Bio Engineer Archive

---

NancyJ wrote:
Over the last week or so there has been some suggestion that the reported formula for additives is incorrect (33/33/33) so I did some tests with the meat and flora I had in my bag and this is what I came up with:

I was making BSNs with meat and flora

55 meat, 25 flora

Meat:
FL 754
PE 378
OQ 651

Flora:
FL 797
PE 159
OQ 762

After weighting the resources my total is:

FL 767
PE 309
OQ 685

add them all up and divide by 3 gives me 587

As a percentage that gives me 58.7 (rounded down is 58%)

Which is excatly the percentage I got to when I maxed my experimentation.

I'm not seeing a problem with this....

-Edit- too early in the morning

Message Edited by NancyJ on 07-08-2004 01:14 PM

---

You have 3 unknowns, %OQ, %PE and %FL, you cannot determine their values with just one equation. If I'm not mistaken you need at least 3 equations to determine the correct values.

---

NancyJ wrote:
I'm not trying to determine the percentages, there would be little value in doing so if I cant first prove that there is a problem.

---

Hmmm, well if you determine the solution then you'll know what the weights are, thiswill tell youwhether there's a problem or not. Of course if the equations aren't linear then solving the problem is much more complicated.

---

NancyJ wrote:
I'm not trying to determine the percentages, there would be little value in doing so if I cant first prove that there is a problem.

---

---

Badeb wrote:

---

NancyJ wrote:
I'm not trying to determine the percentages, there would be little value in doing so if I cant first prove that there is a problem.

---

I believe what NancyJ is trying to get from thepeople is an actual example of why people think the 33/33/33 ratio is incorrect. Her example confirms that the formula seems to be behaving as advertised, so how about folks provide a counter-example? The burden of proof is not on her.

I'm curiousto see a clear explanation myself.

---

That's what I"ve been thinking too.

I think for now at least adding in wood will just complicate the matter unnecessarily.
Do we have any official word on exactly how resources without certain values are *supposed* to work?

What we were told, back when this was changed then reverted, is that when a resource has a 0 value, that portion is based solely on the resources that do have the value.

So for a theoretical BSN:

35 Wood OQ
25 Bean OQ/PE/FL
20 Meat OQ/PE/FL

OQ = [(35*Wood OQ) + (25*Bean OQ) + (20*Meat OQ)] / 80
PE = [(25*Bean PE) + (20*Meat PE)] / 45
FL = [(25*Bean FL) + (20*Meat FL)] / 45

That is if what we have been told holds true. Then if 33/33/33 is correct:

Final Value = (OQ + PE + FL) / 3

[Edit - Typo in my OQ formula]

Message Edited by Seiryuu on 07-08-2004 06:31 PM

I ran a lot of BSN tests tonight. I will list the numbers first.

The slots in order: Organic, Flora, Creature Food.
"A" is the weighted average given 35/25/20 units respectively.
On the same line is the theoretical assembly assuming everything contributes equally.
Next is the success type, starting percentage and value followed by the experimented.

A +15.00 food and chem tool was used in combination with a +43 private food station.

---

1 OQ PE FL
O 847 767 417
F 778 572 575
M 771 261 379
A 806 581 379 -> 61.5% (78)
Great Success 16% (64) -> 66% (80)

2 OQ PE FL
O 771 261 379
F 778 572 575
M 771 261 379
A 795 479 391 -> 55.5% (76)
Great Success 13% (64) -> 60% (78)

3 OQ PE FL
O 724 866 286
F 819 518 723
M 890 560 443
A 795 681 462 -> 64.6% (79)
Great Success 17% (65) -> 69% (80)

4 OQ PE FL
O 890 560 443
F 724 866 286
M 759 810 437
A 805 718 392 -> 63.8% (79)
Great Success 17% (65) -> 69% (80)

5 OQ PE FL
O 819 518 723
F 724 866 286
M 759 810 437
A 777 700 515 -> 66.4% (79)
Great Success 17% (65) -> 69% (80)

6 OQ PE FL
O 1000 839 919
F 968 972 668
M 804 951 948
A 941 968 848 -> 91.9% (87)
Great Success 25% (67) -> 91% (87)

7 OQ PE FL
O 743 1000 798
F 743 1000 798
M 637 929 990
A 717 982 846 -> 84.8% (85)
Great Success 22% (66) -> 82% (84)

8 OQ PE FL
O 778 572 575
F 778 572 575
M 759 810 437
A 773 632 541 -> 64.8% (79)
Great Success 16% (65) -> 68% (80)

9 OQ PE FL
O 978 978 108
F 978 978 108
M 907 953 421
A 960 971 186 -> 70.6% (81)
Amazing Success 20% (66) -> 80% (84)

As far as I can tell, the results are always reproducable. If you use the same ingredients and get the same amount of experimentation, you will have the same result. (It should be noted that even though the product's percentage is two digits, the decimal value is what determines the final result.)

With the exception of (6) and (7), all actual values are slightly better than the theoretical. In (7) we see that OQ is the lowest, and the actual value is under the theoretical one. In (6) all values are very high and close enough together that the actual value matches the theoretical one.

---

If we take the weighted averaged (Avg) of any three tests and set that equal to the percentage on the same line, we can set up a 3x4 matrix and solve for the contribution of OQ, PE and FL. In this case we get 33/33/33. However, this theoretical number is not the one that we get as the final result.

Now, if we take the same weighted average and set it equal to the actual value, we can find the true weights of OQ, PE and FL. The problem is, this is not consistant at all. Some samples:

([1],[2],[3]) = (66, 60, 69) ==> OQ = 39%, PE = 33%, FL = 34%
([1],[2],[4]) = (66, 60, 69) ==> OQ = 40%, PE = 36%, FL = 29%
([3],[4],[5]) = (69, 69, 69) ==> OQ = 65%, PE = 14%, FL = 17%
([7],[8],[9]) = (82, 68, 80) ==> OQ = 49%, PE = 30%, FL = 21%

Some results came back with negative percentages. It is frustrating but not wholly unexpected since these coefficients are fluxuating between samples. It does show that either some randomness is occuring (but it cannot since we can reproduce the results!) or there is another factor we are missing.

I also tried a 9x10 matrix where I did not weight anything, I simply put each OQ, PE, FL in for three resources for all nine tests. Unfortunately this came back with negative and wildly varient numbers.

---

At this point it would appear that OQ does have a greater factor than PE or FL. I am not prepared to say how much because my numbers vary too much. There is some other factor that we are missing.

What I am going to try a new batch of tests but note the DR as well. It may not factor directly into the equation, but I wonder if it might weight either the resource or one of the resource's attributes to contribute more.

---

The conclusion I will make at this point is that each attribute does not contribute 33% for all cases.

[Edited a typo in sample #9]

Message Edited by Seiryuu on 07-13-2004 05:54 PM

I'm glad its not just me that came out with negatives after trying to solve simulatneous euations in 3 dimensions, I figured it was cos I couldnt find my calculator.

I will take a look at the results again after I've woken up a lottle more and see if I can prepare something to question the devs with.