Q: Leftover points - mineral concentrations
Thore Schmechtig
Hi,
the help file says that if you spend leftover points on mineral
concentrations, this will raise the _worst_ concentration by the amount
possible with the spent points.
Now I have joined a team game and had a look on my partners situation. I
took improved concentrations for 42 points, he did not. Normally, I should
be able to have a look at his concentrations, check what mineral has the
lowest conc, then add 14 to this (42 / 3 = 14) and get my concentrations,
right?
Yet this is the _real_ situation: Compared with his homeworld, mine has...
+ 11 on Ironium concentration
+ 32 (!) on Boranium concentration and
+ 11 on Germanium concentration!
So does the help file give incorrect information on this one? I always
thought that (except that concentration bonus by leftover points) all
homeworlds start with the same concentrations?!
P. S. just in case it makes a difference: My partner's homeworld has the Bor
and Ger concentrations on an equal level (both have the same concentration).
*** End of the mail, please stop reading now ;) ***
Greetings from Thore Schmechtig (thore.sc...@metronet.de)
"Every light casts a SHADOW... especially on Z'ha'dum!"
Jason Cawley
Thore Schmechtig wrote:
> the help file says that if you spend leftover points on mineral
Is wrong or old.
> Yet this is the _real_ situation: Compared with his homeworld, mine has...
> + 11 on Ironium concentration
> + 32 (!) on Boranium concentration and
> + 11 on Germanium concentration!
Real thing you get is + 1/4 times points spent to 2 concentrations and +
3/4 times points spent to lowest one. This fellow spent 43 points or so
(or spent 44 and deplete bora 1 already perhaps, etc).
In all, this gives around 10,000 kt all told of extra minerals in the
rock above concentration 30 for typical starts, for around 40 points
invested. Depending somewhat on what the starting concentrations would
otherwise be, of course.
HP races (great factories, poor population) will often spend on this in
order to get a decent G con on the HW no matter what. Since HPs are
more dependent on G cons than other races, this is a useful "insurance
policy". Can often pay for it e.g. with 1 fewer factory operated.
Personally I think it's a great deal at the price and recommend it. It
may be more people don't because of the old info in the help.
Sincerely,
Jason Cawley
Martin Dermody
Thore Schmechtig wrote:
> P. S. just in case it makes a difference: My partner's homeworld has the Bor
> and Ger concentrations on an equal level (both have the same concentration).
I asked Jason recently about when the minimun concentrations tied. Had
not tested, did not know.
I checked with a testbed. It appears that when the min mineral con is
tied you get the larger increase in the order I, then B, then G.
Method:
What I actually did was put 16 races in a tiny universe. 1 was a control
with no pts to mineral con. The other 15 all had 50+ points to mineral
concentrations. When the control showed tied minimun concentrations I
checked all 15 other races to see what they got. In every case the got
the same minerals as each other. There was never a case of one race
getting the larger increase of minerals in a different mineral than the
others. I generated enough universes to get 8 (results not in front of
me, from memory) or so ties. The G was never got the larger increase at
all. The I always got the larger increase. (for (I,B),(I,G), and (I,B,G)
ties). So my conclusion could be wrong from random chance. This was with
current H patch.
As Jason observed... it's nice insurance for low concentrations.
Martin
Robert Croson, Jr
I think that Jason Cawley <jason...@postoffice.worldnet.att.net>
may have written <6ejscn$b...@bgtnsc02.worldnet.att.net>:
> HP races (great factories, poor population) will often spend on this in
> order to get a decent G con on the HW no matter what. Since HPs are
> more dependent on G cons than other races, this is a useful "insurance
> policy". Can often pay for it e.g. with 1 fewer factory operated.
>
> Personally I think it's a great deal at the price and recommend it. It
> may be more people don't because of the old info in the help.
I did this once and got burned pretty bad.
All mineral concentrations were pretty low, with Boranium being the worst.
The extra points, somewhere around 20, didn't even get the Germanium
concentration above 30. :(
Oh well. You win some you lose some.
--
Rob O- rcr...@arcm.com Streamer on Starlink!
Martin Dermody
Robert Croson, Jr wrote:
> I did this once and got burned pretty bad.
>
> All mineral concentrations were pretty low, with Boranium being the worst.
> The extra points, somewhere around 20, didn't even get the Germanium
> concentration above 30. :(
Was this recent? Could there be a change from a previous version. The
result I saw indicated you should have gotten raised to 35 if both B & G
started at 30 con.
Martin
Robert Croson, Jr
I think that Martin Dermody <der...@u.washington.edu>
may have written <350D8E...@u.washington.edu>:
If I remember, I sepnt around 20 some points on mineral concentrations,
and had Germanium and Boranium at around 20% or so. Ironium was around 50%.
Boranium got the big boost. Germ was left at about 25%.
Normally spending a good amount of extra points does give you some
insurance on mineral concentrations. But soemtimes the insurance
company doesn't give you the full value of your policy. :)
Martin Dermody
Robert Croson, Jr wrote:
> Normally spending a good amount of extra points does give you some
> insurance on mineral concentrations. But soemtimes the insurance
> company doesn't give you the full value of your policy. :)
OK... My second thoughts and some math will show through here.
Let's assume my previous post about tied concentrations is correct. Then
since I do not really know how the random mineral concentrations are
chosen I'll use this model for calculations: We have a spinner for each
mineral with 100 concentrations it can stop on. Values 1-30 turn into
30. (Just to get some numbers)
So (.7)*(.7) of the time I>30 or B>30 so only minimun boost for G 51% of
time to start.
Since (.3)*(.3) of the time both I and B are at 30 con and
(.5)*(.51-.09) will Icon>Bcon we see I=30%, B=21%, andG=0% from this
part of calculation.
For the 49% of the time when Icon>30 & Bcon>30 we get 1/4 Gcon > Icon &
Gcon > Bcon. We get 3/8 Icon > Gcon & Icon > Bcon.
Combining these we have :
Boost Icon the most 9% + 21% + .18% = 48%
Boost Bcon the most 21% + 18% = 39%
Boost Gcon the most 12%
These ignore cases when mineral concentrations are equal but above 30
con. They are rounded. They assume my spinner model. Actual values will
vary. Anybody know how the numbers really come out?
Now, that said if you want increased Icon this looks like a good plan.
If you want increased Gcon above the 1/4 addvantage points spent... 12%
aint that hot when I look at it.
Martin
Jason Cawley
Martin Dermody wrote:
>
> Robert Croson, Jr wrote:
But soemtimes the insurance
> > company doesn't give you the full value of your policy. :)
Well, I think the "under 30 even with points spent" is old data from
before some change. I have never seen it in any recent test. If I
spend 36 points, I get 39 at a minimum, etc.
> OK... My second thoughts and some math will show through here.
OK. You missed a case or two I think :-) Looks a lot better than you
think.
> Values 1-30 turn into
> 30. (Just to get some numbers)
Reasonable assumption, seems to me.
> So (.7)*(.7) of the time I>30 or B>30 so only minimun boost for G 51% of
> time to start.
Um, no :-)
Have to look at the whole "decision tree", with probabilities for each
"node" (each "if-then" statement). Multiply probabilities down the
branches; add the probabilities for each ending "leaf" of the branches.
Thus -
*If* G con = 30 (p = .3)
Then
if I con = 30 (p = .3) Note - overall only .09
G gets a minor 1/4 boost
*else* (p = .7)
if B con = 30 (p = .3)
G gets a minor boost
else (p = .7)
G gets a major, 3/4 boost
That half of the branches, on its own, gives 18.3% chance of a large
boost to the G con after a 30 initial G con draw.
Else G con > 30 (p = .7)
If I con <= G con (p ~= .3 + 35/69 * .7) (35/69 for ties)
G gets a minor boost
Else (p = 1 - .3 - (35/69 * .7))
If B con <= G con (p ~= .3 + 35/69 * .7)
(note that this case has only 15.8% *overall* probability)
G gets a minor boost
Else (8.4% overall and 12% of the second part of the tree)
G gets a major boost
Total chance that G gets a major boost is .183 + .084 = .267
Or a little better than 1/4, compared to 1/3 for completely random. Not
a large effect, then, for the tie rules. Some, but not huge.
Plus, you always get the + 1/4. And more likely to have the G be the
thing raised 3/4 in the event of an initial G = 30 draw.
Say you spend 40 points. Then minimum G con = 40 (assuming Rob's case
is old data of course). And what is the chance of drawing exactly 40 G
con? Same sort of tree gives 11.7% or call it 1/9. To get under a 60
starting G con, G has to be the lowest *and* the G con drawn has to be
under 50. Works out to less than 1/3, including the 1/9 "get 40"
chance.
So, recapping. Assuming you always get over 30 and the draws work
according to your model, with a 30% chance of an initial draw 30, etc.
Then if you spend 40 points on min cons, your G con probabilities are
going to look like this -
~1/9 only 40 G con
~1/5 over 40, under 60 G con
~2/5 decent G con drawn (you and others), with 10 added to it
~1/4 + 30 to G con
So 30-40 results are about 1/4th as likely as normal (and always 40 not
30 or 30's), 41-59 results about 1/2 as likely as normal; you get a 1/4
chance of +30 and always +10.
For one factory operated (or one more tech field +1 expense level, etc).
That sure looks like reasonable insurance for HPs to me :-)
Also will help iron-poor starts of course, so using privateer colonizers
will generally be feasible for you, while sometimes less so for others.
Hope this is useful.
Sincerely,
Jason Cawley
Jason Cawley
Jason Cawley wrote:
(slight correction)
> That half of the branches, on its own, gives 18.3% chance of a large boost to the G con after a 30 initial G con draw.
Should be 14.7% of course - .3 * .7 * .7.
> Else (8.4% overall and 12% of the second part of the tree)
Really works out ot 8.3% not 8.4% with the 35/69's included.
Gives total chance of the major boost to G .147 + .083 = 23%.
Still about the 1/4 indicated below, but not quite so small an effect
from the tie cases.
> ~1/9 only 40 G con
> ~1/5 over 40, under 60 G con
> ~2/5 decent G con drawn (you and others), with 10 added to it
> ~1/4 + 30 to G con
Essentially stands since the 26.7 vs. 23 are both "about 1/4" as used
here.
Just trying to be careful...
Sincerely,
Jason Cawley
Martin Dermody
Jason Cawley wrote:
> OK. You missed a case or two I think :-) Looks a lot better than you
> think.
>
Oh Oops. Looking back at my napkin from lunch I ONLY left out the case
where (Gcon=30,Icon>30,Bcon>30). A meer 15% of the total cases. And all
of them ones where the Gcon gets the biggest boost. This made the other
figures larger in the part this was left out from.
Thank you Jason.
Martin
Andrew Turner
>HP races (great factories, poor population) will often spend on this in
>order to get a decent G con on the HW no matter what. Since HPs are
>more dependent on G cons than other races, this is a useful "insurance
>policy". Can often pay for it e.g. with 1 fewer factory operated.
One little problem with this idea, which I found in a testbed last
night.
My initial G concentration must have been *atrocious*, (e.g. "5")
because after spending 36 points on mineral concentrations, I still
had only a concentration of 32 (I and B were in the 60s and 70s, so G
definitely got the larger part of the increase). But 32 isn't
significantly different than the HW effective minimum concentration.
--
Andrew Turner.
Andrew Turner
>My initial G concentration must have been *atrocious*, (e.g. "5")
No, I take it back.
Just ran 30 test universe creations. G Concentration was never below
30, but suspiciously often at precisely that level.
I'll have to look at last night's testbed again to see what I was
*really* doing... can't have been leftovers to mineral concentrations,
can it?
--
Andrew Turner.
Jason Cawley
Robert Croson, Jr wrote:
> Could be. It was a while ago. Maybe under the f or g version?
Donno. Did a test of min con spending on starting cons now. Didn't see
any results under 40 with 40 points spent on min cons (+10 minimum,
applied to 30 min draw).
I did 10 trials with and 10 without Acc BBS (didn't expect that to make
any difference, and no indication it makes any differece) with no points
spent on min cons and with 40 points spent. 40 trials in all, only one
race in each.
Here are my test results -
no points spent, std start
56/30/87, 30/78/58, 60/85/30, 30/98/62, 80/94/103,
40/111/60, 53/84/46, 30/61/87, 72/76/110, 73/64/30
no points, Acc BBS
30/84/78, 30/30/30, 30/68/66, 67/84/84, 30/87/83,
30/73/69, 30/30/88, 81/104/87, 90/76/94, 30/51/30
40 points, std start
100/74/60, 104/60/40, 76/60/103, 55/60/58, 60/98/87,
84/110/91, 98/77/79, 60/40/91, 75/60/40, 77/79/60
40 points, Acc BBS
92/119/87, 106/77/84, 120/67/60, 99/69/60, 60/83/40,
83/81/76, 60/92/40, 97/85/85, 60/80/40, 97/97/88
Some conclusions - the minimum draw is 30; that occurs about 30% of the
time (28.3% ~= 17/60 for the 0 point tests e.g.). The added points are
added after the 30 minimum "floor" has been applied. Thus the minimum
40 observed with 40 points spent (and all the "60" draws in that case
too). Natural draws over 100 are rare but do happen - my guess is
something like 95 or over drawn, add some random number of points to it
- like 1-15/20 say.
Other overall stuff from the test -
average con with no points added: 63.7 (55 expected w/ 30 floor)
average con with 40 points added: 76.7 (72 expected)
observed under 50 with no points: 18/60 = 30%
60 23/60 = 38%
Note - better than expected; good draws in other words.
" " 50 with points: 6/60 = 10%
60 8/60 = 13%
Big reduction in low con draws, even compared to good draws without the
added points.
Worst out of the 20 trials for each -
No points - 30/30/30
40 points - 75/60/40
Worst out of the 20 w/ 40 points was better than the expected 55 average
for no points starts.
Things still unclear - how those natural over 100 draws work - see a
111, for instance. I doubt it has much effect on average expectations,
though, whatever it is.
Certainly looks to me like a good deal.
I hope this is useful.
Sincerely,
Jason Cawley
Robert Croson, Jr
> Martin Dermody wrote:
>>
>> Robert Croson, Jr wrote:
>
> But soemtimes the insurance
>> > company doesn't give you the full value of your policy. :)
>
> Well, I think the "under 30 even with points spent" is old data from
> before some change. I have never seen it in any recent test. If I
> spend 36 points, I get 39 at a minimum, etc.
Could be. It was a while ago. Maybe under the f or g version?
--