbrand
last month.
(5) Persuading: Persuader, [Persuaded], When, Where. There should only
be a small chance that the correct unit being persuaded is revealed.
Otherwise sly persuading becomes very difficult.
John Garth [1301] was seen bribing an unknown unit in Pesbrand
sometime recently.
(6) Terrorizing: Terrorizer, Terrorized, When, Where. I think this is
a lot more public than Persuading.
The Black Company [1301] was bullying Tyr [1501] last month.
(7) Prisoners: Jailer, Prisoner, When, Where.
The Black Company [1301] is holding Tyr [1501] prisoner in
Pesbrand!
(8) Movement of large groups: Who, To Where, When.
John Garth [1301] led 7 units toward Pesbrand last month.
(9) Large Gives: From, To, What, When, Where.
John Garth [1301] gave Tyr [1501] 1200 gold in Pesbrand last month.
Hmmm. Looking at these, you can probably abstract out a "rumor"
representation:
(1) Type of Rumor: Attack, Movement, Purchase, etc.
(2) Actor: acting unit
(3) Victim: victim unit
(4) Location: where it occurred
(5) Time: month it occurred
(6) Object: object involved
(7) Quantity: # of object involved
(8) Destination: direction of movement
Note that for, say, Skill Gained, the object would the skill and the
Quantity would be the level gained. Destination would be where a unit
was moving to, or where it was fleeing, and so on. The corruption routine
would try to introduce reasonable changes, i.e., corrupt locations to
adjacent locations, units to nearby units, quantities up or down by 10%,
and so on.
Probably would be pretty neat.
-- Scott T.