The Patrol Split I have run into many players who think the optimal way to split is to select groups of marines and move them at angles using a patrol command. As the banelings approach the marines, some will stop and start shooting, the rest will keep moving. This allows you to automatically peel of a couple of marines at a time. Also at the end of the patrol the units tend to spread out a little naturally because the marines that are in front want to turn around and the rest want to continue forward. This jumble makes the marines split a little.
Why the Patrol Split is Suboptimal I think that this is an effective method, but I would argue that it is not optimal at high levels of play. I think it is sub optimal because in the patrol is disrupted by any targets that do not move at the speed of banelings, or enter from different angles. In real games there are many variables that could stop all your marines, such as suddenly scanning a nearby creep tumor or a speedling running in. Also if the marines complete the patrol pattern they tend to regroup.
Manual Splitting Manual splitting is simply a series of move commands given to break up a group of marines. We can actually put a direct price on splitting in actions. If we already have the marines selected the actions look like this: 1. Run away from banes 2. Stim 3. Select subgroup of marines 4. Move subgroup to a slightly different location 5. Repeat 3-4 until all marines are split or dead
If you have a group of 8 marines you would spend 2 actions to stim and run then select a subgroup of say 7 marines, move them somewhere else, then select 6 marines, move them to a third location. If we follow this through you can see that we have to make a subgroup for one less than the total number of marines. Each split takes two actions and we have two actions at the start. So we can arrive at this equation:
2+2(marineCount-1) = Total actions in split or 2*marineCount = Total actions in split
Why You Should Care Lets work through a quick example. Pro X can split with 240 APM (lower than max apm because of precision required). If Pro X has 10 seconds to split marines because he sees banelings coming and starts running, then he has 40 actions. 2*marineCount = 40 /2 /2 marineCount = 20 That is to say Pro X can perfectly split 20 marines or at least split his marine army into 20 pieces. This doesn't seem like an unreasonable situation to me, and a 20 piece marine army takes at least 40 banelings to kill (if non are shot down by marines at all). To me this seems far more effective than a patrol split.
Ways to get more out of your actions First suggestion would be to split individual marines after a few large splits. Splitting one marine off is easier than to take a small group because boxing requires a mouse movement and a click, rather than just a click. The second idea I have is to move marines out from the middle of a group: M M M to M ------>M M
Personal Example I am sure some of you are skeptical that this could possibly be a better way to split so here is a replay of me practicing manual marine splitting: http://drop.sc/118838 I get to level 37 in Darglien's "The Split" with stim and speed enabled, no creep. Using the patrol method I was only able to achieve level 30. The replay is long as I am not the greatest player in the world (mid master NA). I was also testing out my ideas as I went in that game, so it goes on for a while.
Outro Please discuss this and tell me your opinions, also please try it for yourself. I am curious to know what top level players think of this. I also think that this is so much more action intensive that in some situations, such as when you are multipronging, using the patrol split could be useful for its APM efficiency. However, I think that a manual split is more appropriate in a major battle.
tl;dr: High level players should manual split because it has more upward potential than patrol split and because it cannot be thrown off.
Edit: Made the first line more representative of my experience and less generalizing.
Patrol Split has never been the conventional wisdom in real games because unless you face 100% baneling your marines will stop to shoot a ling or muta and mess up the micro. How you micro against banelings and infestors really depends on the situation: sometimes you will see the best players micro in various ways depending on a large number of factors. It requires good judgment that comes from a deep understanding of the matchup.
Instead of theorycrafting about it just watch micro moments from players like MVP MKP and Taeja over and over. It's something you have to internalize subconsciously then repeatedly practice and experiment with, not think about.
I just downloaded it myself and ran the newly downloaded file no problem. Can someone else try and report if it worked or not? I ran the map out of the sc2 map editor directly so that I wouldn't be on any server in hopes that it would decrease latency (not sure if that works or not, but it feels like it is more responsive). That could be the issue, but it saved the replay and worked for me still. I don't really know any other way to test it.
On February 24 2012 18:13 Ver wrote: Patrol Split has never been the conventional wisdom in real games because unless you face 100% baneling your marines will stop to shoot a ling or muta and mess up the micro. How you micro against banelings and infestors really depends on the situation: sometimes you will see the best players micro in various ways depending on a large number of factors. It requires good judgment that comes from a deep understanding of the matchup.
Instead of theorycrafting about it just watch micro moments from players like MVP MKP and Taeja over and over. It's something you have to internalize subconsciously then repeatedly practice and experiment with, not think about.
Excellent point about the conventional wisdom line. I will edit that to better reflect what I meant to get at, which is that the method been seen by many lower league players as a viable option.
I think there is a time and place for theory crafting and in this case I think it has given me a better idea about what kinds of things I should do when I split. I think lower level players would benefit from thinking about the theory of it because I see many people around my skill level throw their marines around inefficiently without any regard to what will actually effectively split them, e.g. they declump some units, but then make orders to locations they just sent marines and reclump them.
Also practice is probably better than crafting an equation for action usage in splitting, but I had some time on my hands in which I did not have access to a computer and I felt that sharing my calculations and ideas with the community could help a few people out who.
On February 24 2012 18:13 Ver wrote: Patrol Split has never been the conventional wisdom in real games because unless you face 100% baneling your marines will stop to shoot a ling or muta and mess up the micro. How you micro against banelings and infestors really depends on the situation: sometimes you will see the best players micro in various ways depending on a large number of factors. It requires good judgment that comes from a deep understanding of the matchup.
Instead of theorycrafting about it just watch micro moments from players like MVP MKP and Taeja over and over. It's something you have to internalize subconsciously then repeatedly practice and experiment with, not think about.
a pre-split which i've never found a situation to use is if you load up like 4-5 medivacs then magic box the medivacs in a line perpendicular to your concave, then do a moving-drop with them to create 4-5 'layers' of spread marines infront of your concave.
perfect marine spread... each baneling will only hit 1-2 marine maximum.
of course in most situations where you don't have prep time, i find just boxing small groups of marines and spreading like crazy is the best technique... and like all micro in SC2 i believe this is WAAAAAAAAAAY easier if you're comfortable using high mouse sensitivity. box/moving with high sense is always going to be easier and faster than box/moving with a lower sensitivity.
On February 24 2012 18:13 Ver wrote: Patrol Split has never been the conventional wisdom in real games because unless you face 100% baneling your marines will stop to shoot a ling or muta and mess up the micro. How you micro against banelings and infestors really depends on the situation: sometimes you will see the best players micro in various ways depending on a large number of factors. It requires good judgment that comes from a deep understanding of the matchup.
Instead of theorycrafting about it just watch micro moments from players like MVP MKP and Taeja over and over. It's something you have to internalize subconsciously then repeatedly practice and experiment with, not think about.
MKP uses patrol splitting
Yea there are VODs of MKP playing the baneling micro challenge and he uses a patrol split technique with manual splitting on top of it. I don't know if he does this in real games though.
When MKP was streaming, he actually did the marine split challenge map a few times. He did both patrol and without patrol. One noteworthy thing was that when he used the patrol function, he did it to presplit marines rather than splitting while running from banes. He would patrol the whole clump of marines in 2 close spots and then hit stop a second later, resulting in all the marines being spread out and then spread them out further manually by boxing.
It seemed that he concluded the patrol function was not any good, as when he got to the higher levels, he stopped doing it all together and just split the marines manually with box method.
On February 24 2012 22:33 TheGreenMachine wrote: Heres a vod of Happy during a real game doing some epic marine+tank control. Very relevant to the conversation.
Looks like he uses a combination of moving individual groups of marines back, then A+moving them forward. Maybe some target fire with tanks too.
On February 24 2012 18:13 Ver wrote: Patrol Split has never been the conventional wisdom in real games because unless you face 100% baneling your marines will stop to shoot a ling or muta and mess up the micro. How you micro against banelings and infestors really depends on the situation: sometimes you will see the best players micro in various ways depending on a large number of factors. It requires good judgment that comes from a deep understanding of the matchup.
Instead of theorycrafting about it just watch micro moments from players like MVP MKP and Taeja over and over. It's something you have to internalize subconsciously then repeatedly practice and experiment with, not think about.
MKP uses patrol splitting
Yea there are VODs of MKP playing the baneling micro challenge and he uses a patrol split technique with manual splitting on top of it. I don't know if he does this in real games though.
No lol he was just testing. On his stream he definitely splits manually.
Do a study where you create different combinations of lings/blings/mutas coming in different arrangements. Give people a couple practice trials to acquaint them with the different variations, then throw them 20-30 trials of randomly selected challenges. Make this a map, ask 20-30 Master's League Terrans to try it, analyze their use of splitting methods and their cost effectiveness. Make graphs.
On February 24 2012 18:13 Ver wrote: Patrol Split has never been the conventional wisdom in real games because unless you face 100% baneling your marines will stop to shoot a ling or muta and mess up the micro. How you micro against banelings and infestors really depends on the situation: sometimes you will see the best players micro in various ways depending on a large number of factors. It requires good judgment that comes from a deep understanding of the matchup.
Instead of theorycrafting about it just watch micro moments from players like MVP MKP and Taeja over and over. It's something you have to internalize subconsciously then repeatedly practice and experiment with, not think about.
MKP uses patrol splitting
no he doesn't. he stated it in an interview somewhere, manual splitting is far better
patrol splitting isn't even cost efficient... a baneling still hits about 3-5 marines so it's better than a clumped ball of marines but it's still not good enough.
On February 24 2012 22:33 TheGreenMachine wrote: Heres a vod of Happy during a real game doing some epic marine+tank control. Very relevant to the conversation.
Looks like he uses a combination of moving individual groups of marines back, then A+moving them forward. Maybe some target fire with tanks too.
In real games, you're best off doing just this. 1- stim marines, run directly away 2- select tanks, shift queue all the banelings 3(finally) split your marines up.
Your marines should be relatively pre-split, ad splitting should be prioritized only after focus firing with tanks. And use the manual method. Patrol split is garbage
On February 24 2012 22:33 TheGreenMachine wrote: Heres a vod of Happy during a real game doing some epic marine+tank control. Very relevant to the conversation.
Looks like he uses a combination of moving individual groups of marines back, then A+moving them forward. Maybe some target fire with tanks too.
i used to be a big fan of the patrol splitting ( still use it but try not to). Especially doing a 2 racks pressure with about 8 marines sometimes zerg will go blings and if you do the individual splitting as opposed to the the patrol splitting i find that I always have a few marines left which can clean up the remaining lings that he sends in.
Why You Should Care Lets work through a quick example. Pro X can split with 240 APM (lower than max apm because of precision required). If Pro X has 10 seconds to split marines because he sees banelings coming and starts running, then he has 40 actions. 2*marineCount = 40 /2 /2 marineCount = 20 That is to say Pro X can perfectly split 20 marines or at least split his marine army into 20 pieces. This doesn't seem like an unreasonable situation to me, and a 20 piece marine army takes at least 40 banelings to kill (if non are shot down by marines at all). To me this seems far more effective than a patrol split.
Excuse me? Is that a reasonable assumption? I'd like to see a game where a pro has 10 seconds of time to just sit and split his marines and just wait for the banelings.
You're essentially using theory crafting to state that perfect marine positioning beats out the computer ai of patrol. I don't see how this is a useful argument at all.
Excuse me? Is that a reasonable assumption? I'd like to see a game where a pro has 10 seconds of time to just sit and split his marines and just wait for the banelings.
I think their was a game where puma was playing against a zerg, he scans his base sees banelings coming. And then splits every single one of his marines behind his wall at his natural or main. Banelings come in, destroy the wall only to find every marine already perfectly split.
If i remember correctly puma was behind and won the game. so it does happen
At the very top levels of play this is only worth it when already multi tasking in multiple places. If you have the APM of Demuslim or MKP you patently SHOULDN'T use this method in main battles. Why? Because it gives you less base control. Players of their caliber don't use this, not because it isn't useful, but because their normal mouse speed is faster and more flexible than this method.
In general, players SHOULD use this method when they don't have the APM to spare, but I don't think it should become a crutch or the vogue for Pros.
MKP does not usually patrol split, but sometimes does. Some pros do it. If you have to react quickly, it's good to patrol split real quick, and then fix them. If you are prepared, even by a split second, then you can split 'better', but if it's an OH SHIT moment, patrol can be useful, because it's patrol quick, then fix, whereas otherwise it's intensive at first.
patrol is very situational. Pure baneling vs marine will work, but with zerglings or mutas marines will stop fire at the zerglings and mutas so it wont work. and most players use manual split because IF patrol did help you have to do manual split anyways what is the point of wasting apm on 2 clicks and trying to get the click at the optimum spot.
EPT
