Am I missing something, or is it really only in pretty rare cases that it's worth bribing a non-barbarian enemy unit?

(Number of shields required to build unit) * 2 + n

where
n=0 for shields =< 20
n=1 for shields =<30
n=2 for shields > 30

So a Barb Horseman would cost 41 g and a Legion 82 g.

The AI unit costs are more complicated so the given link would be worth while following up. Variables are distance of the AI unit from its capital and the size of the AI civ's coffers, among other things.

1. Distance of AI unit from its capital
2. The amount of gold the civ has in its coffers
3. Of course the shield-cost to build the unit

I can add to the exceptional cases though.

Sometimes a unit is found wandering far from home and can then be a cheap buy. (A point that may play a part in Tonic's preference for caravan heisting ).

If you have Leo's, and there is a very small backward civ around, it's antique units will be cheap and will soon turn into something more useful. In fact, in the extreme case of a civ still producing warriors in the midst of armour/howies/mobile infantry (which I've seem two or three times) you can work a variation of the recruiting station outside a barbarian city tactic. Warriors are always a cheap buy as compared to riflemen (so I am sure that Tonic is right to think that, like the bribe cost of a barb, initial unit price is part of the formula). In fact, even without Leo's the price is worth it if the units are always "none" because you can use them in the cases where you occupy choke points and have need of a second unit to double up as protection against bribery. I often find myself tying down diplos and spies that way and an antique "none" warrior will free those units up.

Warships are good tagets. No doubt that's partly because they are often far from home but can also be because a single powerful warship can decisively change the balance of power in a particular theatre. I've cheerfully paid a good four figure sum for a battleship before now - and picked up several cities to expand a tenuous beachhead as an immediate result. This is particularly true if you are a democracy, don't have Shakespeare's Theatre and the nearest coastal town to the bribe site belongs to the AI. Once bribed the ship will be a "none" and that's very valuable in such conditions. On some maps a merchant ship is worth paying a high price for in such a case.

Sometimes an antique explorer will turn up in a remote location late in the game when his ability to work well with alpines makes him a good buy.

Lastly, there are cases where the fact that the bribed ex-AI unit gets to move in the turn it's bribed (a familiar point in the context of bribing barbs of course) can be enough to decide an otherwise close battle and thereby prevent a potentially costly counter attack.

A point of technique which this thread has made me realise I have been overlooking is that if the AI's bank balance is a factor in the price then, during the period in the game when it can be done, it would pay to try to extract tribute before sweet-talking the unit. A point which probably applies in the commoner case of bribing cities too.

An oddity that I've noticed is that an approach to bribe settlers produces the most exorbitant prices of all. I've seen some four figure demands. I suspect that partisans are also a premium price. Quite appropriate, I think, that those units should be more loyal.

quote: Originally posted by Blaupanzer on 11-08-2000 01:13 PM Distance from AI capitol does figure in the bribe

Hmm, and i have noticed that on round worlds, the computer can hardly recognize that the world goes on...

1. There is a minimum distance from the AI's capital beyond which the gold required to bribe a particular unit remains the same (minimum). This is approximately 15 squares from the capital.

2. The bribe-amount is proportional to the gold in the civ's coffers. For the case of the minimum (when the unit is > 15 squares from its capital), it is given by the equation:

G = M*S

Where
G = gold required to bribe
S = shield cost of the unit,
and M is given by

M = 2.05 + 0.0028*C

where C is the amount of gold in the coffers.

So the least you would pay for bribing eg an AI Rifleman (40 shields) would be the same as for a Barb Rifleman when the civ has zero or close to zero gold in the coffers ie 82 g. But when the coffers are at 1000 g say, it would be 194 g.

The effect of distance I've only superficially tested. For an arbitrary treasury of 442 g:
2 squares from capital, bribing Rifleman costs 596 g
16 squares away (the minimum cost) = 132 g

The bribe cost gets greatly inflated when the civ is really rich eg the corresponding figures for a tenfold increse in the treasury (4420 g) are
2584 g, and
574 g

From Blaupanzer's post I take it that the number of squares is counted in a line from the spy/diplo through the target to the capital?

Because of the similarity to the position with bribing cities I'm now starting to think the discount in favour of spy over diplo and vet spy over greenie will also hold.

Confirmation that the civ's bank balance is critical to the price will have immediate practical application
It occurs to me that this knowledge increases the value of Marco Polo's and/or the value of establishing an embassy.

quote: <font size=1>Originally posted by East Street Trader on 11-09-2000 06:07 AM</font> From Blaupanzer's post I take it that the number of squares is counted in a line from the spy/diplo through the target to the capital? Because of the similarity to the position with bribing cities I'm now starting to think the discount in favour of spy over diplo and vet spy over greenie will also hold.

There is an important distinction between bribing an AI unit and an AI city. When bribing a city it is important to place your diplo/spy at a square furthest away from the AI's capital to minimise the cost. In bribing a unit, it is the position of the unit (and hence its distance from its capital) that determines the bribe cost and it makes no difference on what square your diplo is!

Hard to figure how there can be a distinction between horizontal and vertical though.

I am using ToT, which could obviously account for numerical differences. When I first tested my numbers, I gave the AI 750 and 1000 gold and found the bribe cost of both of those amounts for a riflemen.

I set up the equations, using tonic's variables:

G = (y + x*C)S
204 = (1000x + y)40
176 = (750x + y)40

The numbers I dervied were:

x = .0028
y = 2.3

Beforehand, I had gone through and gotten G for when C equaled 0, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 175, 200, 250, 500, 1000, 1500 and 2000 gold. I plotted these on my graphing calculator and then tried ot fit my previously derived equation [ G = .0028*C + 2.3)S ] to those points, but the line only fit the two points I had used to get the equation.

If anyone can tell me how I can fit an equation to this I would be very grateful... I can't remember the "line of best-fit" command for my calc, but I don't think that is what I want anyways (it's a TI-83 plus).

I found a couple other things though.

• Government has NO effect on the price to bribe..
  
• size of the capitol has NO effect on the price to bribe.
  
• I found that, on a straight line (squares with touching sides), when the unit is greater than or equal to 10 squares away, distance from capitol no longer has an effect. Square 1 is the most expensive, and it decreases until you reach square 10, and from there on it stays the same for as far away as you go.
  

For anyone interested, I attempted to find if there was a common point for the x term. I used the equation:

x = (G/S - y)/C

Since I already had G, S, and C, I tried it. I used random numbers that I had already derrived, but this way proved inconclusive. The numbers from 10-100 crossed at the same point, but then all the numbers after that began crossing at different points. Could there be another variable that chages with C? Say, when C = 0-100, then z = x(1), or when C = 101-200, then z = x(2).

Good job!

I would never had figured that out by myself.

G = M*S
Where
G = gold required to bribe
S = shield cost of the unit,
and M is given by
M = 2.05 + 0.0028*C
where C is the amount of gold in the coffers.

I want to present a more precise (i hope so...) formula that includes the distance from a capital and to add a formula for bribing cities:

Formula for bribing units
G = DF*GF*S
Where
G = gold required to bribe
S = shield cost of the unit (settler and engineer considered as 80-shield unit)
DF = 3*12.5/(D+2) (distance factor)
GF = (1+4*EG/3000) (gold factor)
EG = enemy's gold
D = the distance (of enemy's unit)from the enemy's capital (maximum 16)

Formula for bribing cities
G = DF*GF*S
Where
S = size of the city
DF = 5*200*SF/(D+3) (distance factor)
SF = 5/6 for a spy; 1 for a diplomat (spy factor)
GF = (1+EG/1000) (gold factor)
EG = enemy's gold
D = the distance (of your dip/spy) from the enemy's capital (maximum 16); (bribing from distance 1 is not allowed)

Is it correct?

debeest, I use dips early for emergency non suppported defence. Good as you say for barbs. Good as 2nd unit in fort(no maintenance,bribe defence\bribe offense) take in trierme to buy cheap unsupported triermes and caravels, AND every city that has a dip is more resistant to bribe/theft/whatever from enemy spies.
I do not make a dip the cornerstone of my defence as you suggest some people do, but if enemy attacks an unprepared city with excessive force, buying 1 or 2 unstacked units to kill the stacked units is quite a viabe tactic.
Other Ideas but I gotta go

Merry Holidays,

quote: <font size=1>Originally posted by La Fayette on 12-27-2000 05:27 AM</font> Slowthinker It seems that your formula for bribing cities is essentially the same as the one I have in mind: S(T+1000)/(D+3), where S=Size of the city T=Treasury of the civ D=Distance to the capital (measured from the dip who is attempting to bribe). Have you tested it? Have you tested 33% discount for vet spies? and 50% discount for city in disorder or empty or previously owned by your civ? What about Dmax? You wrote 16. I have 32 in mind (10 for communist cities).

I found the same information as you on Winkler's page "Scrolls of Ancient Wisdom"(http://ltswww.epfl.ch/~winkler/civ2.html). There was formula for bribing cities only ,not units. Do you have another source?

When I sought formulas (by testing through cheat menu), i didnt know about Winkler's link, so you may consider that formula for cities is confirmed from two sources.
I tested 50% discount for city in disorder or empty, its all.
I think that Dmax=32 is a mistake on Winkler's page. Only elucidation is it depends on the size of a map, but it would affect tonic's constant 2.05 in this thread, but everybody agreed with him and I agree too.
Or Winkler counts some jagged way?
I think definitely there is a mistake in "Scrolls of Ancient Wisdom". I tested communism now, Dmax is 10 then, indeed, so 32 for other governments should be a mistake.