The Scrolls of Wisdom are wrong! You wanted examples?

[debeest]

OK, I've claimed a bunch of times over the past few months that the Scrolls of Wisdom are wrong about things like city bribing cost and caravan payoff. Several times, people have asked me for examples, but I haven't had them at hand. In my last game I kept careful track of some things.

I bribed several cities, all from the same civilization, all within a few turns. Here's the data. I hope it will be readily decipherable; I'm just typing it in online.

government: republic, throughout
gold: 1319, 1100, 948, 716, 508, 254
distance: no capital, so 35 throughout
size: 12, 10, 2, 7, 12, 4
courthouse: none, throughout
disorder: no, no, no, no, yes, no
empty?: no, throughout
ever mine?: no, throughout
using a spy?: yes, throughout
veteran?: no, throughout

actual cost: 1220, 917, 170, 525, 450, 220
Scrolls cost: 662+, 500, 92+, 286, 143+, 119

[edit: well, I see that my columns didn't come out, so I put in commas instead. Each line reads across the values for six cities.]

By first checking the price, then wiping out the defensive units, I did verify that the 50% price reduction for an empty city is correct. By checking one city's costs with both veteran and non-veteran units, I found that the cost dropped from 450 to 316, which does not fit with the Scrolls claim or with any obvious simple calculation.

In every case, the Scrolls predicted cost was about half of the actual cost, but not exactly half or any easily calculable ratio. There was no apparent relationship to the presence of city improvements or units within the city or supported units elsewhere, except for courthouse.

Incidentally, I also tried bribing some units. On the same turn, some partisans in one area cost 115 gold, but partisans in a different place cost 150, even though they were from the same civilization and, since they had no capital, should have been "equidistant" and thus equally expensive. A rifleman with no capital was priced at 254 gold, a difference out of proportion to the relative shield cost of the two units.

I also tracked trade payoffs. In all cases the trade was a demanded product between two of my own cities on separate continents, no airports, no superhighways:

1802, caravan, silk, trades=47+36, distance=5, no railroad yet: Scrolls predict bonus 103.75, trade 10.875 -- actual payoffs = bonus 93, trade 7

1871, freight, oil, trades=27+39, distance=13, railroad but no flight: Scrolls predict bonus 80.5, trade 17.5 -- actual payoffs = bonus 296, trade 4

1871, freight, gold, trades=45+43, distance=13, railroad but no flight: Scrolls predict bonus 56, trade 23 -- actual payoffs = bonus 306, trade 5

1872, freight, silver, trades=58+26, distance=21, railroad but not flight: Scrolls predict bonus 108.5, trade 22 -- actual payoffs = bonus 310, trade 4

1874, freight, silk, trades=49+47, distance=13, railroad and flight: Scrolls predict bonus 61.33, trade 25 -- actual payoffs = bonus 204, trade 5

1874, freight, gold, trades=28+40, distance=16, railroad and flight: Scrolls predict bonus 49.11, trade 18 -- actual payoffs = bonus 172, trade 4

1875, freight, oil, trades=41+57, distance=7, railroad and flight: Scrolls predict bonus 46.27, trade 25.5 -- actual payoffs = bonus 198, trade 6

Note that the Scrolls always predict that the freight trade bonus will be a substantial fraction of the one-time bonus, but it never actually is.

I may have introduced errors by, for example, multiplying by 1.5 where I should have added 50% instead, but my experience tells me that neither method works.

So, I'm still wondering: can anyone actually take apart the code and determine how these things are really calculated?
[This message has been edited by debeest (edited November 20, 2000).]