computing bonuses... new formula??

[oaktown]

For starters, no way will a formula ever give you perfect bonuses for your map; you have to look at how it will play and adjust accordingly. Still, it helps to get you going. I've never been entirely happy with the bonus calculation formulas we've been working with because...

1. I feel they over-value the number of neighboring territories and regions, without considering the size of the neighbors. For instance, in classic Australia has only one neighboring territory and region, which makes it an attractive take, right? Certainly... but since that one neighbor is huge, there's nowhere to go. Image if that neighboring region was only four territories - then it would really be attractive, because you get one bonus and jump to the next. It would make for a powerful start.

2. I feel they don't take into account just how much easier it is to grab and hold a small region early vs. a larger region. With the rising number of two and three territory regions, I think we need to be very careful about how we assign them bonuses. Each territory in a three terit bonus won't be nearly as hard to take as each terit in a 6 terit region, because in the smaller region you don't have to defend as much as you go.

So I've come with the following, which is still very much in the idea stage and I welcome you all to tear into it. It looks confusing on paper, but everything is easier in a spreadsheet.

(#terits - 2) x 5
plus
(# defending terits) x 3
plus
(# attacking terits) x 1
plus
(# neighboring regions) x .5
plus
(size of smallest neighbor region - 2) x .5
and divide the whole thing by 10.

For classic this formula gives you (not rounded):
Australia: +2.0
S. Amer.: +2.0
N. Amer.: +4.9
Asia: _7.35
Europe: +4.85
Africa: +3.5

Africa is the only bonus that doesn't round to what it should, but check it out: 3.5. I've always felt that Africa is a borderline +3/+4 anyway, because you rarely win by making Africa your base.

[InkL0sed]

"Borderline" is the word... even the numbers can't decide!

[benny profane]

awesome, many thanks oaktown!

snap and i will run our numbers through this, see what we get.

(our biggest issue remains, however, what to do with bonuses within bonuses)

[pepperonibread]

Stickify it!

[oaktown]

Stickify it!

I'd like to run some more maps through it first, see how it comes out... there has to be a glitch somewhere!

[BaldAdonis]

there has to be a glitch somewhere!

Somethings going wrong with the formulae. I've got Aus and SA with
(terits, defend, attack, neighbours, smallest neighbour)
Aus = (4, 1, 1, 1, 12)
SA = (4, 2, 2, 2, 6)
and raw numbers, before dividing by 10:
20 + 3 + 1 + 0.5 + 5
20 + 6 + 2 + 1 + 2

Neither of which will get to the numbers you came out with, for any combination of +/-. Maybe my understanding of the definitions is different?

[Coleman]

I got the same numbers as BladAdonis.

[onbekende]

I like mine and I already look at how the continent lies in respect to the rest, meh

it gives around the good numbers for every map recently made, it normally has enough math build into it to keep it that way (whatever crazy thing you people make, I am sure I can alter my formula to suit it)

[DiM]

i want to add just 1 thing. no matter what formula you use the numbers will need adjustment afterwards and this can only done by the human brain not by a calculator. because the calculator doesn't look at the overall map and gameplay possibilities. for example look at land's end in british isles. 1 border +2 bonus. just like australia. seems fine and it is. but what if when you get out of that continent you bump into a similar one? and you can hold both for 1 border? will you still give each of them a +2 knowing both can be easily held with 1 border?

[oaktown]

there has to be a glitch somewhere!

Somethings going wrong with the formulae. I've got Aus and SA with
(terits, defend, attack, neighbours, smallest neighbour)
Aus = (4, 1, 1, 1, 12)
SA = (4, 2, 2, 2, 6)
and raw numbers, before dividing by 10:
20 + 3 + 1 + 0.5 + 5
20 + 6 + 2 + 1 + 2

Neither of which will get to the numbers you came out with, for any combination of +/-. Maybe my understanding of the definitions is different?

Yep, my bad... i typed it up wrong. That's now how I did it at all. Should be (# terits -2) x 5.

[yeti_c]

I'm with DiM here - you need a "expandibility" quotient...

C.

[oaktown]

at the top of the thread I made it very clear that no formula will ever be perfect. I came up with this new formula precisely to give some weight to expandability - in the example that DiM gave, Australia would only compute to 1.55 instead of 2.0 if the neighboring continent was a 4 as well.

What we really need is something like the way the NCAA calculates RPI rankings - your rank is based oon your stats as well as those of your opponents, and their opponents, etc.

[DiM]

at the top of the thread I made it very clear that no formula will ever be perfect. I came up with this new formula precisely to give some weight to expandability - in the example that DiM gave, Australia would only compute to 1.55 instead of 2.0 if the neighboring continent was a 4 as well.

What we really need is something like the way the NCAA calculates RPI rankings - your rank is based oon your stats as well as those of your opponents, and their opponents, etc.

wtf is NCAA and RPI

[InkL0sed]

NCAA is the college sports organization in the US (I think it stands for National College Athletics Association?). RPI is some basketball stat, I guess. Some sport I don't follow, anyway.

[oaktown]

the RPI rankings are fascinating. It considers a team's record against other teams, the record of the teams they've played, and the record of the teams that those opponents have played.

Say Team A plays Team B. Team A is ranked nationally based on its own performance, the performance of Team B, AND the performance of all of the teams that Team B has played, even if Team A never played those teams. This gets tricky because everybody's rankings are circular, depending on the success of teams that are ranked based on the success of the first team.

My head is spinning just trying to describe it.

How this would apply to CC is as follows:
1. Each continent is given a baseline value using # terits, # defendable borders, # attacking terits, and # neighbor regions.
2. The baseline value of the regions neighboring each region is somehow applied to each baseline value to calculate a final value.

Thus Australia's value is based on it's own baseline value, as well as the values of it's neighboring regions. A continent like Australia, for example, with four terits and one border, is given a baseline value of 2.0, but it could be brought up a tick because Asia's baseline value is so large. Meanwhile, a continent like Australia whose only border was another Australia would see its final value brought DOWN a tick because its only border's baseline value is so small.

My "beta" formula, above, takes into consideration neighbor size, but not the other features of the neighbor.

[DiM]

the rpi thing sounds neat. make that into a formula so we can do some testing.

btw i just thought of another thing formulas don't take into consideration. special features of terits.
for example the dustbowl map where some terits get -1
or let's say a continent has a terit that can bombard all other terits. that makes it extra powerfull but the formula doesn't take that into consideration.

[oaktown]

Here's the excel file, if anybody wants to play with it.

http://h1.ripway.com/rparodi/BonusCalc_Oaktown.xls