PDA

View Full Version : Max Army size - Army size "spying"



blitzkrieg
09-15-2005, 05:59 AM
Could it be a good idea to max out an army's troop size at maybe 10,000 - at least you could try to defend against that.

Or maybe, if an army comes toward you in real life, you can get a bit of an idea about it's size. So maybe you could click on an army and get a troop range it is between - like 0-2000, 2000-4000, or larger ranges.

Of course I would then make an army of 2010 and scare opposition by making them think there is probably about 3000 troops coming (not much worry for the scripters), so maybe the range could randomise so that my 2010 could look anything like 10 - 2010 to 2010 to 4010 troops in an info window. Real life spy reports would be like that.

And on spies, if you see an army coming toward you could "spy" it's size while enroute for a cost dependant on how far away it is, or if already outside city walls you should be able to see it's troop range as described above.

Attacking armies should still have no idea of defensive strengths until they attack.

Blitzkrieg
(can I have an advance of some Geos for my suggestions ....... please)

(pretty please)

Brendo
09-15-2005, 06:17 AM
Could it be a good idea to max out an army's troop size at maybe 10,000 - at least you could try to defend against that.

No there shouldn't be any army limits.


Or maybe, if an army comes toward you in real life, you can get a bit of an idea about it's size. So maybe you could click on an army and get a troop range it is between - like 0-2000, 2000-4000, or larger ranges.

I agree, it is a good way of implementing a 'spying' technique.


Attacking armies should still have no idea of defensive strengths until they attack.

I agree once again

Also, on the war/fight summary page (I can't remember what it is called), it should have: the total amount of troops lost/the begining army size/the finishing army size.

rasqual
09-15-2005, 03:20 PM
Not a bad idea.

As ever, I'll urge that any implementation would implicate some randomness. The accuity of observation would not only be a range, but it would be a range subject to variable precision and, frankly, error.

Let's say an approaching army is 5000 troops. The report could look like:

2000 - 12500
4500 - 9000
4500 - 5050
2000 - 5000
5500 - 5900

The middle and last cases, there, are the ones that interest me most. The middle case is very narrow, and gives cause for pleasure. However, IMO such cases should be rare -- a stroke of luck in intel gathering.

The last case is just wrong. But the range is narrow, so actions based on the intel would not likely be absurdly out of proportion to what's coming. But if it's possible for a narrow range to be wrong by some margin, this qualifies one's pleasure with the middle example. It might be wrong!

This is what I'd propose -- wide ranges should give great reason for confidence that the actual troops strength lies somewhere between the upper and lower bounds. In fact, I'd say that the wider the range, the more closely the distribution of possible troop strengths should be to a Gaussian distribution. Mickey's job is to compute bounds which statistically, over time, would match that where ranges are large.

Smaller ranges, as I suggest, should leave doubt as to whether the troop strength even falls within the range. But I'd emphasize this -- they should only fall outside the range, when it's this narrow, rarely. So the doubt would itself be subject to doubt.

A further concern -- as troop strengths increase, the range should generally remain wider, and the number of narrow ranges presented should decrease a lot. I suppose the precision of such reports, given in ranges, should scale linearly but the frequency of narrow ranges provided should drop.

</nerd>