Extractor Trick Economically Inferior

[GoogleCodeExporter]

The simulation uses the 10/10 single extractor trick. Based on real-world 
results, 9OL provides 6 more minerals at 2:15 than single, and 17 more than 
double. I understand the simulation does not include real-world variables so 
the extractor trick may seem to provide a mineral boost, however if we accept 
that 9OL is always the best econ build in real-world environments removing the 
extractor trick might provide even more accurate data.

Source: http://www.teamliquid.net/forum/viewmessage.php?topic_id=133917 

Original issue reported on code.google.com by MPers...@gmail.com on 22 Oct 2010 at 4:08

[GoogleCodeExporter]

Economy is not always the goal. Sometimes the limiting factor in a build is 
larva count, or early timing, or a fast queen, all of which can be supplemented 
for by extractor trick->overpool.

That being said, not wanting to extractor trick could become a feature, along 
the same lines as limited bases.

Original comment by Frit...@gmail.com on 22 Oct 2010 at 5:07

• Changed state: Accepted
• Added labels: Priority-Low, Type-Enhancement
• Removed labels: Priority-Medium, Type-Defect

[GoogleCodeExporter]

The current implementation already implements the economic disadvantages of 
doing an extractor trick.

It still uses them because you can spend otherwise wasted larvae with them.

Original comment by azzur...@gmail.com on 23 Oct 2010 at 2:55

[GoogleCodeExporter]

Economically 9 OL is NOT the best it is actually the worst.

9D + OL + D + 2D + D gains ~100 less minerals than 10D + OL + 3 D
10 D + Extractor Trick + D + OL + 3D is better than the 9D OL combo by ~30 
minerals, but still worse than the 10D OL combo.

Original comment by rpanu...@gmail.com on 3 Nov 2010 at 2:17

[GoogleCodeExporter]

How are you coming to that conclusion, worst by 100 minerals really that doesnt 
even seem remotely possible.
I just ran 2 games on the same starting position @ steppes.

Both builds got the 9th drone @ 31 s, here are the rest of the timings:
10OL:
9 drone @ 31s
10 drone @ 39s
13 drone @ 83s

9OL:
9 drone @ 31s
10 drone @ 58s
12 drone @ 74s
13 drone @ 80s

9OL ends up having 3 more seconds of mining time. It also doesnt waste any 
larva spawn time I ended up getting the 14th larva 6s before with 9OL than 10OL.

Original comment by g.c.tave...@gmail.com on 3 Nov 2010 at 11:31

[GoogleCodeExporter]

9OL may be the the best at 2:15, but the point of overpool with or without 
extractor trick is to get a pool out quickly in order to queen faster, giving 
an economic boost at a critical time. Sometime shortly after 13/14, you become 
larva capped. That is my understanding, but I'm not able to give support at 
this time.

Original comment by Frit...@gmail.com on 3 Nov 2010 at 3:19

[GoogleCodeExporter]

I came to that conclusion simply by doing it.  Going to a 15 drone start as is 
what most people build on unless of course you're going a non-econ opening 
which is pretty silly to discuss in an econ thread.

If you go 14 drone base just to add up the minerals in a game for all 3 of 
those openings you will see exactly what I mean.  Play a game, watch exactly 
what point you're at when the 14th drone pops.  You'll see exactly the same 
result I did.  (14 start is a pretty basic start if you aren't 6/10 pooling)

Original comment by rpanu...@gmail.com on 4 Nov 2010 at 12:32

[GoogleCodeExporter]

Did the tests again this time up to 15 will try to make it easier for you to 
understand.

10OL:
9 drone @ 31s
10 drone @ 40s (1s slower than first time)
13 drone @ 82s (1s faster)
14 drone @ 98s
15 drone @ 113s

9ol:
9 drone @ 31s - same speed as 10ol
10 drone @ 57s - 17s slower than 10 OL thus -17s of mining time compared to 10OL
12 drone @ 74s - 8s faster than 10 OL thus +16s of mining timing compared to 
10OL (-1 total)
13 drone @ 80s - 2s faster than 10OL ... (+1s total)
14 drone @ 94s - 4s faster than 10OL ... (+5s total)
15 drone @ 109s - 5s faster than 10OL ... (+10s total)

Your test isn't fair because 10OL gets drones slower due to slowing larvae 
spawn so you can't use population as a timing.

Original comment by g.c.tave...@gmail.com on 4 Nov 2010 at 1:21

[GoogleCodeExporter]

Okay, try this.

Overpool at 24 drones
11 pool at 24 drones
12 pool at 24 drones
13 pool at 24 drones
14 pool at 24 drones

And the trick is this. Get a queen as fast as possible on top of it.

Overpool is about long term larva production, not drone production. The result 
in this will be drone production, but you get the point.

Original comment by Frit...@gmail.com on 6 Nov 2010 at 1:01

• Changed state: Discuss

[GoogleCodeExporter]

Overpool wastes larva, so an earlier queen needs to be fast enough to offset 
that loss of larva, and it's not.

Overpool gets the queen earliest theoretically possible at 3:12. Overpool 
wastes slightly more than 1larva while gathering minerals for the spawning 
pool. A queen generates 5.3 larva per minute, so you would need to have the 
queen about 12 seconds earlier to _offset_ this fact.

9 OL gets queen earliest at 3:21, which is only 9 seconds slower. This does not 
offset the larva loss, which means that there is no larva advantage, and 
thereby no economic advantage.

Those are theoretical numbers, and in game the difference is even larger with 
overpool being at almost a while larva disadvantage.

Original comment by AudioL...@gmail.com on 8 Nov 2010 at 6:28

[GoogleCodeExporter]

Also worth mentioning that 9ol gets a second hatch faster than overpool, and 
does not waste larva. That makes the larva advantage undoubtedly in favour of 
9ol.

Original comment by AudioL...@gmail.com on 8 Nov 2010 at 7:29

[GoogleCodeExporter]

Just to add a less technical point: the mechanics behind th extractor trick are 
valid internally, so whether it is in fact worth doing will be determined as 
part of the fitness.

Arguing that the extractor trick is not economically sound is totally 
irrelevant - if it helps for a given build it will be used, if it doesn't, it 
won't. That is the beauty behind an optimizer like this, it doesn't make 
assumptions before testing.

It is worth noting that a large percentage of rush builds save around 15 
seconds through building a 9 or 10 extractor (some cancel for the trick, some 
leave but do not put 3 in gas early).

Original comment by kel.bolden@gmail.com on 9 Nov 2010 at 2:10

[GoogleCodeExporter]

You can disable extractor trick in the current version, so this issue should be 
moot. If you have economic concerns, then disable it. A future version will 
allow for economic weighting, and that should give you even further ability to 
find economic builds.

Original comment by qwerty10...@gmail.com on 9 Nov 2010 at 2:18

• Changed state: WontFix