I’m quite sure a lot of you have already noticed, but if not, I will make sure you do now, that there is a huge inconsistency in the game regarding the role of different units. Namely that there are worker units and battle units, which have their own, distinctive role in the game, that is, mining and attacking other units or structures, respectively. This would be all nice and great, but there is a small issue here. The fact that while combat units can only be used to attack, workers can gather resources and also attack (see SCV pull, Drone drill, etc.). No way it’s correct like this! Why would an SCV be worth more than a Marine just because he was given mining equipment instead of a combat suit? With all the rage about equality nowadays, one can wonder how the hell the developers ended up with a game of such discrepancy between unit types.
Thank God, it has recently been discovered that it need not be the case. A way has been found, and now also used in professional competition, to allow any unit to be a useful part of the economy by generating mineral income. The past model of “SCVs mining, Marines attacking” is no longer valid; every Marine now can pursue his own dreams of being the hero of not only the battlefield but also the mineral patches, freeing them from the burden of fear of being killed by a Psionic Storm while contributing to the Terran victory…
Now that you think this whole post is a big joke, let me tell you, it’s not. It is actually about the “patrol-in-the-mineral-line-to-boost-your-income” glitch that has now been discovered for quite a while, but only recently seen use in professional play. You may have heard about it already, but if you read it here first, chances are, you don’t believe it because it’s so absurd. But it’s true. The way it works is exactly as written in the previous sentence: you order a unit to patrol in the mineral line, and this somehow makes the workers gather minerals a little bit faster. Like this:
This effect is quite marginal in comparison to normal income rates, except for one case. It is known that when MULEs are dropped on a close mineral patch, they can almost complete their 10th trip, but they expire before that. With this technique, this is no longer the case.
This is most likely the consequence of a glitch, and is used by at least Flash and TY, see the video below.
As I usually do, I wrote a short analysis of the phenomenon which can be read in the spoiler tags or here in pdf format.
To summarize, these are the facts I know about this glitch:
It is caused by a moving unit in the mineral line.
The unit can be patrolling, but a simple well-timed move command will have the same effect.
The unit has to be close to the mineral patch; if it is patrolling at the CC, it has no effect.
The patrolling unit interacts with the worker, which is shoved perpendicular to the line in which the worker moves between the mineral and the CC (usually in the direction the patrolling unit is facing, but not always!), and also gets a sudden speed increase towards the CC. This increase allows it to complete its trip back faster.
This allows MULEs to return their 10th 30 minerals to the CC on close patches, increasing their effectiveness by 11.1%.
It seems the condition is that the collision radius of the worker and the patrolling unit overlap when the worker starts to move back towards the CC, but it’s only a guess.
It seems that if a MULE gets such an impact at least twice, it will complete its 10th trip.
This behavior is 100% reproducible.
As stated in the pdf, on a map with double close mineral patches, like Overgrowth, the patrolling unit can boost the income by as much as 56 minerals per game minute (with two Orbital Commands), which is more than what any regular worker could produce on any regular patch. This leads to the absurd situation that not only can a Marine (but also a Thor or any ground unit) now mine faster than any SCV, but in some cases it is worthwhile to order your SCV to patrol in the mineral line, instead of actually mining with it. This is really impressive considering the SCV is equipped with proper mining tools but the Marine has to mine with its Gauss rifles...
So now Marines can fight better than SCVs, and also mine faster. Where is the equality again?
Disclaimer: I did not discover this behavior; I have already seen it in two videos, one is Tefel's one above. I just didn’t find any posts covering this topic, and since it seemed that even mOOnGLaDe and Valdes didn't know what this is all about, I thought it may be informative.
And yes, this post contains sarcasm and irony, but the facts stated are true nonetheless.
I think this becomes problematic when you factor unit costs into how they otherwise could be used. In many situations it is true, defensive, idle units can patrol and boost mineral income. However, I think when the collision size of the unit is greater than the workers there is a chance they get boosted farther in the direction away from the command center, plus the movement speed from the patrolling unit has a random effect as well since they more frequently cause random intercation with workers. Plus, the turn rate of units (such as a thor) is rather slow and might cause more havoc, but this requires mathemathical testing to be proven (Plus, a thor might not ever patrol a mineral line.) So the numbers can only be trusted, as far as I see, by units of similar movement stats: SCV, drone, probe, marine, zergling, zealot, etc. It becomes harder and less efficienct with bigger units as far as I have tested, done similarily. I think using zealots to patrol in mineral line can become a problem, also using marines when they might want to be grouped versus oracles etc.
How useful it actually becomes is up to each game and how they evolve. Personally I think it is only "actually" useful with the mule to make it return the 30 minerals. Utilizing a single unit for 2 patches might be useful too, but you can always make patrols of all close mineral patches as well.
Btw, the is not usually efficient to do with long mineral patches because lack of movement area for the patrolling unit, plus i am sure the "accleration" (op defines the vector speed quite well) is altered differently when there is a longer distance to the town center/CC.
On January 31 2015 01:40 TheoMikkelsen wrote: I think this becomes problematic when you factor unit costs into how they otherwise could be used. In many situations it is true, defensive, idle units can patrol and boost mineral income. However, I think when the collision size of the unit is greater than the workers there is a chance they get boosted farther in Y direction, away from the command center, plus the movement speed from the patrolling unit has a random effect as well since they more frequently cause random intercation with workers. Plus, the turn rate of units (such as a thor) is rather slow and might cause more havoc, but this requires mathemathical testing to be proven (Plus, a thor might not ever patrol a mineral line.) So the numbers can only be trusted, as far as I see, by units of similar movement stats: SCV, drone, probe, marine, zergling, zealot, etc. It becomes harder and less efficienct with bigger units as far as I have tested, done similarily. I think using zealots to patrol in mineral line can become a problem, also using marines when they might want to be grouped versus oracles etc.
How useful it actually becomes is up to each game and how they evolve.
Very good post though and very useful.
I specifically tried a Thor, and it had the same effect visually. I don't know how effective it actually is with bigger/slower units, maybe different, maybe not, because I did not gather actual numerical data. Of course you will never patrol Thors in the mineral line, but since Marines are cheap and cost little supply, you can patrol with one or two on specific close patches, and use the MULEs there (just as Flash did).
And yeah, it only makes real sense on close patches in conjunction with MULEs, the boost on SCVs is only like an added bonus .
On January 31 2015 01:40 TheoMikkelsen wrote: I think this becomes problematic when you factor unit costs into how they otherwise could be used. In many situations it is true, defensive, idle units can patrol and boost mineral income. However, I think when the collision size of the unit is greater than the workers there is a chance they get boosted farther in Y direction, away from the command center, plus the movement speed from the patrolling unit has a random effect as well since they more frequently cause random intercation with workers. Plus, the turn rate of units (such as a thor) is rather slow and might cause more havoc, but this requires mathemathical testing to be proven (Plus, a thor might not ever patrol a mineral line.) So the numbers can only be trusted, as far as I see, by units of similar movement stats: SCV, drone, probe, marine, zergling, zealot, etc. It becomes harder and less efficienct with bigger units as far as I have tested, done similarily. I think using zealots to patrol in mineral line can become a problem, also using marines when they might want to be grouped versus oracles etc.
How useful it actually becomes is up to each game and how they evolve.
Very good post though and very useful.
I specifically tried a Thor, and it had the same effect visually. I don't know how effective it actually is with bigger/slower units, maybe different, maybe not, because I did not gather actual numerical data. Of course you will never patrol Thors in the mineral line, but since Marines are cheap and cost little supply, you can patrol with one or two on specific close patches, and use the MULEs there (just as Flash did).
And yeah, it only makes real sense on close patches in conjunction with MULEs, the boost on SCVs is only like an added bonus .
Well the effect is probably the same with thor if you test it with only 1, centered mineral patch. If you simulate this with a full 8 mineral patch, realistic test, I would believe not only does unit unstack close mineral patches so your workers start mining the far, inefficienct mineral patches, but this only works with mules because mules are usually in lower numbers for terrans and thus not provoked to go to other mineral patches when ignored scvs mining. Normal workers does not do this. (EDIT: actually you can make patrols that does not unstack if the workers does not get too much boost.)
The thor should achieve the same effect with just 1 mineral patch since the edge(priferi) of it´s collision is the same as the marine, but I think it becomes troubling in a full-size mineral line. If you do the mathemathical observations it might still gain mineral income but it should be different since the speeds is what causes, I do not see what else it can be, the "acceleration boosts" when the "collision sizes" of units collide. You could test it, though.
But I tested myself some time back with two close mineral patches and, especially then, is there a great chance the workers start choosing other mineral patches. However, I do believe this happens more rarely/never if you already have 16 workers with 2 on each patch, so if you sit with 20 workers and 2 only on close patches, they should not start going to other patches with only 2 etc. Only if there is 16 workers with two in each it MAY happen depending on the accuracy of the patrol etc, but if there are mathemathical definitions to this I am sure they have a better explanation.
In this picture, the select stalker jsut pushed a probe towards the mineral patch below it interfering slightly with the probe there. A different patrol would likely cause it to hit the other stalkers and the patrol randomness starts to kick in. This is why I think bigger collision sizes in "actual" mineral lines can become a problem but, surely if you are extremely accurate with the patrolling it should always gain benefit. Getting mathemathical observerations from 4 units patrolling like this is harder and more exhausting though I am sure. The test map cant make the best mineral patches to test with though...
For overgrowth patches specifically, with 2 patches next to each other, 2 units to get 4 patch boosts more frequently than if you used 4 pathces with 4 units seems to be a plausibly efficient choice, especially considering the +52 mineral possiblity. By the time you have your first unit, these patches are usually stacked though, and by the time you have full saturation, you probably need units to defend or scout or attack somewhere. How this trick can be used besides mules is something I doubt.
But I believe since they interfere with probe/worker stacking, units patrolling that is, this actually becomes useless since you likely want these units elsewhere by the time you are fully saturated. For mules though, always good.
For patches with SCV and Mules, then I think it is always worth it. perhaps you can test whether it is worth to boost the mule even if it breaks a close mineral patch stack? Maybe if you have a patrol that doesn´t make random steps you can keep the stacked workers for a long time, if not infinite, but it is quite hard i would assume.
When I just went to test big units, it actually seems like as long as the collision of the unit patrolling constantly is withing the collision of the worker, there is always a gaurantee to the boost but seemingly "slower" than units with higher movement speeds such as marine, so it does not seem to be true that bigger unit collisions can push the units away as long as you patrol correctly. If the thor was fast though, might be a different story. But then again, none of these units likely patrols mineral lines.
On January 31 2015 01:30 Blargh wrote: I think an important question to ask is... Why is Flash still losing if he has been using this totally imba mining trick??
he's actually too busy microing workers to protect his combat shield from dark templars
On January 31 2015 02:00 TheoMikkelsen wrote: But I believe since they interfere with probe/worker stacking, units patrolling that is, this actually becomes useless since you likely want these units elsewhere by the time you are fully saturated. For mules though, always good.
Are you sure they also interfere with worker stacking? I mean, it is already a mystery for me why workers sometimes suddenly unstack; and if this glitch exists in the first place, then it is also possible they do interfere with stacking as well, but I'm not sure.
On January 31 2015 02:02 OtherWorld wrote: I believe Tefel did a video on that some time ago too. It's interesting.
On January 31 2015 02:00 TheoMikkelsen wrote: But I believe since they interfere with probe/worker stacking, units patrolling that is, this actually becomes useless since you likely want these units elsewhere by the time you are fully saturated. For mules though, always good.
Are you sure they also interfere with worker stacking? I mean, it is already a mystery for me why workers sometimes suddenly unstack; and if this glitch exists in the first place, then it is also possible they do interfere with stacking as well, but I'm not sure.
On January 31 2015 02:02 OtherWorld wrote: I believe Tefel did a video on that some time ago too. It's interesting.
Thanks, OP updated with the video.
They did once, but when I am running almost 2 full mineral line exhausts right now it has not happened more than once. I think the stacking unstacking actually more likely has to do with the stacking rather than the patrolling. I think it happens if the worker has not returned minerals from the same patch several times so like it "checks" for other patches, like the bug you could do when scouting so that your worker starts choosing other patches if it is already taken. Maybe this patch has fixed it?
Or, since it is only 1 of the 2 returns from 2 workers that is out of sync, maybe if you do an incorrect stacking (the two workers are not in the exact same position and movement path) they will unstack if they hit the 1 of the two desyncs. This is also why you get 12 minerals from a 3rd worker on a close mineral patch. I think this is the most likely explanation.
Never the less, if the units never unstack workers even without full saturation, the patrol can work as long as it is done correctly.
On January 31 2015 02:02 OtherWorld wrote: I believe Tefel did a video on that some time ago too. It's interesting.
There's videos and guides up way back in 2010... of course back then there was even more broken way to mine more. But this patrol trick haven't been fixed evar. I guess people will continue rediscovering it until it does a shitstorm big enough on reddit for Blizzard to fix it (like the out of map tank trick or the mineral patch rally scouting trick).
On February 01 2015 01:36 sc2chronic wrote: even though it was never stated, im assuming that this works for all races, right?
Yeah, it does work for all races, but it becomes really effective when used in conjunction with MULEs to complete their 10th mining cycle. Simply boosting your regular workers' mining speed with this trick gives only a very small advantage, but an advantage nonetheless.
On February 01 2015 05:16 nkr wrote: I'm glad there's stuff like this in the game. Hopefully they don't remove it.
It's nice to have some little things like this to gain advantage of (see worker stacking for example), but this one is so illogical and just makes zero sense I wouldn't be so sad to see it removed (not to mention that it is clearly Terran favored). On the other hand, it doesn't really grant a huge advantage, but still I'm sure it is not an intended mechanism.
If anyone is still interested, I realized what a "closer" reason of this glitch might be. When I examined this behavior, I was under the impression that an SCV works on a patch for about 3.25 gs, before returning.
I only recently realized that a worker actually only works for 2.786 gs, and then waits another 0.5 gs. During this waiting time the next worker can already begin mining, but the SCV does not start moving (this data is "officially" available on Liquipedia). Looking at the data with this knowledge, it is clear that the strange stuff happens during this 0.5 gs. It is possible that the mineral walking, which should make it not to interact with other units at all, is somehow not (fully) active during this period, since the SCV is not actually moving. I can't say more about this, but I think someone who knows how the workers are coded to mine should know why this is happening, with this in mind.
Also, I'm curious whether this will be fixed in LotV (my guess is not).
EPT![[image loading]](http://i.imgur.com/fO0tXBb.jpg)
![[image loading]](http://i.imgur.com/YTq8IOd.jpg)
![[image loading]](http://i.imgur.com/rNQCKwu.jpg)
![[image loading]](http://i.imgur.com/HlLRQBi.jpg)
![[image loading]](http://i.imgur.com/WDjwHss.jpg)
![[image loading]](http://i.imgur.com/VKFdD3t.jpg)
