I've been silently reading TL for a long time, I figured it is time to contribue! I've been trying to do this as proper and scientific as possible, although more tests should probably be conducted to draw definite conclusions.
/Tenn (Sweden)
Edit
I've updated some conclusions and explanations for test 5 to better explain how I did the actual extractor trick.
--
Introduction
So, each game you play you have a different set of options to choose from, especially as zerg with the (so called) extractor trick element. Granted, this has little effect on the overall game (or has it?), but it is still 1-2 minutes when you have nothing else to do, so you might as well be as efficient as possible!
I've always been curious if the extractor trick is really worth it or not, seeing some top players using it while others doesn't. Moreover, I've been curious if microing the workers to make sure they go to an empty mineral patch is really worth it or not. While I was at it, I decided to see how 9 OL builds compares to 10 OL builds.
For those of you who don't know, the extractor trick is done by building an extractor when you're supply blocked, building a drone and then cancelling the extractor, giving you an extra supply until an OL pops out (usually used to get 11/10 supply).
Method
I used one of the build testers to be able to get a reliable timer for the different builds. After some thinking I settled on going for timing when the 15th drone spawns (stopping building drones when I've started the 15th), as well as the current minerals when that happens.
With the exception of Test 1 I will not be using any mineral micro since it adds another parameter to the test, and the the disadvantage of not microing should be roughly the same anyway.
Test 1: New drone first or moving all drones to mineral line first
Conclusion: If you build a new drone first you will get between 0 and 20 more minerals after the 15th drone pops out.
+ Show Spoiler +
Method: I've been using no micro but putting the rally point at the closest mineral patch.
Build: 9 OL
Create drone -> move workers
Attempt Time Minerals
1 2:06 270
2 2:06 265
3 2:07 265
Move workers -> create drone
Attempt Time Minerals
1 2:05 255
2 2:06 255
3 2:07 270
Granted, this is only viable when I am playing. Other players might be much more efficient when moving workers first, but I decided to post it anyway.
Test 2: Efficiency of mineral micro
Conclusion: You barely gain anything on microing drones (workers?) to empty mineral patches.
+ Show Spoiler +
Method: When microing I did the following:
Splitting the drones by first moving all drones to one mineral patch, and then, while they're on their way, moving 3 drones to a mineral patch at least 3 mineral patches away from the first one (so that all drones will quickly find an empty patch). This is how I've seen most top players do it, correct me if I'm wrong. Moreover, I've been putting rally point on empty patches as much as possible.
Build: Create drone - move workers - 9 OL
No micro
Attempt Time Minerals
1 2:06 270
2 2:06 265
3 2:07 265
Using micro
Attempt Time Minerals
1 2:06 270
2 2:06 270
3 2:06 260
If you compare the stats, you see that you indeed get a little advantage using micro, but it is minimal. But you have to bear in mind the risk when microing of misclicking etc.
More tests (by more players) should probably be made to get more accurate figures on this.
Test 3: 9 OL build or 10 OL build
Conclusion: 9 OL build is more economic.
+ Show Spoiler +
Micro used: Moving all workers to closest patch and rallying to the same patch.
Create drone - move workers - 9 OL
Attempt Time Minerals
1 2:06 270
2 2:06 265
3 2:07 265
Create drone - move workers - 10 OL
Attempt Time Minerals Minerals at 2:06
1 2:10 300 265
2 2:12 320 260
3 2:11 310 265
The 15th drone pops out roughly 5 seconds later than if doing the 9 OL build, but the mineral count up until 2:06 is roughly the same. This makes me believe that 9 OL build is more economic, but the 10 OL build might have other advantages as being able to afford spawning pool earlier. Although I never tested it, I think you get your first 200 minerals faster if doing 10 OL build.
Test 4: Extractor trick before overlord (11 OL)
Conclusion: Doing extractor trick gives you 10-15 minerals LESS than if skipping it.
Edit: Note: This isn't how you're supposed to do ETs. Test 5 is more appropriate
+ Show Spoiler +
Micro used: Moving all workers to closest patch and rallying to the same patch.
Create drone - move workers - 10 OL
Attempt Time Minerals Minerals at 2:06
1 2:10 300 265
2 2:12 320 260
3 2:11 310 265
Create drone - move workers - 10 OL - 11 Extractor (extractor trick)
Attempt Time Minerals
1 2:06 250 (254)
2 2:07 250 (254)
3 2:06 255 (259)
Note that you cannot use 4 minerals after you've done the extractor trick since everything you can buy is rounded by 10 (uh, best way i could explain it). The only way to get some of those 4 minerals is to make another extractor trick (at 18 supply for example).
As you can see the minerals after the 15th drone pops out are much lower than when doing nothing at all. Interesting, seeing many top players doing this extractor trick!
Test 5: Extractor trick after OL (10 OL)
Conclusion: If doing extractor trick this way you have 0-30 minerals LESS than if skipping it.
Edit: Some additional tests have been made, and it seems that you can make it the ET better and gain 5-10 minerals on that. But to do that you need to have practiced doing this alot in my experience.
+ Show Spoiler +
Micro used: Moving all workers to closest patch and rallying to the same patch.
Edit: When doing the extractor trick I tried to limit the time the extractor being built, as well as the time the drone have to wait as long as possible. So basically I waited until I had 75 minerals before starting doing the extractor, and I moved the drone closest to the gas as soon as it had left its minerals at the hatch. I believe this is how you do it best, but again, I might have missed 5 or 10 minerals.
Create drone - move workers - 9 OL
Attempt Time Minerals
1 2:06 270
2 2:06 265
3 2:07 265
Create drone - move workers - 9 OL - 10 Extractor (extractor trick)
Attempt Time Minerals
1 2:06 240 (244)
2 2:06 235 (239)
3 2:07 265 (269)
Again:
Note that you cannot use 4 minerals after you've done the extractor trick since everything you can buy is rounded by 10 (uh, best way i could explain it). The only way to get some of those 4 minerals is to make another extractor trick (at 18 supply for example).
As you can see the minerals after the 15th drone pops out are much lower than when doing nothing at all. Interesting, seeing many top players doing this extractor trick!
Final conclusion
It seems like the AI is intelligent enough without our efforts to help it out Extractor trick doesn't seem viable in any situation which is quite interesting IMO. In addition, the more micro you do the more likely you are to screw something up. For example, it took me like 5 tries to get 10 OL - extractor trick to get 3 perfect shots at it ( the numbers posted here), since it is so much more complex than just hitting "1sd" over and over.
Things to bear in mind
3 tests for each build is probably not enough to get reliable results, but I believe you can see some patterns still, and the numbers doesn't seem to differ very much within each tests so I think it's ok.
Larva timing is difficult to measure. For future tests it might be a good idea to measure when the next larva will spawn. The larva count has been at 1 when the 15th drone pops for all the different builds.