Hi, I'm back with another analysis, now about worker pairing. This time it is a short one. The question about worker micro is how big an advantage you can get, to ultimately decide if it is worth at all. Here is what I found.
TL;DR: Worker pairing, in the best case, can provide an extra 50 minerals early in the game, but it is likely that it will be less. This is a completely free 50 minerals, though, without any drawback. In higher leagues, it may actually matter a little bit, and it should not be a problem managing.
I'm looking forward to your opinions, criticisms, etc.
Worker pairing is pretty important. The extra 50 minerals allows you to get your first building 6~10 seconds earlier which is huge for holding off early cheese.
When you see pros worker stack, they don't always force workers onto a close mineral patch because like you said there is a downtime. Instead they let the rallied worker mine from a far mineral patch for the first 1 or 2 waves and then force them onto a closer mineral patch for less downtime.
Nice! It may seem small, but if you pair your workers at the start that's the difference of dropping your barracks at 1:34 instead of 1:40, which results in a faster OC etc etc
Except when I try to pair them they fuck off after 2 or so successful trips. Also sometimes I misclick and send all my miners to one patch and ruin all my pairings. Lol.
Sometimes it's just such a pain to pair them. Even when it seems they are perfectly paired (same interval between gathering minerals from both workers), one will suddenly "catch up" to the other and wander away for no apparent reason...
On August 07 2014 22:35 loft wrote: Pretty awesome work Sholip! Answering those questions people never dared to test.
How do you make the workers stay paired?
I will pair them (by spam clicking the mineral till it catches) and then they sometimes break away after 2 runs.
Keeping them paired is one of the hardest parts about it because if you just put the worker from the base onto it and it starts mining asap it will end up leaving in a few runs.
HOWEVER, if you start mining the side of the mineral patch first (Or the corner, i can't remember exactly) then it lines it up perfectly from the quarter second extra it needs to travel back initially for the 2 to stay mining constantly.
EDIT: there are also closer and further mineral patches as in not all close mineral patches are the same distance. This causes some issues as well with the pairing so be on the look out for that :3
My stargate build in PvP actually requires it for the timings: with worker pairing I can cut no probes and get my cyber core started exactly on time, without it my cyber core is slightly late and I have to cut a probe for a short while.
On August 07 2014 22:35 loft wrote: How do you make the workers stay paired? I will pair them (by spam clicking the mineral till it catches) and then they sometimes break away after 2 runs.
Well, I rarely do (gold league ftw ), but when so, I do the same. I think there is some better method, but I can't remember what it is. I, too, noticed that, on particular maps, they sometimes do "break away". I don't know why that is, though.
Jakatak did a video on this on how to keep them paired, I'm on mobile so I can't link it but I'm sure if someone searched for 30 seconds they could find it
From what I gather it's just because the human clicking speed doesn't get it always within the perfect pairing range so the desync after not many returns.
edit:slow
edit: I enjoy your work OP, I would be interested to know how effective worker mineral harass is (the bit where your opponent click on your mineral patch lots with their worker stopping you from mining and if that's more effective than just sending the worker home asap considering they have to pull a worker or two to deal with it. My other thing is how many SCV's make effective repairing on buildings when burning and not, because keeping one not mining for ages or lots not mining for one trip could make a difference It must do when burning?
On August 07 2014 23:00 DinoMight wrote: Except when I try to pair them they fuck off after 2 or so successful trips. Also sometimes I misclick and send all my miners to one patch and ruin all my pairings. Lol.
awesome work indeed. i dont agree with your last sentence though: "why then do they occassionally pick up 5 minerals while scouting, but dont bother to get 50 minerals for free?"
afaik the pickup up minerals of scouting workers have the main purpose of allowing mineral glitching, which might be the difference between losing and saving the scout.
btw is there any KOREAN pro that doesnt stack his workers?
On August 08 2014 00:10 Black Gun wrote: awesome work indeed. i dont agree with your last sentence though: "why then do they occassionally pick up 5 minerals while scouting, but dont bother to get 50 minerals for free?"
afaik the pickup up minerals of scouting workers have the main purpose of allowing mineral glitching, which might be the difference between losing and saving the scout.
btw is there any KOREAN pro that doesnt stack his workers?
You will prob get kicked out of the house for not stacking your workers .
50 minerals so early is potentially exponential, no? The extra worker which can be produced allows for another worker to be produced earlier, etc. etc.
Another thing that should be pointed out is that the ai does not distribute the workers evenly. In my games, I frequently see one patch with 3 workers and another patch with only 1. According to the tables, a worker that is send to a patch with two workers already instead of a patch with only one worker generates a loss of 15-33 minerals per minute, which is actually surprising, I thought it would be much more.
On August 08 2014 00:10 Black Gun wrote: awesome work indeed. i dont agree with your last sentence though: "why then do they occassionally pick up 5 minerals while scouting, but dont bother to get 50 minerals for free?"
afaik the pickup up minerals of scouting workers have the main purpose of allowing mineral glitching, which might be the difference between losing and saving the scout.
btw is there any KOREAN pro that doesnt stack his workers?
Do you mean mineral walking by mineral glitching? If so, you are right. Also, now it occures to me that some players mine from the opponent's minerals so as to screw up their worker stacking.
I don't know any players in particular, I just remember that many a time I would watch games between high-level pro players and I would notice that one stacks their workers, and the other one does not. Also, I checked the first game that literally came into my mind, Jaedong vs TaeJa in IEM Shenzhen semifinals. It seems to me that TaeJa didn't properly stack his SCVs in game one, but did so consistently in the following games. However, workers seemed to wander away after some time, breaking the perfect pairing.
Watching the video really enlightened me and now I am inclined to agree, I used to stack "normally" meaning my workers often bounced off pretty fast. This in turn made it bounce to another mineral patch bouncing away another worker and all that together made it feel more a waste of mine time than something that could actually give you an advantage.
Really liked the post and especially now that I've learnt how to properly do it I might incorporate it more into my play.
On August 08 2014 00:36 tili wrote: 50 minerals so early is potentially exponential, no? The extra worker which can be produced allows for another worker to be produced earlier, etc. etc.
dunno, that's how it works in my head
not exactly exponential, because time is a limiting factor, but it can matter a lot. e.g. if your 15 hatch is 7 seconds earlier than the one of your opponent in zvz, it means earlier queens, earlier injects, more drones and so on. 7 seconds are HUGE at higher levels. (ofc there won't be such a big discrepancy, but still...)
Startale Life goes 10 OL into 3 Drones, which has been proven to be less efficient than 9 OL. I guess he doesn't care, because I think it's foolish to say that a top Korean doesn't know something like that! I could be wrong, though.
On August 08 2014 01:31 reikai wrote: Startale Life goes 10 OL into 3 Drones, which has been proven to be less efficient than 9 OL. I guess he doesn't care, because I think it's foolish to say that a top Korean doesn't know something like that! I could be wrong, though.
It's marginally safer against early aggression because you have the extra drone to fight with, and he knows he's good enough in longer games that he's willing to take the slight hit.
Pairing on the close patches makes a huge difference. The earlier the timing you're trying to hit the more important it is to pair them up, at least in my experience. Its a big deal when you're, say, proxying 2 rax to have the second one start at 1:55 instead of 2:04.
Ive found the best way to keep them on the patches is to start from the outermost patches and work inwards, and to keep adjusting your worker rally point. If you rally to a far patch in between two close patches, the newly spawned workers can sometimes bump and disrupt the paired workers...it only takes a little disruption to break your pairing. So keep your worker rally as far away from your paired patches as possible.
I dont start worker pairing usually until after the second trip, and I try to sync it up so that the worker I'm forcing onto a patch is one that is close by, and on the return, to avoid blowing the extra minerals I'll get by wasting more worker time than the pairing is worth.
As for the APM it takes to do this...I can hit the same early game production timings and benchmarks as a korean pro (first 3-4 minutes) averaging 20 APM or so. This is with all SCVs paired up on the close patches while my 9 supply scv (total supply, not 9th produced) is still building. Its realistically as fast as you can do it without blowing all your gains by inefficiently pairing, and I usually average waaay less than 20apm doing this.
Its always funny to me when watching replays to see my opponent blazing away with 200+ APM for the first 3 minutes doing NOTHING, while I manage to have a higher income and better build timings with 1/10 the actions. If youre one of those people who needs to "warm up" or who is so insecure that you need to spam at 200apm for the first 3 min so no one questions the size of your e-peen when they see your APM on the score screen...just relax for the first minute, and focus on pairing, then move on to your nonsense spamming. Your build can't help but be better as a result.
id just like to add my observations to this on a general game.
the minigame i like to play EVERY game on a 14/15pool opening is what time does it go down!
The earliest ive ever done a 15 pool(if supply clicks to 15) is 1.56 and i got that by NOT pairing the workers but sending them to their own mineral patches but of course letting the new drones go to their closest patch relative to their larva position.
Sometimes ive gone mass try hard and rallied all drones before the pool to the closest patches relative to the base and hit 1.57-2.05. it didnt make sense. closer patch mining is optimal right?
I found out, sometimes the pairing actually isnt efficient if you dont get the drone to latch on just as the other drone is leaving, if you get them to pair up just after the other drone arrives you lose quite a lot of time. The drone would have made the trip already if you didnt spam the click onto the patch! The closer mining is instantly negated.
This is to say, ive had much earlier pool put downs by putting each drone to their own mineral patch first, further or closer.
The game just to be clear automatically assigns to drones latch on to the patch and will jump if the position is occupied. although 16 drones on the patches is regarded as the most efficient its not the most mineral income. 3 drones on far patches and 2 drones on close is and they indefinitely lock on as you give the drone NO where to go and dont get the unlinking of the drone.
When you do this right you give the computer NO chance to jump patches as every single patch is locked. this is shown when you only have the one patch left and you have 20 drones on 1 patch . .why dont they bounce around . . because they cant!
for me and for what ive noticed over the 10000's of games ive played rallying the drones to their own patch, prioritising the closest one to the larva spawn and then the closest patch to the base THEN trying to get them to rally Just after the last drone leaves is the most efficient way to set up early game mineral collection and be consistent.
To add one more thing to this, if you spam box, double click and get it wrong, then there was never no point. i must see one top streamer a day do this at least once.
There is another little trick which is still possible and that is to shift click a point before the base and the patch to stop the deceleration of the drone to the patch, i thought they took this out with hots but to me its still there.
Nice research, you're work is really interesting and appreciation. Finding out small things about this game is setting we really need to do. Thanks for this.
On August 08 2014 05:02 Yrr wrote: The total difference in minerals is only true for terran that didn't transform to orbital right? And did I mention gas workers and building workers..
I can't see 50 minerals or even 7 seconds making a real difference in a game. The only difference may be in extreme cheese (7 pool, 11 11 rax, 2 gate proxy, cannon rush, etc) where games last 4 minutes long. If you're anal about timings it can have a placebo-like effect I guess?
It's not going to get you ahead in workers because you should already be making workers with 100% uptime or 95% uptime in the early game anyway. Zerg is limited by larva in the beginning so you're not going to get ahead there either.
Scouting is actually much more important, as is perfect chrono, injects, and mules. With that said, yeah if it doesn't hurt and you've got all other aspects of the game down pat then why not?
My main is Z and I almost always do worker pairing. Like you say, it is not very critical but that extra minerals really make a difference when you are going for, like, 14gas 14pool speedling expand. I think it is useful to hone your skills if you are diamond or low masters.
On August 07 2014 22:52 TokyoGirl wrote: When you see pros worker stack, they don't always force workers onto a close mineral patch because like you said there is a downtime. Instead they let the rallied worker mine from a far mineral patch for the first 1 or 2 waves and then force them onto a closer mineral patch for less downtime.
This.
There are little other macro tricks as Zerg, like building an extractor and then inmediately rally three drones to it (they reach as soon as the extractor finishes). Or making 1 or 2 Overlords right after you do an inject.
On August 07 2014 22:35 loft wrote: Pretty awesome work Sholip! Answering those questions people never dared to test.
How do you make the workers stay paired?
I will pair them (by spam clicking the mineral till it catches) and then they sometimes break away after 2 runs.
You just need to get them to go in at the same angle. If they aren't coming from the right direction, shift-click close to the mineral patch in-line with the already mining worker. If both probes are mining from the same spot, they'll never split up by themselves.
On August 08 2014 10:10 GiveMeCake wrote: I can't see 50 minerals or even 7 seconds making a real difference in a game. The only difference may be in extreme cheese (7 pool, 11 11 rax, 2 gate proxy, cannon rush, etc) where games last 4 minutes long. If you're anal about timings it can have a placebo-like effect I guess?
It's not going to get you ahead in workers because you should already be making workers with 100% uptime or 95% uptime in the early game anyway. Zerg is limited by larva in the beginning so you're not going to get ahead there either.
Scouting is actually much more important, as is perfect chrono, injects, and mules. With that said, yeah if it doesn't hurt and you've got all other aspects of the game down pat then why not?
Below Masters being extremely precise with your timings will be extremely effective. So effective that you won't be below masters for very long at all. Even at low masters, I would expect to win a game (or gain an insurmountable advantage) with any timing I use if I I hit it correctly and don't f*** up my micro. I have won matches against players with good macro and better micro than I purely from being ~10 seconds ahead.
Of course, the longer a game goes the less 10 seconds matter, but you can get that 10 second advantage REALLY early and although this one mostly only applies to early game and taking bases, you can repeatedly grab an advantage so that you are further ahead.
EDIT: Of course, if everybody is doing it (eg: the pro level), then you won't get any advantage - but if you don't do it you will lose, which is often considered sub-optimal in Starcraft.
Below Masters being extremely precise with your timings will be extremely effective. So effective that you won't be below masters for very long at all. Even at low masters, I would expect to win a game (or gain an insurmountable advantage) with any timing I use if I I hit it correctly and don't f*** up my micro. I have won matches against players with good macro and better micro than I purely from being ~10 seconds ahead.
Of course, the longer a game goes the less 10 seconds matter, but you can get that 10 second advantage REALLY early and although this one mostly only applies to early game and taking bases, you can repeatedly grab an advantage so that you are further ahead.
EDIT: Of course, if everybody is doing it (eg: the pro level), then you won't get any advantage - but if you don't do it you will lose, which is often considered sub-optimal in Starcraft.
Post some replays proving that 10 seconds made the difference between winning and losing...
On August 07 2014 22:55 Jer99 wrote: Nice! It may seem small, but if you pair your workers at the start that's the difference of dropping your barracks at 1:34 instead of 1:40, which results in a faster OC etc etc
Below Masters being extremely precise with your timings will be extremely effective. So effective that you won't be below masters for very long at all. Even at low masters, I would expect to win a game (or gain an insurmountable advantage) with any timing I use if I I hit it correctly and don't f*** up my micro. I have won matches against players with good macro and better micro than I purely from being ~10 seconds ahead.
Of course, the longer a game goes the less 10 seconds matter, but you can get that 10 second advantage REALLY early and although this one mostly only applies to early game and taking bases, you can repeatedly grab an advantage so that you are further ahead.
EDIT: Of course, if everybody is doing it (eg: the pro level), then you won't get any advantage - but if you don't do it you will lose, which is often considered sub-optimal in Starcraft.
Post some replays proving that 10 seconds made the difference between winning and losing...
do you really think being at the enemies base 10 seconds earlier makes no difference? assume hellion harass vs protoss, 10 sec earlier and he doesnt get that forcefield > dead workers > win game
do you really think being at the enemies base 10 seconds earlier makes no difference? assume hellion harass vs protoss, 10 sec earlier and he doesnt get that forcefield > dead workers > win game
I believe the difference is so minimal that it would not decide whether you won the game or not. What hellion timing are you thinking of that would strike before the sentry had enough energy to FF?
Again, I'm still waiting for those replays showing worker stacking and split second timing being the deciding factor.
id just like to add my observations to this on a general game.
the minigame i like to play EVERY game on a 14/15pool opening is what time does it go down!
The earliest ive ever done a 15 pool(if supply clicks to 15) is 1.56 and i got that by NOT pairing the workers but sending them to their own mineral patches but of course letting the new drones go to their closest patch relative to their larva position.
Sometimes ive gone mass try hard and rallied all drones before the pool to the closest patches relative to the base and hit 1.57-2.05. it didnt make sense. closer patch mining is optimal right?
I found out, sometimes the pairing actually isnt efficient if you dont get the drone to latch on just as the other drone is leaving, if you get them to pair up just after the other drone arrives you lose quite a lot of time. The drone would have made the trip already if you didnt spam the click onto the patch! The closer mining is instantly negated.
This is to say, ive had much earlier pool put downs by putting each drone to their own mineral patch first, further or closer.
The game just to be clear automatically assigns to drones latch on to the patch and will jump if the position is occupied. although 16 drones on the patches is regarded as the most efficient its not the most mineral income. 3 drones on far patches and 2 drones on close is and they indefinitely lock on as you give the drone NO where to go and dont get the unlinking of the drone.
When you do this right you give the computer NO chance to jump patches as every single patch is locked. this is shown when you only have the one patch left and you have 20 drones on 1 patch . .why dont they bounce around . . because they cant!
for me and for what ive noticed over the 10000's of games ive played rallying the drones to their own patch, prioritising the closest one to the larva spawn and then the closest patch to the base THEN trying to get them to rally Just after the last drone leaves is the most efficient way to set up early game mineral collection and be consistent
On August 08 2014 10:10 GiveMeCake wrote: I can't see 50 minerals or even 7 seconds making a real difference in a game. The only difference may be in extreme cheese (7 pool, 11 11 rax, 2 gate proxy, cannon rush, etc) where games last 4 minutes long. If you're anal about timings it can have a placebo-like effect I guess?
It's not going to get you ahead in workers because you should already be making workers with 100% uptime or 95% uptime in the early game anyway. Zerg is limited by larva in the beginning so you're not going to get ahead there either.
Scouting is actually much more important, as is perfect chrono, injects, and mules. With that said, yeah if it doesn't hurt and you've got all other aspects of the game down pat then why not?
Below Masters being extremely precise with your timings will be extremely effective. So effective that you won't be below masters for very long at all. Even at low masters, I would expect to win a game (or gain an insurmountable advantage) with any timing I use if I I hit it correctly and don't f*** up my micro. I have won matches against players with good macro and better micro than I purely from being ~10 seconds ahead.
Of course, the longer a game goes the less 10 seconds matter, but you can get that 10 second advantage REALLY early and although this one mostly only applies to early game and taking bases, you can repeatedly grab an advantage so that you are further ahead.
EDIT: Of course, if everybody is doing it (eg: the pro level), then you won't get any advantage - but if you don't do it you will lose, which is often considered sub-optimal in Starcraft.
10 seconds are only gamechanging if one or both players go for very early aggression. if both are opening up for a longer game, it wont make much of a difference.
Good read. This is something I had wondered about. Microing workers is still good practice for APM and micro in general even if you don't care about the few extra minerals though
do you really think being at the enemies base 10 seconds earlier makes no difference? assume hellion harass vs protoss, 10 sec earlier and he doesnt get that forcefield > dead workers > win game
I believe the difference is so minimal that it would not decide whether you won the game or not. What hellion timing are you thinking of that would strike before the sentry had enough energy to FF?
Again, I'm still waiting for those replays showing worker stacking and split second timing being the deciding factor.
Any "early" pool where zerg pop 4 lings to try to catch the SCV building the CC on the low ground, often those are at 1 or 2 zergling attack + SCV position randomness. So yeah that pool or cc 6 second earlier is a huge deal.
Thanks for the analysis, 50 minerals is a big deal early game since it snowballs and affects early defense as well. Could you make an analysis of players pulling ther mules before their last "wasted" trip, and how much that actually saves?
this is cool could you do an analysis on minerals lost based on a failed pairing please :D
my drones seem to spend more tiem messing about when i start trying to optomise so i woder how many faied pairings before you negate the 50 advantage :D
My favorite way of explaining the actual (not mental) significance of worker-pairing is the timing of a 12 Rax on Xel'naga Caverns. Benchmarks: really good worker-pairing 1:33, standard good worker-pairing 1:34, if you dont worker-pair but still make rax exactly when you have money and still pick which mineral patch the scv goes to (not to pair, but to make sure they dont bounce around) 1:37.
EPT![[image loading]](http://i.imgur.com/04U0e48.jpg)
![[image loading]](http://i.imgur.com/w5jULME.jpg)
