Originally Posted by
Tynskel
As part of my PhD, I work on climate simulation modeling. Crazy crazy stuff- when I finish (years from now!) the model that I will be running will take almost a full year to execute.
Here's the thing about all this math math math:
There are many other factors involved- things that effect how the math equations are used.
Losing LOS takes equations out, being cover applies new equations, placement of units in different locations cause orders of operations.
The problem with the math hypothesis:
No data.
Computer models are completely WORTHLESS without some ground truth. In fact, there is no point to modeling unless you already have an idea of the outcome. The models I work with in climate science are painstakenly designed to incorporate checks against actual data-- CO2 ratios, plant matter, known global events. If the model doesn't match these things, it is worthless!
These math models of Warhammer are not theorizing- these are just running a bunch of table of numbers--- without realistic boundary conditions.
A better way apply the math is to use the simple equations with simple problems.
Ex.
Space Marines using Lascannon vs Twin-linked Assault Cannon against armor 14.
4 (shots)*.888 (to hit)*1/6*1/3(to pen)= 0.197 Pens
1 (shots)*2/3 (to hit)*1/6(to Pen) = 0.111 Pens
The Assault Cannon is better.
If you expand the simple equations to more complex problems, you need a series of constraints to give you a predictable result.
Ex.
Baal Predator (hv bolter sponsons) Vs Destructor w/Lascannon Sponsons against armor 14
145 point tank vs 135 point tank
4 (shots)*.888 (to hit)*1/6*1/3(to pen)= 0.197 Pens
2 (shots)*2/3 (to hit)*1/6(to Pen) = 0.222 Pens
Now the Lascannons are a better choice.
I used the point values as a baseline for comparison, types of firepower, and a specific target.
I can expand this complex situation for all armor values (which is just a series of individual equations).
Look at how complicated the situation gets:
To compare the Baal vs the Destructor, against all armor I need: 10 equations.
Note that I do not consider armor facing, terrain, ect. These are first order equations, and can realistically applied on first order problems.
But, what I cannot do is apply this math to a full game or even a turn. There are simply just too many factors involved to try to model this correctly- we are talking about the Destructor have 5 equations for shooting armor, not including LOS, and Cover. You would need to know the boundary conditions: terrain, placement, point values, armor values, ect, ect--- way too complicated. Now factor in that EVERY unit has probably ~10 equations (probably more) associated with them. (just look at the first post- there must be ~30 units: 300 equations!!!) So many boundary conditions that it becomes futile to even bother. Second, with so many factors involved, you would need to run the model many times to reach a steady state in which the results can be statistically analyzed for trends. With so many factors, we are talking about 100s of iterations (perhaps even thousands).
This just gets redonkulous. You might as well develop a computer game and get paid for it-- you'll need the venture funds for amount of time spent programming.
Essentially, the math becomes a reference tool for the 'in-game' decision making. You break down the individual turn to a series of instantaneous probabilities, and make your decisions for what to shoot (although, this is something I wouldn't recommend using all the time unless you are training for a tournament-- kinda takes away from the fun of the game).