A ground-breaking experiment for macro that I'm going to reveal shortly, and what was its results, this experiment answers 2 questions:
What is the maximum number of workers I can have from a specific build (ex: 12 Nexus/16 CC) at a specific time (ex: 9:00) ? What is the quickest road/time to a specific number of workers (ex:50) I can have from a specific build ?
Before I begin to judge pro players macro, I just want to admit that they're very good @multitasking, micro, strategy and decision making, building-placement, etc.., but are just good @macro. And no one of them is Superman, they just generally exceed other players in one or more of the game aspects mentioned above.
Are you saying that pro players aren't the best macro players already, then who is ?! I'm not saying that, I'm just saying that their weakest skill is still macro and idle workers, we can see idle workers from observed games in ASL/KSL, but we can't see idle workers status or workers cut status or macro status accumulatively during or even after the game, then we say: WOW, this player isn't human, he is !!!, blah blah blah.
Fortunately for us, even top players like Flash, don't know the max number of workers he can get from 2:02 CC, 7:07 CC @9:00, he might only know his best record, so if someone patient enough the early game and can get the max number of workers of his build he can out macro Flash and win.
I mean can anyone believe that Flash cuts workers for 4:07 in one game, the early game (the 1st 9 minutes), that a 70 EAPM player can make ~20 workers more than Flash @9:00 from the same build ?! ?! ?!
I can't blame the top players for not able to see or sense many things that happen in the game as BWAPI can, the human mind can't do that (even if they record the screen or watch a replay frame by frame), they can't compare 2 games' macro adequately, they don't know if they break previous records by means other than the supply count or the game timer, and SCR, unfortunately, wants to establish this more and hide vital game infos from the players.
For Bisu, he knows that from 1 base he can have 27 Probes @5:00, but couldn't know how many seconds he stopped making workers (if 1 base is idle for 12.5s, 2 bases idle for 6s or 3 bases idle for 4s that means -1 worker), they also don't know how many seconds they've been above 500 or 750 minerals, how much time/minerals lost due to idle workers, etc... =============================================== To assure the macro defect of the previous era by examples: =============================================== "Best" which is known by many as one of the best macro players, has stopped making workers for 3:36 (Bisu=2:28, same build) in the early 9 minutes of a game, that means as the below pic says that he could've ~17 more workers, if he doesn't miss Macro. I think if Best knows that number, he might try to cut one of his fingers rather than cutting that number of workers again.
There is an option in the CoachAI called autoTrainWorkers, anyone can try to imitate a specific build and leave the AI to build workers by itself, to know the max number of workers from that build (that's the core idea of the topic).
This topic aims at encouraging the players-at-all-levels by saying "Yes, you can do it.", and reveal to them in numbers what a 100% macro skill means.
Is it the time to rethink of "building workers non-stop" ?
Before I jump to conclusions, have you verified that they weren't saving money for something more important than Probes, or that they simply don't have money for the Probe? For example, it is necessary to stop Probe production for a short amount of time after placing 11 Assimilator in PvT. Another example is when you intentionally cut SCVs to hit a specific timing, such as in the Hiya 4 Fac. Do you remove that number from your calculations, or do you blindly count CC/Nexus idle time and nothing else?
If the latter, then this topic is a perfect example of a problem that seems to pop up periodically on these forums. People who don't understand certain fundamental ideas pertaining to a build/unit/the game in general invest a lot of time based on misguided or wrong assumptions, make bad conclusions, and then they usually double-down on these ideas and become obstinate when people tell them that their ideas are misguided. This whole process, which has repeated time and again on these forums, could be easily cut short if people just used extant resources like Liquipedia or the SQSA thread to find out if their ideas are viable or not. Of course, it could be a case of "you don't know what you don't know" or not knowing what questions they need to ask, but still it's frustrating to constantly see time that could have been spent making something truly important or interesting for the community go squandered on silly things.
If it is the former, then that is admittedly impressive and I'm curious how you did it.
EDIT: Removed some text because I misunderstood your concluding thoughts.
Anyway, what's the point of knowing how many Probes you CAN have in a vacuum from a certain build, if playing the game against an opponent makes it impossible or even unfavorable for you to have "perfect" macro? In other words, why does it matter how many workers you could have, if in order to have that number of workers you have to always put worker production and sending them to minerals at the top of the list of priorities, over macro, micro, harass, etc.? It doesn't.
I honestly can’t follow anything in this post and I dont have any idea what I’m looking at with all these graphs and stuff. What is a player supposed to learn from this?
It’s nice that you are making a program to benefit the game but trust me, these worker cuts by pro players and even good foreign players are not by mistake, these are planned decisions based on scouting. Many players above 1900 mmr on ladder would have no problem macroing workers nearly flawlessly on 3 bases but the worker cuts you’re seeing in high level replays are not because these pros are “only human” it’s because they’ve fine tuned their game so well that they know when it’s a good idea to sacrifice worker production for something more important at that given time
Can you provide us an example of a game where a top player loses 3 minutes or more on worker production in the first 9 minutes? We can tell you how much of that was intentional. I don't think that your tool is useless - I like the idea, but these stats need to be able to cut out certain portions of the game where worker production is halted.
On November 02 2018 11:18 Lumire wrote: Many builds deliberately cut workers for added units/tech, how does this distinguish a deliberate probe cut from simply missing production rounds?
If the responses seem overwhelmingly negative it's because most of us had seen this misconception that you "have" to continuously produce workers many times before. Also that clickbait title and the sheer presumptioness probably doesn't help either.
"Continuously build workers" is a great heuristic to learn the game and improve yourself, but once you can continuously make workers automatically, you should throw it away and it's not to be used to analyse pro replays anyways. From your picture, 9 mins into the game Best has 3 mining bases and 63 probes. That's actually already over the "optimal" number of mining probes.
Does he need an extra 10 probes when he already have enough probes for 3 bases 9 mins into the game? Probes aren't free to make. Those 10 cut probes? Pretty much guaranteed they were cut so to make units, nexus, production buildings and tech up to hit certain important timings or simply not die. Maybe he is massing gateways in preparation for a big push to deny the third, or just straight up tech to carriers and arbiters.
Props to you though if you actually made an AI that perfectly copies the build with the the same tech and buildings and units and still came out 10 probes ahead though, that would be amazing. But it doesn't seem to do that.
On November 02 2018 11:05 Dazed. wrote: complete waste of time^
If you understood what this topic is trying to prove and have any logical reason for your hate or why its useless, say it, don't just hate and run. Its not a thing you can achieve before you make your own experiment, unless you already know the max number of workers from 12 Nexus @specific time, or want to play with no goals !
On November 02 2018 11:18 Lumire wrote: Many builds deliberately cut workers for added units/tech, how does this distinguish a deliberate probe cut from simply missing production rounds?
It doesn't, it even calculates the initial/inevitable cut when you've 5 workers and still don't have 50 minerals yet to build the 6th, also when the 1st supply (Pylon, Depot) is still in progress and you can't build the 10th/11th worker, or even when you've a floating CC.
On November 02 2018 11:28 orvinreyes wrote: All humans have lapses and errors. Even the most skilled ones. Humans are not AI.
But any intermediate player can do it, it doesn't require a lot APM, just more focus/practice, and the player that knows that this is possible, is not equal to who doesn't even know that he could have that number early.
On November 02 2018 11:30 Jealous wrote: Before I jump to conclusions, have you verified that they weren't saving money for something more important than Probes, or that they simply don't have money for the Probe? For example, it is necessary to stop Probe production for a short amount of time after placing 11 Assimilator in PvT. Another example is when you intentionally cut SCVs to hit a specific timing, such as in the Hiya 4 Fac. Do you remove that number from your calculations, or do you blindly count CC/Nexus idle time and nothing else?
If the latter, then this topic is a perfect example of a problem that seems to pop up periodically on these forums. People who don't understand certain fundamental ideas pertaining to a build/unit/the game in general invest a lot of time based on misguided or wrong assumptions, make bad conclusions, and then they usually double-down on these ideas and become obstinate when people tell them that their ideas are misguided. This whole process, which has repeated time and again on these forums, could be easily cut short if people just used extant resources like Liquipedia or the SQSA thread to find out if their ideas are viable or not. Of course, it could be a case of "you don't know what you don't know" or not knowing what questions they need to ask, but still it's frustrating to constantly see time that could have been spent making something truly important or interesting for the community go squandered on silly things.
If it is the former, then that is admittedly impressive and I'm curious how you did it.
EDIT: Removed some text because I misunderstood your concluding thoughts.
Anyway, what's the point of knowing how many Probes you CAN have in a vacuum from a certain build, if playing the game against an opponent makes it impossible or even unfavorable for you to have "perfect" macro? In other words, why does it matter how many workers you could have, if in order to have that number of workers you have to always put worker production and sending them to minerals at the top of the list of priorities, over macro, micro, harass, etc.? It doesn't.
haha, that would be extremely complicated, WorkerCut only goes for economy 1st, it's not a big defect in macro if you cut in the favour of tech, army or specific timing, but not for more than 2:00 I guess, at least Pros shouldn't do that.
The point of knowing how many Workers a player CAN have from a certain build, is that a player can macro better than other player who doesn't estimate the seconds lost (and eventually the number of workers that he could've made) due to cut, if you know for sure that there is a paradise and precise counter for good deeds you made, wouldn't that motivate/encourage you to do better in this life, if your don't believe in hell and a precise counter for bad deeds, could anything prevent you from doing worse ?
"putting worker production and sending them to minerals at the top of the list of priorities, over macro", I guess worker production IS the big part of macro, the other is spending resources.
On November 02 2018 17:30 Nightshade3 wrote: Implying that brood war players aren't 'aware' of how many workers they can make in a vacuum with a build
lol
ofc they don't, if you've looked @Bisu replay, you can see that he wasn't doing much works other than attacking the Bunkers with Dragoons, which mean low micro, what if he was handling drops or heavy micro actions, he would be not just 10 workers late, but more (or just have more idle workers).
On November 02 2018 20:55 mierin wrote: What I took from this was that Nexii give 9 Psi.
Funny, you still remember this.
On November 02 2018 21:06 Yanokabo wrote: I honestly can’t follow anything in this post and I dont have any idea what I’m looking at with all these graphs and stuff. What is a player supposed to learn from this?
You can macro better or run 100 meters faster if you have a watch in your hand, and you can also break a world record.
On November 02 2018 21:23 castleeMg wrote: It’s nice that you are making a program to benefit the game but trust me, these worker cuts by pro players and even good foreign players are not by mistake, these are planned decisions based on scouting. Many players above 1900 mmr on ladder would have no problem macroing workers nearly flawlessly on 3 bases but the worker cuts you’re seeing in high level replays are not because these pros are “only human” it’s because they’ve fine tuned their game so well that they know when it’s a good idea to sacrifice worker production for something more important at that given time
The most helpful comment till now, although I think one of the hardest things in this game is concentration, and pros still lose this to some extent, you might know everything you want to do before the game, but during the game, its something else.
On November 02 2018 21:42 Rodya wrote: Can you provide us an example of a game where a top player loses 3 minutes or more on worker production in the first 9 minutes? We can tell you how much of that was intentional. I don't think that your tool is useless - I like the idea, but these stats need to be able to cut out certain portions of the game where worker production is halted.
The replay/pic I've mentioned about Flash, is on this small pack.
On November 02 2018 22:49 Dangermousecatdog wrote: If the responses seem overwhelmingly negative it's because most of us had seen this misconception that you "have" to continuously produce workers many times before. Also that clickbait title and the sheer presumptioness probably doesn't help either.
"Continuously build workers" is a great heuristic to learn the game and improve yourself, but once you can continuously make workers automatically, you should throw it away and it's not to be used to analyse pro replays anyways. From your picture, 9 mins into the game Best has 3 mining bases and 63 probes. That's actually already over the "optimal" number of mining probes.
Does he need an extra 10 probes when he already have enough probes for 3 bases 9 mins into the game? Probes aren't free to make. Those 10 cut probes? Pretty much guaranteed they were cut so to make units, nexus, production buildings and tech up to hit certain important timings or simply not die. Maybe he is massing gateways in preparation for a big push to deny the third, or just straight up tech to carriers and arbiters.
Props to you though if you actually made an AI that perfectly copies the build with the the same tech and buildings and units and still came out 10 probes ahead though, that would be amazing. But it doesn't seem to do that.
You mean Bisu replay, the 9mins isn't imperative, you can change it to whatever you like, if this is over the "optimal" number of mining probes, the player can take 4th base, or better, only build, for example, 2nd base and test to see what the optimal time to reach 50 workers from a specific 2nd base-timing (maybe it 7mins, not 9).
I think its more than doable that a human (not even AI) can imitate same Bisu build taking the economy 1st approach, and have a bigger army/tech, but he (might) want to delay some things few seconds especially early on when he doesn't have the money to make everything.
Edit: I think I'm exaggerating on the topic title, but sometimes you need to do this to make people listen, I hope I didn't deceive anyone.
OP, I know what you're getting at here but your research is unfinished. Here's what you need to do from here: * Use your tools to map out where certain builds stand on the aggression-eco axis (difficult to gauge when early harass occurs) * Have a bot follow said build orders with variations, map those as well * Compare with the real data, to see if the used builds really are optimal, maybe some of the tradeoffs are not worth it and there's a better build that does the same thing just around the corner. Look for tech timings, specific army compositions, and economy size. ( hey, maybe there's a reaver build that's just three seconds later with much better eco behind it, unlikely but who knows ) * Fish out the most interesting alternatives to the pro builds. Study the builds beforehand so that you know what their intent is. * Nicely ask some people with better game knowledge to go over said alternatives, as they will know what the caveats are. * Understand that builds are map bound, subject to entropy as game time progresses, and in general just a guideline. Knowing what to focus on is a more important skill than trimming your probe timing by a few miliseconds.
Your time would probably be better spent just trying to write a good AI.
What's your source for claiming that these pros don't know off the top of their heads how many workers they would have at a given point in a game? Especially, LOL, since you proceed to do a simple arithmetic calculation to arrive at that number?
A way more interesting and useful endeavour would be to find out what happens if one player gets ahead on workers which are subsequently lost. For example, say in a PvP mirror one player gets their expansion down way faster, transfers probes and starts producing more off both Nexus's, then a short while later loses 8 probes to a reaver drop. For the sake of simplicity say they were also 8 probes ahead at that time. How much ahead are they still after losing those probes? Sure, both players may be even on probes again, but then the player who temporarily had more probes should still be ahead for having mined out more resources. BUT how much ahead? And obviously in this scenario there are lots of variables and things to be tested. But you're not gonna do that, you're gonna dick around telling us a Nexus gives 9 supply and that pros cut workers.
As people have mentioned already, pros cut workers intentionally as part of their build order. The thesis of this thread seems to be "make as many workers as possible," which isn't the only factor to winning a game.
This thread reminds me of SC2 where it was theory crafted that building CC's was more efficient than building depots for supply lol
On November 03 2018 01:29 badpenny wrote: What's your source for claiming that these pros don't know off the top of their heads how many workers they would have at a given point in a game? Especially, LOL, since you proceed to do a simple arithmetic calculation to arrive at that number?
A way more interesting and useful endeavour would be to find out what happens if one player gets ahead on workers which are subsequently lost. For example, say in a PvP mirror one player gets their expansion down way faster, transfers probes and starts producing more off both Nexus's, then a short while later loses 8 probes to a reaver drop. For the sake of simplicity say they were also 8 probes ahead at that time. How much ahead are they still after losing those probes? Sure, both players may be even on probes again, but then the player who temporarily had more probes should still be ahead for having mined out more resources. BUT how much ahead? And obviously in this scenario there are lots of variables and things to be tested. But you're not gonna do that, you're gonna dick around telling us a Nexus gives 9 supply and that pros cut workers.
My source is the replays. simple arithmetic calculation? that's funny, actually, I didn't see any other tool that even shows how much idle workers you have, not even in SCR.
Your case of PvP doesn't interfere with the usefulness of this WorkerCut standard, you can just compare the 2 players macro.
And your 8 Probe theory is also off topic, the player should secure his workers, you are trying to prove that WorkerCut measure is not useful but you failed, because your only lead is hate, you can try again, don't lose hope.
On November 03 2018 01:51 deathgod6 wrote: As people have mentioned already, pros cut workers intentionally as part of their build order. The thesis of this thread seems to be "make as many workers as possible," which isn't the only factor to winning a game.
This thread reminds me of SC2 where it was theory crafted that building CC's was more efficient than building depots for supply lol
Some people would say pros do their best that no other human can physically do, or that they cut workers on purpose (even if they're cutting workers intentionally, that would still be counted against them), the truth is: they don't focus optimally or they just follow a non-optimal BO.
And its not "make as many workers as possible," but don't be -1 worker for no reason, got it ?
This whole thread is a little odd, it seems you have good intentions to create a program and help the community which people should appreciate. You also seem to be very knowledgeable in making such a program by yourself which I’m sure many of us can’t do, but I really get the impression that you’re lacking essential game knowledge and seriously underestimating pros and even high level foreigners ability to multitask and keep a healthy consistent economy as well as modifying their builds accordingly to adapt to certain strategies.
Please provide some good examples comparing your method and the "normal" method for different build order. Also I guess this is not zerg oriented?
I find this thing interesting, knowing the perfect drone count for each build order, meaning you can't make more unit in X amount of time having n number of drones, but it needs deep analyse before I take it into account. I have low knowledge in maths but it sounds like some optimization algorithms and this could get highly complex. Is it really worth it?
On November 03 2018 02:43 Moataz wrote: Some people would say pros do their best that no other human can physically do, or that they cut workers on purpose (even if they're cutting workers intentionally, that would still be counted against them), the truth is: they don't focus optimally or they just follow a non-optimal BO.
And its not "make as many workers as possible," but don't be -1 worker for no reason, got it ?
I think you're still not understanding what people are telling you because you're probably not that great at the game, thus can't put yourself in the shoes of a higher level player. I'm not saying that I can completely understand what they do and why they do it either, but there are some posters in this thread like eonzerg that are worth listening to. Let me try to explain in more simple terms.
Let's say we are in the middle of a PvZ game, Protoss just moved out with a timing attack and is trying to establish their 3rd. Let's see what they need to do in the next 1-2 minutes.
What are the Protoss's priorities at this time?
1. Move army in such a way that Templar aren't exposed or too far behind the rest of the units. Losing your army or even getting a bad engagement could be fatal. (Checking every few seconds, potentially having to focus on this for longer during the engagement) 2. Harassing/scouting with Corsair and not getting hit by Scourge. Losing Corsair makes you vulnerable to Mutalisk, prevents you from scouting well, diminishes your harass potential. (Checking every few seconds) 3. Save 400 for Nexus, place Pylon for Cannon, place Nexus, make Cannons when Pylon finishes, send Probes to new minerals. (This would potentially result in "lost mining time" for you, no? Only a few actions here, so let's say that this is a spread of 5+ separate seconds out of the next minute)* 4. Maintaining unit production after Nexus. (This is also done relatively quickly, so let's say it's a few more seconds in the next minute; this could also force Probe cuts because it is more important to re-supply your army so that if your attack doesn't go well, you have enough units to hold your new 3rd and defend your natural). 5. Adding Gateways to coincide with third saturation. (A few seconds) 6. Making Probes. 7. Sending Probes to minerals. 8. Possibly other stuff I can't come up with right now.
*Now that I think about it, wouldn't 8 Probes transferring from main to natural result in 8x the amount of time spent traveling being counted as lost mining time? And from Main + Nat to 3rd, wouldn't this result in even more? Or did you find some way to overcome this?
While the above order is definitely arguable, I don't think anyone would argue that making Probes and sending them to minerals is more important than any of the actions above it. Thus, if sending a Probe to a mineral happens at the same time as the engagement, doesn't it make more sense to focus on the engagement and send the Probe after you have completed the engagement? Your counter would still consider this as "lost mining time" and you chide the progamer for "not having focus" but his decision to prioritize the engagement over sending a Probe to a mineral was correct because if he had looked away at that specific moment to handle the Probe, he could have missed an opportunity to Storm, gotten his Templar sniped, etc.?
How about if his Corsairs are harassing Overlords in the natural of the Zerg, and he sees a big swarm of Scourge, all nicely spread out and cloned, coming for his Corsairs. Oops! My Probe just finished, I have to go back to send him to the minerals! *Looks back* Where did my Corsairs go?
In other words, at any point in time, if the need to send a Probe to a mineral overlaps with the need to do something else, almost always the pro player will correctly choose to ignore the Probe until a more pressing issue is handled.
This is all, of course, ignoring the obvious cases where cutting Probes is necessary to make a build/timing work, etc. This is why people in this thread are saying that you've wasted your time with this approach.
On November 02 2018 11:30 Jealous wrote: Before I jump to conclusions, have you verified that they weren't saving money for something more important than Probes, or that they simply don't have money for the Probe? For example, it is necessary to stop Probe production for a short amount of time after placing 11 Assimilator in PvT
While I agree that there are many situations where you cut workers in order to use minerals for something else, placing your assimilator at 11 supply in PvT is not one of them. You can easily place the assimilator at 11 without ever having zero probes in your Nexus after your 6th probe.
I agree with what Jealous said. Pros are the best because they know what to prioritize and when. You simply can't be perfectly macroing at all times while also managing your army, scouting, harassing etc. And the player who does those latter things well will crush the player who just perfectly macros in their base at the expense of everything else. If the game were as simple as "more units automatically wins" then perhaps perfect macro would make sense. But so much depends on how well you use those units that you've made.
As said, there are plenty of reasons why workers are not building all the time. But that does not mean that Flash doesn't know how many scvs it's possible to have.
On November 02 2018 11:30 Jealous wrote: Before I jump to conclusions, have you verified that they weren't saving money for something more important than Probes, or that they simply don't have money for the Probe? For example, it is necessary to stop Probe production for a short amount of time after placing 11 Assimilator in PvT
While I agree that there are many situations where you cut workers in order to use minerals for something else, placing your assimilator at 11 supply in PvT is not one of them. You can easily place the assimilator at 11 without ever having zero probes in your Nexus after your 6th probe.
Maybe I'm just bad at splitting or deciding which mineral to send new Probes to, but it seems that when the 11th Probe finishes I always have 30-40 minerals and have to wait a tick or two to start the next one.
On November 03 2018 02:43 Moataz wrote: My source is the replays. simple arithmetic calculation? that's funny, actually, I didn't see any other tool that even shows how much idle workers you have, not even in SCR.
From your original post:
"even top players like Flash, don't know the max number of workers he can get from 2:02 CC, 7:07 CC @9:00"
"if 1 base is idle for 12.5s, 2 bases idle for 6s or 3 bases idle for 4s that means -1 worker"
"Flash replay (@9:00 workersCut=4:07, #workers=53) Flash replay remake (@9:00 workersCut=0:22, #workers=73) 04:07-00:22= 225s/12.5= Flash was behind by 18 workers"
Looks like you're doing arithmetic to me. Anyone with basic math knowledge can use the knowledge that 1 worker takes 12.5 second to produce, calculate the total time that a player had 3 completed town halls by 9 minutes in a game, and thus the ideal number of workers they could produce in 9 minutes with that particular 2nd and third town hall timing. After all, isn't that exactly how you (using this program) come to the conclusion that "Flash is behind by 18 workers" despite the fact also that 73-53=20 ???
Also how do those replays indicate what Flash actually knows and not just what he did in that particular game?
Edit: I also see on your graphic that you write "you can divide 3:36 by 12.5s (worker train time) to get the ideal amount of workers". That would be a simple arithmetic calculation in my books buddy.
I can tell you right now if Bisu is not making workers when the only thing he has are Dragoons shooting a bunker, he probably has a very good reason to do so.
Thread would be a lot better if you posted the exact replays you used so people can look into it themselves. You're basically holding back the most important piece of evidence.
I just looked into the Flash vs Free replay, and here's the breakdown of the first 9 minutes.
1. Flash goes for CC first, as he does in every game of this pack 2. He then cuts to make his rax at 15 supply. This is deliberate as he needs to adjust the timing of his marines to compensate for the extra economy 3. Another small cut after getting the refinery down 4. Another cut to afford a bunker and a factory at the same time. Again, priorities. 5. Another cut when his e-bay finishes, to get turrets up 6. Another cut to afford an Academy 7. Yet another one to make a second Factory 8. Another when he makes an Armory
During this, he does lose a few seconds between cycles, due to: 1. Microing his scout 2. Microing his army when he scans the DT at the natural 3. Microing his army to take a third base 4. Scanning Free's base
I'd say that adds up to a total of under 20 seconds (as opposed to the 225s you arrived at). He also loses a second here and there waiting for 100 minerals, in order to start SCVs in both CCs at the same time, which is a common macro technique used to make things smoother.
Beware though, that even some of the cuts while microing are unavoidable, as micro is something that often requires you to react to what the opponent is doing, so it can't be streamlined the same way as macro can.
For example when he scans the DT, Free may try to run away in a number of directions, or he may try to attack his marines, all of which require different inputs by Flash to optimize the exchange. That's not to say he couldn't have made a worker there, but that it's not that simple.
It's an interesting initiative your have, but what you're going for has a level of complexity eons higher than you seem to think. If I had your skills I'd make a bot that allows people to tell the computer what to do. Something similar to the training stage in Street Fighter, where you can control the AI character, press Record, do a move or combo, then go back to your own character and practice against that move over and over.
Something like this would be very useful to create micro scenarios on the fly, instead of having to deal with map editing and triggers, not to mention the limitations of both of those things.
On November 03 2018 03:04 castleeMg wrote: This whole thread is a little odd, it seems you have good intentions to create a program and help the community which people should appreciate. You also seem to be very knowledgeable in making such a program by yourself which I’m sure many of us can’t do, but I really get the impression that you’re lacking essential game knowledge and seriously underestimating pros and even high level foreigners ability to multitask and keep a healthy consistent economy as well as modifying their builds accordingly to adapt to certain strategies.
I'm not underestimating anyone, I'm a newbie @mechanics & multitasking BTW, I'm just trying to say there is a level up for everyone, no one even Top pro-games has reached the end of the perfection road yet, but they're certainly the best.
On November 03 2018 03:09 iFU.pauline wrote: Please provide some good examples comparing your method and the "normal" method for different build order. Also I guess this is not zerg oriented?
I find this thing interesting, knowing the perfect drone count for each build order, meaning you can't make more unit in X amount of time having n number of drones, but it needs deep analyse before I take it into account. I have low knowledge in maths but it sounds like some optimization algorithms and this could get highly complex. Is it really worth it?
I don't understand "comparing your method and the "normal" method for different build order", unfortunately, there are fewer features for Zerg, there is no WorkerCut measure for them. as their way of producing is different, there is no queue for the production or certain structure that is as easy as the 2 other races.
On November 03 2018 04:14 Piste wrote: As said, there are plenty of reasons why workers are not building all the time. But that does not mean that Flash doesn't know how many scvs it's possible to have.
So how much SCVs Flash can get from 4:00 CC, 8:00 CC, (note that the 8:00 CC will be ready after the default 9mins)
Can we make Flash comment on this, or answer this simple question, or how much time can Flash take to answer this?
On November 03 2018 02:43 Moataz wrote: Some people would say pros do their best that no other human can physically do, or that they cut workers on purpose (even if they're cutting workers intentionally, that would still be counted against them), the truth is: they don't focus optimally or they just follow a non-optimal BO.
And its not "make as many workers as possible," but don't be -1 worker for no reason, got it ?
I think you're still not understanding what people are telling you because you're probably not that great at the game, thus can't put yourself in the shoes of a higher level player. I'm not saying that I can completely understand what they do and why they do it either, but there are some posters in this thread like eonzerg that are worth listening to. Let me try to explain in more simple terms.
Let's say we are in the middle of a PvZ game, Protoss just moved out with a timing attack and is trying to establish their 3rd. Let's see what they need to do in the next 1-2 minutes.
What are the Protoss's priorities at this time?
1. Move army in such a way that Templar aren't exposed or too far behind the rest of the units. Losing your army or even getting a bad engagement could be fatal. (Checking every few seconds, potentially having to focus on this for longer during the engagement) 2. Harassing/scouting with Corsair and not getting hit by Scourge. Losing Corsair makes you vulnerable to Mutalisk, prevents you from scouting well, diminishes your harass potential. (Checking every few seconds) 3. Save 400 for Nexus, place Pylon for Cannon, place Nexus, make Cannons when Pylon finishes, send Probes to new minerals. (This would potentially result in "lost mining time" for you, no? Only a few actions here, so let's say that this is a spread of 5+ separate seconds out of the next minute)* 4. Maintaining unit production after Nexus. (This is also done relatively quickly, so let's say it's a few more seconds in the next minute; this could also force Probe cuts because it is more important to re-supply your army so that if your attack doesn't go well, you have enough units to hold your new 3rd and defend your natural). 5. Adding Gateways to coincide with third saturation. (A few seconds) 6. Making Probes. 7. Sending Probes to minerals. 8. Possibly other stuff I can't come up with right now.
*Now that I think about it, wouldn't 8 Probes transferring from main to natural result in 8x the amount of time spent traveling being counted as lost mining time? And from Main + Nat to 3rd, wouldn't this result in even more? Or did you find some way to overcome this?
While the above order is definitely arguable, I don't think anyone would argue that making Probes and sending them to minerals is more important than any of the actions above it. Thus, if sending a Probe to a mineral happens at the same time as the engagement, doesn't it make more sense to focus on the engagement and send the Probe after you have completed the engagement? Your counter would still consider this as "lost mining time" and you chide the progamer for "not having focus" but his decision to prioritize the engagement over sending a Probe to a mineral was correct because if he had looked away at that specific moment to handle the Probe, he could have missed an opportunity to Storm, gotten his Templar sniped, etc.?
How about if his Corsairs are harassing Overlords in the natural of the Zerg, and he sees a big swarm of Scourge, all nicely spread out and cloned, coming for his Corsairs. Oops! My Probe just finished, I have to go back to send him to the minerals! *Looks back* Where did my Corsairs go?
In other words, at any point in time, if the need to send a Probe to a mineral overlaps with the need to do something else, almost always the pro player will correctly choose to ignore the Probe until a more pressing issue is handled.
This is all, of course, ignoring the obvious cases where cutting Probes is necessary to make a build/timing work, etc. This is why people in this thread are saying that you've wasted your time with this approach.
I'm not saying these standards should be meaningful the whole game, the default 9min of WorkerCut isn't imperative, you can change it to whatever you like.
I know that there is a lot of tasks need to be done, but 2 games (Flash, Bisu, not Best) from what I mentioned, there were no heavy micro happening, the most noticeable thing for Bisu replay was, attacking the Bunkers with Dragoons, and Bisu was late by 10 workers as I said.
I'm not going to tell you when its suitable for you to build workers, whether you want to Storm efficiently, or run away with your Corsair, it is up to u when to make workers, I'm just saying don't forget, cause every second count.
I'm not chiding anyone for no reason. please tell me when I was chiding anyone? if you mean when I said Blah, Blah, Blah in the OP, that was for the casters and others that say: "Pro players aren't humans, but Gods".
What I really think is waste of time, is when you try to explain tool to someone who is not intended to buy it, or even test it for free, and then he comes to complain with very few infos he knows, and you have to correct some misunderstanding for him, when you're also have written some good manual of what this tool does.
On November 03 2018 02:43 Moataz wrote: My source is the replays. simple arithmetic calculation? that's funny, actually, I didn't see any other tool that even shows how much idle workers you have, not even in SCR.
From your original post:
"even top players like Flash, don't know the max number of workers he can get from 2:02 CC, 7:07 CC @9:00"
"if 1 base is idle for 12.5s, 2 bases idle for 6s or 3 bases idle for 4s that means -1 worker"
"Flash replay (@9:00 workersCut=4:07, #workers=53) Flash replay remake (@9:00 workersCut=0:22, #workers=73) 04:07-00:22= 225s/12.5= Flash was behind by 18 workers"
Looks like you're doing arithmetic to me. Anyone with basic math knowledge can use the knowledge that 1 worker takes 12.5 second to produce, calculate the total time that a player had 3 completed town halls by 9 minutes in a game, and thus the ideal number of workers they could produce in 9 minutes with that particular 2nd and third town hall timing. After all, isn't that exactly how you (using this program) come to the conclusion that "Flash is behind by 18 workers" despite the fact also that 73-53=20 ???
Also how do those replays indicate what Flash actually knows and not just what he did in that particular game?
Edit: I also see on your graphic that you write "you can divide 3:36 by 12.5s (worker train time) to get the ideal amount of workers". That would be a simple arithmetic calculation in my books buddy.
What you're saying is "ink on papers" the execution is slightly harder (just like StarCraft1), if that easy as you've said why nobody has made it before? Why don't you try to help with ideas or code or specific algorithm, btw you can fork the CoachAI (its open source), or make a similar tool from scratch? but that is not as easy as you think.
FYI Flash lost 2 SCVs before 9mins in the mentioned replay, I hope this answers your question.
On November 03 2018 03:04 castleeMg wrote: This whole thread is a little odd, it seems you have good intentions to create a program and help the community which people should appreciate. You also seem to be very knowledgeable in making such a program by yourself which I’m sure many of us can’t do, but I really get the impression that you’re lacking essential game knowledge and seriously underestimating pros and even high level foreigners ability to multitask and keep a healthy consistent economy as well as modifying their builds accordingly to adapt to certain strategies.
On November 03 2018 02:43 Moataz wrote: Some people would say pros do their best that no other human can physically do, or that they cut workers on purpose (even if they're cutting workers intentionally, that would still be counted against them), the truth is: they don't focus optimally or they just follow a non-optimal BO.
And its not "make as many workers as possible," but don't be -1 worker for no reason, got it ?
I think you're still not understanding what people are telling you because you're probably not that great at the game, thus can't put yourself in the shoes of a higher level player. I'm not saying that I can completely understand what they do and why they do it either, but there are some posters in this thread like eonzerg that are worth listening to. Let me try to explain in more simple terms.
Let's say we are in the middle of a PvZ game, Protoss just moved out with a timing attack and is trying to establish their 3rd. Let's see what they need to do in the next 1-2 minutes.
What are the Protoss's priorities at this time?
1. Move army in such a way that Templar aren't exposed or too far behind the rest of the units. Losing your army or even getting a bad engagement could be fatal. (Checking every few seconds, potentially having to focus on this for longer during the engagement) 2. Harassing/scouting with Corsair and not getting hit by Scourge. Losing Corsair makes you vulnerable to Mutalisk, prevents you from scouting well, diminishes your harass potential. (Checking every few seconds) 3. Save 400 for Nexus, place Pylon for Cannon, place Nexus, make Cannons when Pylon finishes, send Probes to new minerals. (This would potentially result in "lost mining time" for you, no? Only a few actions here, so let's say that this is a spread of 5+ separate seconds out of the next minute)* 4. Maintaining unit production after Nexus. (This is also done relatively quickly, so let's say it's a few more seconds in the next minute; this could also force Probe cuts because it is more important to re-supply your army so that if your attack doesn't go well, you have enough units to hold your new 3rd and defend your natural). 5. Adding Gateways to coincide with third saturation. (A few seconds) 6. Making Probes. 7. Sending Probes to minerals. 8. Possibly other stuff I can't come up with right now.
*Now that I think about it, wouldn't 8 Probes transferring from main to natural result in 8x the amount of time spent traveling being counted as lost mining time? And from Main + Nat to 3rd, wouldn't this result in even more? Or did you find some way to overcome this?
While the above order is definitely arguable, I don't think anyone would argue that making Probes and sending them to minerals is more important than any of the actions above it. Thus, if sending a Probe to a mineral happens at the same time as the engagement, doesn't it make more sense to focus on the engagement and send the Probe after you have completed the engagement? Your counter would still consider this as "lost mining time" and you chide the progamer for "not having focus" but his decision to prioritize the engagement over sending a Probe to a mineral was correct because if he had looked away at that specific moment to handle the Probe, he could have missed an opportunity to Storm, gotten his Templar sniped, etc.?
How about if his Corsairs are harassing Overlords in the natural of the Zerg, and he sees a big swarm of Scourge, all nicely spread out and cloned, coming for his Corsairs. Oops! My Probe just finished, I have to go back to send him to the minerals! *Looks back* Where did my Corsairs go?
In other words, at any point in time, if the need to send a Probe to a mineral overlaps with the need to do something else, almost always the pro player will correctly choose to ignore the Probe until a more pressing issue is handled.
This is all, of course, ignoring the obvious cases where cutting Probes is necessary to make a build/timing work, etc. This is why people in this thread are saying that you've wasted your time with this approach.
and... this ^^ Basically summarizes why people are giving you negative feedback.
On November 03 2018 02:43 Moataz wrote: Some people would say pros do their best that no other human can physically do, or that they cut workers on purpose (even if they're cutting workers intentionally, that would still be counted against them), the truth is: they don't focus optimally or they just follow a non-optimal BO.
And its not "make as many workers as possible," but don't be -1 worker for no reason, got it ?
I think you're still not understanding what people are telling you because you're probably not that great at the game, thus can't put yourself in the shoes of a higher level player. I'm not saying that I can completely understand what they do and why they do it either, but there are some posters in this thread like eonzerg that are worth listening to. Let me try to explain in more simple terms.
Let's say we are in the middle of a PvZ game, Protoss just moved out with a timing attack and is trying to establish their 3rd. Let's see what they need to do in the next 1-2 minutes.
What are the Protoss's priorities at this time?
1. Move army in such a way that Templar aren't exposed or too far behind the rest of the units. Losing your army or even getting a bad engagement could be fatal. (Checking every few seconds, potentially having to focus on this for longer during the engagement) 2. Harassing/scouting with Corsair and not getting hit by Scourge. Losing Corsair makes you vulnerable to Mutalisk, prevents you from scouting well, diminishes your harass potential. (Checking every few seconds) 3. Save 400 for Nexus, place Pylon for Cannon, place Nexus, make Cannons when Pylon finishes, send Probes to new minerals. (This would potentially result in "lost mining time" for you, no? Only a few actions here, so let's say that this is a spread of 5+ separate seconds out of the next minute)* 4. Maintaining unit production after Nexus. (This is also done relatively quickly, so let's say it's a few more seconds in the next minute; this could also force Probe cuts because it is more important to re-supply your army so that if your attack doesn't go well, you have enough units to hold your new 3rd and defend your natural). 5. Adding Gateways to coincide with third saturation. (A few seconds) 6. Making Probes. 7. Sending Probes to minerals. 8. Possibly other stuff I can't come up with right now.
*Now that I think about it, wouldn't 8 Probes transferring from main to natural result in 8x the amount of time spent traveling being counted as lost mining time? And from Main + Nat to 3rd, wouldn't this result in even more? Or did you find some way to overcome this?
While the above order is definitely arguable, I don't think anyone would argue that making Probes and sending them to minerals is more important than any of the actions above it. Thus, if sending a Probe to a mineral happens at the same time as the engagement, doesn't it make more sense to focus on the engagement and send the Probe after you have completed the engagement? Your counter would still consider this as "lost mining time" and you chide the progamer for "not having focus" but his decision to prioritize the engagement over sending a Probe to a mineral was correct because if he had looked away at that specific moment to handle the Probe, he could have missed an opportunity to Storm, gotten his Templar sniped, etc.?
How about if his Corsairs are harassing Overlords in the natural of the Zerg, and he sees a big swarm of Scourge, all nicely spread out and cloned, coming for his Corsairs. Oops! My Probe just finished, I have to go back to send him to the minerals! *Looks back* Where did my Corsairs go?
In other words, at any point in time, if the need to send a Probe to a mineral overlaps with the need to do something else, almost always the pro player will correctly choose to ignore the Probe until a more pressing issue is handled.
This is all, of course, ignoring the obvious cases where cutting Probes is necessary to make a build/timing work, etc. This is why people in this thread are saying that you've wasted your time with this approach.
and... this ^^ Basically summarizes why people are giving you negative feedback.
On November 03 2018 03:04 castleeMg wrote: This whole thread is a little odd, it seems you have good intentions to create a program and help the community which people should appreciate. You also seem to be very knowledgeable in making such a program by yourself which I’m sure many of us can’t do, but I really get the impression that you’re lacking essential game knowledge and seriously underestimating pros and even high level foreigners ability to multitask and keep a healthy consistent economy as well as modifying their builds accordingly to adapt to certain strategies.
I'm not underestimating anyone, I'm a newbie @mechanics & multitasking BTW, I'm just trying to say there is a level up for everyone, no one even Top pro-games has reached the end of the perfection road yet, but they're certainly the best.
On November 03 2018 03:09 iFU.pauline wrote: Please provide some good examples comparing your method and the "normal" method for different build order. Also I guess this is not zerg oriented?
I find this thing interesting, knowing the perfect drone count for each build order, meaning you can't make more unit in X amount of time having n number of drones, but it needs deep analyse before I take it into account. I have low knowledge in maths but it sounds like some optimization algorithms and this could get highly complex. Is it really worth it?
I don't understand "comparing your method and the "normal" method for different build order", unfortunately, there are fewer features for Zerg, there is no WorkerCut measure for them. as their way of producing is different, there is no queue for the production or certain structure that is as easy as the 2 other races.
On November 03 2018 04:14 Piste wrote: As said, there are plenty of reasons why workers are not building all the time. But that does not mean that Flash doesn't know how many scvs it's possible to have.
So how much SCVs Flash can get from 4:00 CC, 8:00 CC, (note that the 8:00 CC will be ready after the default 9mins)
Can we make Flash comment on this, or answer this simple question, or how much time can Flash take to answer this?
On November 03 2018 02:43 Moataz wrote: Some people would say pros do their best that no other human can physically do, or that they cut workers on purpose (even if they're cutting workers intentionally, that would still be counted against them), the truth is: they don't focus optimally or they just follow a non-optimal BO.
And its not "make as many workers as possible," but don't be -1 worker for no reason, got it ?
I think you're still not understanding what people are telling you because you're probably not that great at the game, thus can't put yourself in the shoes of a higher level player. I'm not saying that I can completely understand what they do and why they do it either, but there are some posters in this thread like eonzerg that are worth listening to. Let me try to explain in more simple terms.
Let's say we are in the middle of a PvZ game, Protoss just moved out with a timing attack and is trying to establish their 3rd. Let's see what they need to do in the next 1-2 minutes.
What are the Protoss's priorities at this time?
1. Move army in such a way that Templar aren't exposed or too far behind the rest of the units. Losing your army or even getting a bad engagement could be fatal. (Checking every few seconds, potentially having to focus on this for longer during the engagement) 2. Harassing/scouting with Corsair and not getting hit by Scourge. Losing Corsair makes you vulnerable to Mutalisk, prevents you from scouting well, diminishes your harass potential. (Checking every few seconds) 3. Save 400 for Nexus, place Pylon for Cannon, place Nexus, make Cannons when Pylon finishes, send Probes to new minerals. (This would potentially result in "lost mining time" for you, no? Only a few actions here, so let's say that this is a spread of 5+ separate seconds out of the next minute)* 4. Maintaining unit production after Nexus. (This is also done relatively quickly, so let's say it's a few more seconds in the next minute; this could also force Probe cuts because it is more important to re-supply your army so that if your attack doesn't go well, you have enough units to hold your new 3rd and defend your natural). 5. Adding Gateways to coincide with third saturation. (A few seconds) 6. Making Probes. 7. Sending Probes to minerals. 8. Possibly other stuff I can't come up with right now.
*Now that I think about it, wouldn't 8 Probes transferring from main to natural result in 8x the amount of time spent traveling being counted as lost mining time? And from Main + Nat to 3rd, wouldn't this result in even more? Or did you find some way to overcome this?
While the above order is definitely arguable, I don't think anyone would argue that making Probes and sending them to minerals is more important than any of the actions above it. Thus, if sending a Probe to a mineral happens at the same time as the engagement, doesn't it make more sense to focus on the engagement and send the Probe after you have completed the engagement? Your counter would still consider this as "lost mining time" and you chide the progamer for "not having focus" but his decision to prioritize the engagement over sending a Probe to a mineral was correct because if he had looked away at that specific moment to handle the Probe, he could have missed an opportunity to Storm, gotten his Templar sniped, etc.?
How about if his Corsairs are harassing Overlords in the natural of the Zerg, and he sees a big swarm of Scourge, all nicely spread out and cloned, coming for his Corsairs. Oops! My Probe just finished, I have to go back to send him to the minerals! *Looks back* Where did my Corsairs go?
In other words, at any point in time, if the need to send a Probe to a mineral overlaps with the need to do something else, almost always the pro player will correctly choose to ignore the Probe until a more pressing issue is handled.
This is all, of course, ignoring the obvious cases where cutting Probes is necessary to make a build/timing work, etc. This is why people in this thread are saying that you've wasted your time with this approach.
I'm not saying these standards should be meaningful the whole game, the default 9min of WorkerCut isn't imperative, you can change it to whatever you like.
I know that there is a lot of tasks need to be done, but 2 games (Flash, Bisu, not Best) from what I mentioned, there were no heavy micro happening, the most noticeable thing for Bisu replay was, attacking the Bunkers with Dragoons, and Bisu was late by 10 workers as I said.
I'm not going to tell you when its suitable for you to build workers, whether you want to Storm efficiently, or run away with your Corsair, it is up to u when to make workers, I'm just saying don't forget, cause every second count.
I'm not chiding anyone for no reason. please tell me when I was chiding anyone? if you mean when I said Blah, Blah, Blah in the OP, that was for the casters and others that say: "Pro players aren't humans, but Gods".
What I really think is waste of time, is when you try to explain tool to someone who is not intended to buy it, or even test it for free, and then he comes to complain with very few infos he knows, and you have to correct some misunderstanding for him, when you're also have written some good manual of what this tool does.
On November 03 2018 02:43 Moataz wrote: My source is the replays. simple arithmetic calculation? that's funny, actually, I didn't see any other tool that even shows how much idle workers you have, not even in SCR.
From your original post:
"even top players like Flash, don't know the max number of workers he can get from 2:02 CC, 7:07 CC @9:00"
"if 1 base is idle for 12.5s, 2 bases idle for 6s or 3 bases idle for 4s that means -1 worker"
"Flash replay (@9:00 workersCut=4:07, #workers=53) Flash replay remake (@9:00 workersCut=0:22, #workers=73) 04:07-00:22= 225s/12.5= Flash was behind by 18 workers"
Looks like you're doing arithmetic to me. Anyone with basic math knowledge can use the knowledge that 1 worker takes 12.5 second to produce, calculate the total time that a player had 3 completed town halls by 9 minutes in a game, and thus the ideal number of workers they could produce in 9 minutes with that particular 2nd and third town hall timing. After all, isn't that exactly how you (using this program) come to the conclusion that "Flash is behind by 18 workers" despite the fact also that 73-53=20 ???
Also how do those replays indicate what Flash actually knows and not just what he did in that particular game?
Edit: I also see on your graphic that you write "you can divide 3:36 by 12.5s (worker train time) to get the ideal amount of workers". That would be a simple arithmetic calculation in my books buddy.
What you're saying is "ink on papers" the execution is slightly harder (just like StarCraft1), if that easy as you've said why nobody has made it before? Why don't you try to help with ideas or code or specific algorithm, btw you can fork the CoachAI (its open source), or make a similar tool from scratch? but that is not as easy as you think.
FYI Flash lost 2 SCVs before 9mins in the mentioned replay, I hope this answers your question.
BTW, thanks for being slightly nice this time.
Because the information is well known already by the majority of the sc population. They know macro > micro. Building a program that explains how many workers someone failed to make is honestly pointless. The body is limited and because of that people are forced to prioritize some things over others, for example producing marines or tanks or whatever the case may be then immediately jumping back to micro and then maybe responding to defend a drop and then back to micro ect ect ect... depending on the situation.
Must keep in mind that these are players with over 400 APM. That is absolutely nuts if you actually think about it. That's a lot of physical work on top of having to actively strategize against your opponent. This is why most sc programmer careers start to decline around mid 20's. So many players are left with carpal tunnel.
I think some of the things you might be pointing out are physically impossible to perfect when you're exchanging with a player of equal skill.
This "tool" does one thing: tells us progamers don't have literally perfect macro. We know... no one thinks they did, because that's impossible. What stunning discovery will your next thread contain?
On November 03 2018 05:13 Moataz wrote: What I really think is waste of time, is when you try to explain tool to someone who is not intended to buy it, or even test it for free, and then he comes to complain with very few infos he knows, and you have to correct some misunderstanding for him, when you're also have written some good manual of what this tool does.
You're right, my bad. We're all idiots here who don't understand the magnitude, the sheer impact, of what you have accomplished. Sorry for wasting your time.
Actually at the Italy event when foreigners asked FBH for advice FBH told them unironically to work on constant worker production and quickly sending them to mine and that it was something even he and other pros can improve on ¯\_(ツ)_/¯
I think you can actually make this tool much better by improving the algorithm in the following ways:
1. Always check if the player has enough minerals to make a worker(not just in the beginning) 2. If the minerals go above 50, check what the player spends it on next, and how much is left afterwards. For example if a player accumulates 220 minerals and then makes a Factory, it means this was a deliberate cut. 3. Implement a threshold variable, so people can change it from 50 to a different number, to account for uneven mining rates and other factors. 4. Make the program check for available larva. This would make it useful to Zerg users as well.
With this I believe your numbers would be a lot more accurate.
another key thing to go along with "perfect macro" is that theres no such thing after a certain point in a game. there are way too many factors, you can pump as many workers as you want but if you dont have a cc or nexus to assign them to then whats the point? in fact they are detrimental to you, you'd rather not even have them, they're taking up much needed supply, what happens if you're getting hydra all inned and you need to cut probes for cannons and zealots? that would go against "perfect macro" but if you dont cut probes and defend then you die, i could make a long list of why the tool is obsolete but if you want to think that its going to change the way people macro then you should be the one to prove it. sorry you clearly have skills in making these types of programs, but this particular one has no use or very little to me and many others
EDIT: people are coming off as harsh because the title gives the impression that you made a program that is going to revolutionize the way people macro and potentially change the game, which is far from the truth, sorry but this is the reality of it
On November 03 2018 06:17 rand0MPrecisi0n wrote: I think you can actually make this tool much better by improving the algorithm in the following ways:
1. Always check if the player has enough minerals to make a worker(not just in the beginning) 2. If the minerals go above 50, check what the player spends it on next, and how much is left afterwards. For example if a player accumulates 220 minerals and then makes a Factory, it means this was a deliberate cut. 3. Implement a threshold variable, so people can change it from 50 to a different number, to account for uneven mining rates and other factors. 4. Make the program check for available larva. This would make it useful to Zerg users as well.
With this I believe your numbers would be a lot more accurate.
Most importantly, make programs compatible with remastered.
On November 03 2018 02:21 [sc1f]eonzerg wrote: Mostaza u re 3/3 in useless threads grats !
xD Eloquently put!
Moataz, you need to re-read the comments because based on your replies it seems as though you have a hard time understanding what the more experienced players are telling you (a common theme with you).
1) Yes, you are right, even the very best players are still human. Artosis and Tasteless say that they are 'gods' because it's their job as sports commentators to get excited and to get everyone watching excited as well. If you listen to them, you know that they also mention that it's impossible to play a 'perfect' game of Starcraft. The idea is that you decide which mistakes you make, and which mistakes you avoid. That's the tricky thing to master, and that's what these players are insanely good at
2) Yes, sometimes chosing to make the mistake of not making a worker is the right choice
3) Yes, probably a lot of situations of which you think they are mistakenly not making workers are intentional and your algorithm doesn't account for the complexity of that
edit1:
On November 03 2018 06:15 REDBLUEGREEN wrote: Actually at the Italy event when foreigners asked FBH for advice FBH told them unironically to work on constant worker production and quickly sending them to mine and that it was something even he and other pros can improve on ¯\_(ツ)_/¯
True. And we have acknowledged as much:
On November 03 2018 06:07 Sero wrote: This "tool" does one thing: tells us progamers don't have literally perfect macro. We know... no one thinks they did, because that's impossible. What stunning discovery will your next thread contain?
But the thing is guys... It's awesome that new people are tearing apart games and replays to figure out how things work. Maybe it's nothing new for us, but it's awesome that Moataz put in the effort. He thought he discovered something new... And this is why this game is so awesome, because we can always dig in and discover new things. So congrats Moataz and keep on digging! BW is a wonderful and complex game.
edit2: And don't let your enthusiasm be curbed just because your discoveries are not actually as revolutionary as you have thought. This game is is old, people have been thinking about it for 20 years... Maybe one day you'll discover something people will be amazed at, and write a tool people will love!
On November 03 2018 06:15 REDBLUEGREEN wrote: Actually at the Italy event when foreigners asked FBH for advice FBH told them unironically to work on constant worker production and quickly sending them to mine and that it was something even he and other pros can improve on ¯\_(ツ)_/¯
honestly man he probably saw alot of noobs playing in that event and realized that they first need to learn how to macro and play mechanically before they start learning more in depth strategy and that was the best advice to give them on the spot. i dont know many mid-high level italian players, there are a few i guess but i dont even see any in BSL this season
On November 03 2018 05:13 Moataz wrote: What I really think is waste of time, is when you try to explain tool to someone who is not intended to buy it, or even test it for free, and then he comes to complain with very few infos he knows, and you have to correct some misunderstanding for him, when you're also have written some good manual of what this tool does.
You're right, my bad. We're all idiots here who don't understand the magnitude, the sheer impact, of what you have accomplished. Sorry for wasting your time.
Exactly! That's all he's trying to say, we're just idiots.
On November 03 2018 04:45 rand0MPrecisi0n wrote: I can tell you right now if Bisu is not making workers when the only thing he has are Dragoons shooting a bunker, he probably has a very good reason to do so.
Thread would be a lot better if you posted the exact replays you used so people can look into it themselves. You're basically holding back the most important piece of evidence.
I just looked into the Flash vs Free replay, and here's the breakdown of the first 9 minutes.
1. Flash goes for CC first, as he does in every game of this pack 2. He then cuts to make his rax at 15 supply. This is deliberate as he needs to adjust the timing of his marines to compensate for the extra economy 3. Another small cut after getting the refinery down 4. Another cut to afford a bunker and a factory at the same time. Again, priorities. 5. Another cut when his e-bay finishes, to get turrets up 6. Another cut to afford an Academy 7. Yet another one to make a second Factory 8. Another when he makes an Armory
During this, he does lose a few seconds between cycles, due to: 1. Microing his scout 2. Microing his army when he scans the DT at the natural 3. Microing his army to take a third base 4. Scanning Free's base
I'd say that adds up to a total of under 20 seconds (as opposed to the 225s you arrived at). He also loses a second here and there waiting for 100 minerals, in order to start SCVs in both CCs at the same time, which is a common macro technique used to make things smoother.
Beware though, that even some of the cuts while microing are unavoidable, as micro is something that often requires you to react to what the opponent is doing, so it can't be streamlined the same way as macro can.
For example when he scans the DT, Free may try to run away in a number of directions, or he may try to attack his marines, all of which require different inputs by Flash to optimize the exchange. That's not to say he couldn't have made a worker there, but that it's not that simple.
It's an interesting initiative your have, but what you're going for has a level of complexity eons higher than you seem to think. If I had your skills I'd make a bot that allows people to tell the computer what to do. Something similar to the training stage in Street Fighter, where you can control the AI character, press Record, do a move or combo, then go back to your own character and practice against that move over and over.
Something like this would be very useful to create micro scenarios on the fly, instead of having to deal with map editing and triggers, not to mention the limitations of both of those things.
Thanks rand0MPrecisi0n for taking the time to write this, Bisu replay was already there in the OP from start, and this the FPV (if you're interested).
Can I ask you what tool you're using for analyzing replays? Can you look @remake of Flash replay @9:00 and tell us how can anyone have ~20 workers more than Flash, is it some kind of cheat/hack ?
The answer is simple, just click this, and search the page for the word "tripled".
I'm not good @c++, and the lowest BWAPI developer might've more knowledge than me, they're just crazy people fighting each other the whole day (I'm kidding, nobody tells them, or they won't help me again)
On November 03 2018 04:45 rand0MPrecisi0n wrote: I can tell you right now if Bisu is not making workers when the only thing he has are Dragoons shooting a bunker, he probably has a very good reason to do so.
Thread would be a lot better if you posted the exact replays you used so people can look into it themselves. You're basically holding back the most important piece of evidence.
I just looked into the Flash vs Free replay, and here's the breakdown of the first 9 minutes.
1. Flash goes for CC first, as he does in every game of this pack 2. He then cuts to make his rax at 15 supply. This is deliberate as he needs to adjust the timing of his marines to compensate for the extra economy 3. Another small cut after getting the refinery down 4. Another cut to afford a bunker and a factory at the same time. Again, priorities. 5. Another cut when his e-bay finishes, to get turrets up 6. Another cut to afford an Academy 7. Yet another one to make a second Factory 8. Another when he makes an Armory
During this, he does lose a few seconds between cycles, due to: 1. Microing his scout 2. Microing his army when he scans the DT at the natural 3. Microing his army to take a third base 4. Scanning Free's base
I'd say that adds up to a total of under 20 seconds (as opposed to the 225s you arrived at). He also loses a second here and there waiting for 100 minerals, in order to start SCVs in both CCs at the same time, which is a common macro technique used to make things smoother.
Beware though, that even some of the cuts while microing are unavoidable, as micro is something that often requires you to react to what the opponent is doing, so it can't be streamlined the same way as macro can.
For example when he scans the DT, Free may try to run away in a number of directions, or he may try to attack his marines, all of which require different inputs by Flash to optimize the exchange. That's not to say he couldn't have made a worker there, but that it's not that simple.
It's an interesting initiative your have, but what you're going for has a level of complexity eons higher than you seem to think. If I had your skills I'd make a bot that allows people to tell the computer what to do. Something similar to the training stage in Street Fighter, where you can control the AI character, press Record, do a move or combo, then go back to your own character and practice against that move over and over.
Something like this would be very useful to create micro scenarios on the fly, instead of having to deal with map editing and triggers, not to mention the limitations of both of those things.
Thanks rand0MPrecisi0n for taking the time to write this, Bisu replay was already there in the OP from start, and this the FPV (if you're interested).
Can I ask you what tool you're using for analyzing replays? Can you look @remake of Flash replay @9:00 and tell us how can anyone have ~20 workers more than Flash, is it some kind of cheat/hack ?
The answer is simple, just click this, and search the page for the word "tripled".
I'm not good @c++, and the lowest BWAPI developer might've more knowledge than me, they're just crazy people fighting each other the whole day (I'm kidding, nobody tells them, or they won't help me again)
I didn't use any tool besides the game itself. I'll have a look at the remake
On November 02 2018 11:30 Jealous wrote: Before I jump to conclusions, have you verified that they weren't saving money for something more important than Probes, or that they simply don't have money for the Probe? For example, it is necessary to stop Probe production for a short amount of time after placing 11 Assimilator in PvT
While I agree that there are many situations where you cut workers in order to use minerals for something else, placing your assimilator at 11 supply in PvT is not one of them. You can easily place the assimilator at 11 without ever having zero probes in your Nexus after your 6th probe.
Maybe I'm just bad at splitting or deciding which mineral to send new Probes to, but it seems that when the 11th Probe finishes I always have 30-40 minerals and have to wait a tick or two to start the next one.
Or do you scout very early? That also has an impact on minerals and probe production early on. There are so many factors that play into everything, pros definitely have their build orders calculated down to a lot of those things that we just do by intuition.
Ok I looked into it, and I have to say it gave me a laugh.
It's not a remake if the player isn't building the same things, at the same time. For example, you took your first gas at 30+ supply, while Flash does it at 16. That means you have 3 extra workers on minerals for a few minutes. You're also building less things, which means less workers busy building stuff, and more mining.
You also didn't do any of the deliberate worker cuts he did, so obviously you're gonna have more stuff. For example he makes his CC at 13, then makes a worker with the next 50 minerals, whereas you make the worker first and delay the CC.
Lastly, you take the second CC at the natural position, whereas Flash takes it in his main, as part of a wall. That means you have access to seven more mineral patches, which will increase your mining rate considerably.
On November 02 2018 22:49 Dangermousecatdog wrote: If the responses seem overwhelmingly negative it's because most of us had seen this misconception that you "have" to continuously produce workers many times before. Also that clickbait title and the sheer presumptioness probably doesn't help either.
"Continuously build workers" is a great heuristic to learn the game and improve yourself, but once you can continuously make workers automatically, you should throw it away and it's not to be used to analyse pro replays anyways. From your picture, 9 mins into the game Best has 3 mining bases and 63 probes. That's actually already over the "optimal" number of mining probes.
Does he need an extra 10 probes when he already have enough probes for 3 bases 9 mins into the game? Probes aren't free to make. Those 10 cut probes? Pretty much guaranteed they were cut so to make units, nexus, production buildings and tech up to hit certain important timings or simply not die. Maybe he is massing gateways in preparation for a big push to deny the third, or just straight up tech to carriers and arbiters.
Props to you though if you actually made an AI that perfectly copies the build with the the same tech and buildings and units and still came out 10 probes ahead though, that would be amazing. But it doesn't seem to do that.
You mean Bisu replay, the 9mins isn't imperative, you can change it to whatever you like, if this is over the "optimal" number of mining probes, the player can take 4th base, or better, only build, for example, 2nd base and test to see what the optimal time to reach 50 workers from a specific 2nd base-timing (maybe it 7mins, not 9).
I think its more than doable that a human (not even AI) can imitate same Bisu build taking the economy 1st approach, and have a bigger army/tech, but he (might) want to delay some things few seconds especially early on when he doesn't have the money to make everything.
Edit: I think I'm exaggerating on the topic title, but sometimes you need to do this to make people listen, I hope I didn't deceive anyone.
It's not about X amount of probes at X amount of times in a vacuum. You are under the assumption that probes are free to make. If probes are free to make, you would be absolutely right, that those probes not made should had absolutely been made, but they are not, and so you are wrong. If you think 9 mins is not the right time, that's the time in your pic and written examples; it's up to you to go choose another time.
I just watched your replay with your AI program. All your AI does is making probes continuously and no other unit, but a single dragoon (lol) 9 mins into the game. You are under the assumption that probes are free to make. They are not. You can't say your AI build is better, when you would just die to anything attacking you.
Probes aren't free to make so you can't just make a 4th base, because that too costs 400 minerals. It's like TvZ, you have to cut marines instead of scvs if you want to 19-20CC as opposed to 21 CC for a better macro later. Same with protoss. Sometimes the build in early P requires you to wait for the pylon to finish and you can't build another probe, but that means you can make units immediately after and not die to an unexpected push. Now imagine that with waiting to gather enough minerals for various production buildings, for pylons, for units, for tech ups, there will be plenty of time that Best will be forced to wait if he want those buildings and production at certain timings.
On November 03 2018 06:07 Sero wrote: This "tool" does one thing: tells us progamers don't have literally perfect macro. We know... no one thinks they did, because that's impossible. What stunning discovery will your next thread contain?
Thanks, I don't know yet What stunning discovery will my next thread contain?, stay tuned.
On November 03 2018 05:13 Moataz wrote: What I really think is waste of time, is when you try to explain tool to someone who is not intended to buy it, or even test it for free, and then he comes to complain with very few infos he knows, and you have to correct some misunderstanding for him, when you're also have written some good manual of what this tool does.
You're right, my bad. We're all idiots here who don't understand the magnitude, the sheer impact, of what you have accomplished. Sorry for wasting your time.
No problem.
On November 03 2018 06:18 castleeMg wrote: another key thing to go along with "perfect macro" is that theres no such thing after a certain point in a game. there are way too many factors, you can pump as many workers as you want but if you dont have a cc or nexus to assign them to then whats the point? in fact they are detrimental to you, you'd rather not even have them, they're taking up much needed supply, what happens if you're getting hydra all inned and you need to cut probes for cannons and zealots? that would go against "perfect macro" but if you dont cut probes and defend then you die, i could make a long list of why the tool is obsolete but if you want to think that its going to change the way people macro then you should be the one to prove it. sorry you clearly have skills in making these types of programs, but this particular one has no use or very little to me and many others
EDIT: people are coming off as harsh because the title gives the impression that you made a program that is going to revolutionize the way people macro and potentially change the game, which is far from the truth, sorry but this is the reality of it
I'm not crazy to the degree that I would say "build workers even when you're dying", these specific cases you guys talking about is certainly wiser in it, to cut workers, but I'm talking about normal games, like Bisu or Flash one, Bisu was not doing much micro other than attacking Bunkers.
On November 03 2018 06:17 rand0MPrecisi0n wrote: I think you can actually make this tool much better by improving the algorithm in the following ways:
1. Always check if the player has enough minerals to make a worker(not just in the beginning) 2. If the minerals go above 50, check what the player spends it on next, and how much is left afterwards. For example if a player accumulates 220 minerals and then makes a Factory, it means this was a deliberate cut. 3. Implement a threshold variable, so people can change it from 50 to a different number, to account for uneven mining rates and other factors. 4. Make the program check for available larva. This would make it useful to Zerg users as well.
With this I believe your numbers would be a lot more accurate.
Most importantly, make programs compatible with remastered.
Unfortunately, this is up to Blizzard, but you can still use both SCR & 1.16, if you want.
On November 03 2018 05:13 Moataz wrote: What I really think is waste of time, is when you try to explain tool to someone who is not intended to buy it, or even test it for free, and then he comes to complain with very few infos he knows, and you have to correct some misunderstanding for him, when you're also have written some good manual of what this tool does.
You're right, my bad. We're all idiots here who don't understand the magnitude, the sheer impact, of what you have accomplished. Sorry for wasting your time.
Exactly! That's all he's trying to say, we're just idiots.
Another quality thread by Moataz.
I didn't say that.
On November 03 2018 07:06 MamiyaOtaru wrote: I don't even visit much anymore but I do know when I see a Moataz thread it's going to be amazing
That's some effort you've put in!
Well, this might be even more popular, stay tuned, share it, like it, subscribe, etc, everything you do will help the thread, whether you've logic or not, it doesn't matters.
On November 03 2018 07:17 rand0MPrecisi0n wrote: Ok I looked into it, and I have to say it gave me a laugh.
It's not a remake if the player isn't building the same things, at the same time. For example, you took your first gas at 30+ supply, while Flash does it at 16. That means you have 3 extra workers on minerals for a few minutes. You're also building less things, which means less workers busy building stuff, and more mining.
You also didn't do any of the deliberate worker cuts he did, so obviously you're gonna have more stuff. For example he makes his CC at 13, then makes a worker with the next 50 minerals, whereas you make the worker first and delay the CC.
Lastly, you take the second CC at the natural position, whereas Flash takes it in his main, as part of a wall. That means you have access to seven more mineral patches, which will increase your mining rate considerably.
I didn't say its complete remake, it was just to prove that a player other than Flash could have ~20 workers more, you can try to mimic the build exactly but that is not the point.
And you're analyzing the replay strategically, while the purpose is to look at the macro only.
And unfortunately, you didn't answer "how can anyone have ~20 workers more than Flash, is it some kind of cheat/hack ?"
Lastly, did you look @the game time @9:00 and see 73 SCV ?
On November 03 2018 07:29 Dangermousecatdog wrote:
On November 02 2018 22:49 Dangermousecatdog wrote: If the responses seem overwhelmingly negative it's because most of us had seen this misconception that you "have" to continuously produce workers many times before. Also that clickbait title and the sheer presumptioness probably doesn't help either.
"Continuously build workers" is a great heuristic to learn the game and improve yourself, but once you can continuously make workers automatically, you should throw it away and it's not to be used to analyse pro replays anyways. From your picture, 9 mins into the game Best has 3 mining bases and 63 probes. That's actually already over the "optimal" number of mining probes.
Does he need an extra 10 probes when he already have enough probes for 3 bases 9 mins into the game? Probes aren't free to make. Those 10 cut probes? Pretty much guaranteed they were cut so to make units, nexus, production buildings and tech up to hit certain important timings or simply not die. Maybe he is massing gateways in preparation for a big push to deny the third, or just straight up tech to carriers and arbiters.
Props to you though if you actually made an AI that perfectly copies the build with the the same tech and buildings and units and still came out 10 probes ahead though, that would be amazing. But it doesn't seem to do that.
You mean Bisu replay, the 9mins isn't imperative, you can change it to whatever you like, if this is over the "optimal" number of mining probes, the player can take 4th base, or better, only build, for example, 2nd base and test to see what the optimal time to reach 50 workers from a specific 2nd base-timing (maybe it 7mins, not 9).
I think its more than doable that a human (not even AI) can imitate same Bisu build taking the economy 1st approach, and have a bigger army/tech, but he (might) want to delay some things few seconds especially early on when he doesn't have the money to make everything.
Edit: I think I'm exaggerating on the topic title, but sometimes you need to do this to make people listen, I hope I didn't deceive anyone.
It's not about X amount of probes at X amount of times in a vacuum. You are under the assumption that probes are free to make. If probes are free to make, you would be absolutely right, that those probes not made should had absolutely been made, but they are not, and so you are wrong. If you think 9 mins is not the right time, it's up to you to go choose another time. I just watched your replay with your AI program. All your AI does is making probes continuously and no other unit, but a single dragoon (lol) 9 mins into the game. You are under the assumption that probes are free to make. They are not. You can't say your AI build is better, when you would just die to anything attacking you.
Probes aren't free to make so you can't just make a 4th base, because that too costs 400 minerals. It's like TvZ, you have to cut marines instead of scvs if you want to 19-20CC as opposed to 21 CC for a better macro later. Same with protoss. Sometimes the build in early P requires you to wait for the pylon to finish and you can't build another probe, but that means you can make units immediately after and not die to an unexpected push. Now imagine that with waiting to gather enough minerals for various production buildings, for pylons, for units, for tech ups, there will be plenty of time that Best will be forced to wait if he want those buildings and production at certain timings.
The replay was just to prove that you (can) have 10 Probes more than Bisu from that specific base-timing, he wasn't microing heavily, so I guess the only reason for him not to macro better, is that he doesn't know that he could macro better.
I don't have the time to mimic the build exactly, but you might, I'm sure its doable to have the same units/buildings/tech as Bisu in 9:00 and more workers and more everything, because Bisu himself in some reply I don't remember right now was only 1:30 WorkerCut @9:00.
You still don't understand. You haven't proved that you can have 10 Probes more than Bisu from that specific base-timing, because you haven't followed that specific base-timing at all. In fact you haven't followed any timing at all except the nexus timings. Do you know what tech, units and buildings he is aiming to have? Maybe there is leeway in his build to wait and scout and respond to his opponent? Do you have his tech, his units, his production? No you don't; all you have is a cyber core and a gateway and a dragoon.
You have no idea what Best is trying to acheive, what timings he is trying to hit all before the 9 mins and beyond the 9 mins, and how neccessary it is to cut probes or not, nor do you seem to have any idea what probe count he is trying to hit. Heck for all we know, he might had hit all his timing exactly and he was aiming to have exactly 63 probes at 9 mins and never makes anymore probes for the rest of the game.
All you know is that he cut workers, but you don't know why, yet you assume that he was aiming to make as many workers as possible at the expense of everything else. You got a massive knowledge gap which I am telling you exactly what you are lacking, but you are refusing to acknowledge it.
On November 03 2018 08:16 Moataz wrote: I didn't say its complete remake, it was just to prove that a player other than Flash could have ~20 workers more, you can try to mimic the build exactly but that is not the point.
And you're analyzing the replay strategically, while the purpose is to look at the macro only.
And unfortunately, you didn't answer "how can anyone have ~20 workers more than Flash, is it some kind of cheat/hack ?"
Lastly, did you look @the game time @9:00 and see 73 SCV ?
xD
I think we can clearly say that we are getting trolled hard based on this response :D Nice thread
This entire post and all the responses you've given just sound super condescending and makes me think you don't really play the game that much. The title itself makes me roll my eyes. Nobody plays this game perfectly. If you had just left it as "this is a way to help you practice making SCVs in a no-rush 15 game" then I would've accepted it and moved it. But to say shit like
On November 03 2018 08:16 Moataz wrote: The replay was just to prove that you (can) have 10 Probes more than Bisu from that specific base-timing, he wasn't microing heavily, so I guess the only reason for him not to macro better, is that he doesn't know that he could macro better.
is just irritating. You really think that Bisu, a guy who is actually good at Starcraft doesn't know he could always macro better?
Also,
On November 03 2018 08:16 Moataz wrote: I didn't say its complete remake, it was just to prove that a player other than Flash could have ~20 workers more, you can try to mimic the build exactly but that is not the point.
And you're analyzing the replay strategically, while the purpose is to look at the macro only
Sorry, but Starcraft isn't a game where you can just macro and be done with it. Corners are cut for a reason. If you're going to look at somebody's macro, you need context.
The graphs are cool and all, (certainly filled with data), but I don't exactly see how this would be practical for a learning player or a pro to utilize in an actual game.
Maybe OP could enlighten me, but I feel like comparing an AI's "perfect" gameplay and implementing that straight to the human's play isn't optimal or the best approach. Not saying you can't learn from it, but I wouldn't go as far as saying this is going to be ground-breaking. We're all humans, after all.
I guess the Mathematica point here is to not cut workers, since everybody does this at some point. But players know they need to protect themselves with units and army which can come at a high cost. Idle workers is never good and we already knew that from a mathematical point.
On November 03 2018 09:40 Kurao wrote: The graphs are cool and all, (certainly filled with data), but I don't exactly see how this would be practical for a learning player or a pro to utilize in an actual game.
Maybe OP could enlighten me, but I feel like comparing an AI's "perfect" gameplay and implementing that straight to the human's play isn't optimal or the best approach. Not saying you can't learn from it, but I wouldn't go as far as saying this is going to be ground-breaking. We're all humans, after all.
Ok I'll enlight you, you see that it's not practical because, you didn't try it, as simple as that, many of the people that are talking here are just talking about it, with a very few knowledge about what they're talking about, while they mightn't gave it a 5 mins tryout. fair enough to judge?
On November 03 2018 08:35 Dangermousecatdog wrote: You still don't understand. You haven't proved that you can have 10 Probes more than Bisu from that specific base-timing, because you haven't followed that specific base-timing at all. In fact you haven't followed any timing at all except the nexus timings. Do you know what tech, units and buildings he is aiming to have? Maybe there is leeway in his build to wait and scout and respond to his opponent? Do you have his tech, his units, his production? No you don't all; you have is cyber core and a gateway and a dragoon.
You have no idea what Best is trying to acheive, what timings he is trying to hit all before 9 mins and beyond, and how neccessary it is to cut probes or not, nor do you seem to have any idea what probe count he is trying to hit. Heck for all we know, he might had hit all his timing exactly and he was aiming to have exactly 63 probes at 9 mins and never makes anymore drones for the rest of the game.
All you know is that he cut workers, but you don't know why, yet you assume that he was aiming to make as many workers as possible at the expense of everything else. You got a massive knowledge gap which I am telling you exactly what you are lacking, but you are refusing to acknowledge it.
I know if I made the exact replay with same & more units, more supply, more workers , some of you would still argue, Its up to guys, not the CoachAI (even if Bisu coachs someone that isn't ambitious enough, it won't help)
And you're saying "Best" again while I've told you here that he is Bisu, you're also confusing Probes with Drones, whatever. It doesn't matter why they cut workers, at least that's not the CoachAI job, you have eyes, and can conclude.
On November 03 2018 08:16 Moataz wrote: I didn't say its complete remake, it was just to prove that a player other than Flash could have ~20 workers more, you can try to mimic the build exactly but that is not the point.
And you're analyzing the replay strategically, while the purpose is to look at the macro only.
And unfortunately, you didn't answer "how can anyone have ~20 workers more than Flash, is it some kind of cheat/hack ?"
Lastly, did you look @the game time @9:00 and see 73 SCV ?
xD
I think we can clearly say that we are getting trolled hard based on this response :D Nice thread
Yes, it is nice thread, it tells how many argumentative/destructive/envy/spiteful persons in this world, thanks for clarifying that more, I'm out of this thread, as you suggest, but will continue to develop the CoachAI, if you don't mind.
I just remember a similar dead topic in TL that has less than tithe the features that in the CoachAI, this topic has +300 comment and all guys there are looking at it as a miracle ! Sorry, I could only fix bugs, but not envy/grudge.
Please also note that I'm taking about Korean pros in this topic, not the poor guys below, but even if a Korean pro denied that this tool might help him and that he has all the focus in game nevertheless and he doesn't need any statistical info about his games !, other players may use this against him and see what he can't, and it would still be a helper tool for them, Bye
I don't think it's wrong to suggest that these are things even pros could be better at, even things as "basic" as worker macro. I think OP's tool is too blunt, but I presume his point is that a perfect AI could make 10, 15, 30 more workers over the course of a game while carrying out exactly the same timings as a human because the AI could calculate the margins and squeeze in workers every single time the mineral count went above 50 in a way that didn't deviate from the human's actual timings. Let's say you're on an X mineral patch map with Z scvs saving for 200 to make a factory. A computer could calculate exactly how long it would take Z scvs mining X mineral patches and their pathing to those patches to get some multiple of 8 above 200 minerals and exactly when to send the worker out from the mineral line to arrive right at that time. A human can't; even if their timing looks right it'd still be microseconds off. And let's say the human gets distracted for 1 second microing his SCV in the opponent's base, and when he makes the fac he's actually at 250+ minerals. That means, technically, he could have made an SCV before even starting to save for the fac and probably had it out and mining by the time the fac actually started. Over the course of the next 5 minutes that SCV would bring in a lot of money. But that's just the reality of human abilities; they're not machines. And because real people don't have 10,000 APM and can't control multiple things literally simultaneously, they prioritize certain things over others. The 1 second of saving the SCV might yield valuable scouting information that then negates the minor disadvantage of having not made that extra SCV. Who knows. SC is a game of variables and prioritization. Obviously perfection is the goal but it's unattainable.
Moataz, I understand your point of view, so please understand players' point of view: - we have no map hack, we often skip worker production because we don't feel safe, we don't want to die, so we prioritize safety over greed - and the other way around, sometimes we want to make as many units as possible, to hit a certain timing, hoping the opponent is gonna over-macro
If your intention is to say that we are mistaken to cut worker production due to safety, timings etc, you are WRONG If your intention is to make us pay more attention during the game to not skip worker production BY ACCIDENT, you are helping. But then again, sometimes it's hard to distinguish a human's error from a safety precaution.
So it's mostly about decision making.
Also, I would find it more helpful if you posted a couple of tables with calculations how much minerals we ended up behind when continuosly making workers, but not sending them to minerals right away. For example if I have made 5 idle probes, and send them to minerals with the 6th one just popping out. A table depending on how many probes I currently have in the mineral line and how many mineral fields are in the base. That would be some valueable knowledge for me.
On November 03 2018 07:17 rand0MPrecisi0n wrote: Ok I looked into it, and I have to say it gave me a laugh.
It's not a remake if the player isn't building the same things, at the same time. For example, you took your first gas at 30+ supply, while Flash does it at 16. That means you have 3 extra workers on minerals for a few minutes. You're also building less things, which means less workers busy building stuff, and more mining.
You also didn't do any of the deliberate worker cuts he did, so obviously you're gonna have more stuff. For example he makes his CC at 13, then makes a worker with the next 50 minerals, whereas you make the worker first and delay the CC.
Lastly, you take the second CC at the natural position, whereas Flash takes it in his main, as part of a wall. That means you have access to seven more mineral patches, which will increase your mining rate considerably.
I didn't say its complete remake, it was just to prove that a player other than Flash could have ~20 workers more, you can try to mimic the build exactly but that is not the point.
And you're analyzing the replay strategically, while the purpose is to look at the macro only.
And unfortunately, you didn't answer "how can anyone have ~20 workers more than Flash, is it some kind of cheat/hack ?"
Lastly, did you look @the game time @9:00 and see 73 SCV ?
Wouldn't you always just straight up lose the game if you don't build a refinery until 30 supply?
I appreciate anything that people like you do to keep the BW scene fresh@Moataz and i can see why the negative feedback can offend you, it's work and time you put in but don't take it personal. I wasn't trying to come off as mean spirite. In my opinion I just think it's more complex than " macro better "
On November 03 2018 10:18 Moataz wrote: Please also note that I'm taking about Korean pros in this topic, not the poor guys below, but even if a Korean pro denied that this tool might help him and that he has all the focus in game nevertheless and he doesn't need any statistical info about his games !, other players may use this against him and see what he can't, and it would still be a helper tool for them, Bye
Did you need help sticking your head this far up your own ass, or did you achieve it all on your own?
That's pretty cool stuff, Moataz! It's nice to have some replay analysis tools like this. Make sure you blogpost on your findings; it'll look good on a dev cv. Keep it up, and don't listen to the naysayers.
What some people might find useful is to add a feature that splits the idle time of a Nexus/CC/Hatchery into two categories:
"Calculated" (deliberate) idle time
"Wasted" (unnecessary) idle time
Essentially what you would do is detect when players build something other than a worker and determine if it is a deliberate worker cut or if the player fucked up. This would allow players to pinpoint moments where worker production halted because of a distraction or a mistake and not because of the build order they were following in that game. It would be up to the player to determine which worker production halts were for good reason (e.g. there was a massive battle happening somewhere at that time) and which were because they messed stuff up. That would give players a good idea on what to focus and work on during their following games.
I've included below some pseudocode to give you an idea of what it should look like, there might be a few bugs so a bit of tweaking would be required in that case:
Is the player currently making a worker? Yes > OK No > Does the player currently have enough minerals to make a worker? No > OK Yes > Since the player isn't currently making a worker and has the minerals to do so, there are two possible scenarios going on: a) The player is cutting worker production in order to make something else as part of their build order ("Calculated" Idle Time). b) The player forgot/is distracted/sucks ("Wasted" Idle Time). Set boolean flag isPlayerCuttingWorkersOnPurpose = false Do the following until the player makes another worker { Did the player build/train/research something other than a worker? Yes (e.g. the player made a Gateway) > After deducting the cost of building/training/researching, did the player have 50 or more minerals left over? No > OK (this means the player is following a precise build order and a calculated cut in worker production): Set isPlayerCuttingWorkersOnPurpose = true Calculate time between minerals > 50 and now, add this time to "calculated" idle time Yes > The player fucked up, add idle time to "wasted" idle time (this means that the player could have made a worker anyways). No (e.g. the player made a Probe) > Did the player cut workers before this in order to build/train/research something else? Yes (isPlayerCuttingWorkersOnPurpose == true) > Did the player have more than 50 minerals when making the worker? No > OK (no "wasted" idle time) Yes > The player fucked up, calculate time between minerals > 50 and now, add this time to "wasted" idle time. No (isPlayerCuttingWorkersOnPurpose == false) > The player fucked up, calculate time between minerals > 50 and now, add this time to "wasted" idle time. }
On November 03 2018 06:15 REDBLUEGREEN wrote: Actually at the Italy event when foreigners asked FBH for advice FBH told them unironically to work on constant worker production and quickly sending them to mine and that it was something even he and other pros can improve on ¯\_(ツ)_/¯
Oh why didn't I see that early,
So if I've FBH's fame/skill when I've made this topic, that would make much sense ? really ?, I mean rarely anyone would disagree with famous casters (because if someone does: How dare u, stupid newbie, these guys understand the game more than anyone else, what do u know kid ?),
Also, I'm not giving or asking game/strategic advises here, I'm just a programmer, so the talk about (why) players cut workers or the amount of tasks the player need to do isn't relevant/useful, it doesn't help/harm what the tool does, and doesn't lessen the discovery.
workerCutLimit, you can set this to 120 (as a goal/target), this will end the game immediately if you cut workers for 2 minutes in the default 1st 9 minutes, and you'll get the message: "WorkerCut too high !"
When you enable the custom countdownTimer (dontDrift value) and it reaches 00:15, visual/sound warning will start (this helped me remembering to save the game-state @lets say 6:00), in order to master the game-phases/tasks partially 1st, then as a whole.
Minerals-threshold sound warning "Spend more minerals", if 500-750 every 3s, if > 750 every 1s, until 15 minutes of the game.
So wise guys, ahem ahem... "Lets agree to disagree"
There is a reason behind cutting workers at specific times in those builds Bisu / Best / Flash were doing, please believe that they know a lot more about the feel / what is going on exactly in the game more so than you do, there are reasonings behind cutting workers around 9 minutes, interesting post but yet again, just another wild Moataz Theory.
I love how he is still plugging his idea that Bisu is bad at macro because he has 120 seconds of idle time totalled across 3 nexus at 9 mins, instead of that Bisu simply had enough probes by that time.
A lot of people here keep rehashing the same point: that people intentionally cut worker production. Congratulations on being smart enough to notice that, but not being patient enough to realize that the OP now undertands this. That doesn't mean that your worker production is optimized. I feel like this tool could be really useful for coming up with new builds.
If there were still pro teams, then coaches would take this tool seriously and try to see if there was a way they could exploit it to improve team performance. And that is really all the OP is saying: "here I made a flexible tool with many features, try to see what use you can make of it, and let me know how it goes!". Once you realize this, many of the responses in this thread are perplexing.
On December 23 2018 04:18 Rodya wrote: A lot of people here keep rehashing the same point: that people intentionally cut worker production. Congratulations on being smart enough to notice that, but not being patient enough to realize that the OP now undertands this. That doesn't mean that your worker production is optimized. I feel like this tool could be really useful for coming up with new builds.
If there were still pro teams, then coaches would take this tool seriously and try to see if there was a way they could exploit it to improve team performance. And that is really all the OP is saying: "here I made a flexible tool with many features, try to see what use you can make of it, and let me know how it goes!". Once you realize this, many of the responses in this thread are perplexing.
What about the point that making workers and sending them to mine when there are more pressing things to attend to is an inevitable part of Brood War even at pro level, thus having constant worker production/sending to mine as a foundation for some super optimized build is unrealistic and demonstrated the lack of perspective that Moataz has?
What if I made a bot that had superhuman Probe micro, maybe pro coaches would realize that Probe rush is the best strategy in the game and that pros should just micro their Probes better? Or that we can skip Corsair for scouting because if your Probe micro is so good that it can't ever be killed by slow lings... Maybe pro Protoss need to just focus on microing their Probe in the Zerg main over anything else, like how Moataz suggests that pros need to just make and send workers to mine over anything else? After all, even 70 apm player can do that right?
I logged into this account for the first time in years to post this as my now only comment that you are an idiot who needs to L2P, everyone who is telling you that your head is up your ass and your idea/tool sucks is 100% correct and all the people in here supporting you are also hopeless idiots. Bye forever!
On December 23 2018 04:18 Rodya wrote: A lot of people here keep rehashing the same point: that people intentionally cut worker production. Congratulations on being smart enough to notice that, but not being patient enough to realize that the OP now undertands this. That doesn't mean that your worker production is optimized. I feel like this tool could be really useful for coming up with new builds.
If there were still pro teams, then coaches would take this tool seriously and try to see if there was a way they could exploit it to improve team performance. And that is really all the OP is saying: "here I made a flexible tool with many features, try to see what use you can make of it, and let me know how it goes!". Once you realize this, many of the responses in this thread are perplexing.
That's the thing. He actually doesn't, otherwise he still wouldn't be plugging his silly 9 mins, 2 mins no workers idea. You don't have to be a pro to look at that if a pro player has 63 probes at 9 mins to figure that probably he stopped making probes because he has enough of them already.
On December 23 2018 06:00 XenOsky wrote: this is a good tool for low level players, i don't get all the hate.
It's pointless because anybody with basic maths can in less than a minute do the exact same thing. In fact, if you would just watch and copy a pro build, it'll be more useful than this program.
On December 23 2018 06:00 XenOsky wrote: this is a good tool for low level players, i don't get all the hate.
It's pointless because anybody with basic maths can in less than a minute do the exact same thing. In fact, if you would just watch and copy a pro build, it'll be more useful than this program.
Everyone who is decent at this game has already copied pro builds. The question is what comes next? Look at the pictures in the first post. If you are an experienced player (say around 2k) and you wanted to take your game to the next level, one way you can do that is to take a vod of a pro player using a build, and plug in this CoachAI to see how many more idle seconds you have compared to them at each point in the game.
You can use a pro replay to see when exactly they cut workers, and you can then compare how much army production you have as them exactly in 3 minutes.
You can look at your own games, and see the difference in idle time between you and Flash, and you can figure out exactly how many minerals you are missing as a result of that. I imagine the tool could be updated so that barracks and factories can be watched for idle time. I don't see how this tool is useless, personally. It seems like it can give you specific numbers for things that people already look for in a replay. When I look at my replay, I just say "oh, too much worker idle time", or "oh, my macro isn't fast enough". It would be nice to have a specific number whose progress I could track and compare it to my favorite players.
On December 23 2018 10:24 Dangermousecatdog wrote:
On December 23 2018 06:00 XenOsky wrote: this is a good tool for low level players, i don't get all the hate.
It's pointless because anybody with basic maths can in less than a minute do the exact same thing. In fact, if you would just watch and copy a pro build, it'll be more useful than this program.
Everyone who is decent at this game has already copied pro builds. The question is what comes next? Look at the pictures in the first post. If you are an experienced player (say around 2k) and you wanted to take your game to the next level, one way you can do that is to take a vod of a pro player using a build, and plug in this CoachAI to see how many more idle seconds you have compared to them at each point in the game.
You can use a pro replay to see when exactly they cut workers, and you can then compare how much army production you have as them exactly in 3 minutes.
You can look at your own games, and see the difference in idle time between you and Flash, and you can figure out exactly how many minerals you are missing as a result of that. I imagine the tool could be updated so that barracks and factories can be watched for idle time. I don't see how this tool is useless, personally. It seems like it can give you specific numbers for things that people already look for in a replay. When I look at my replay, I just say "oh, too much worker idle time", or "oh, my macro isn't fast enough". It would be nice to have a specific number whose progress I could track and compare it to my favorite players.
I do agree that if Moataz can make it measure idle worker time and number workers over time (like in SC2), that would be a great contribution to the Starcraft community.
On another note, I don't think that was main purpose of CoachAI (even though it is probably more useful than the actual purpose, imo), which I think was to show how optimal macro would look like. Moataz said "a 70 EAPM player can make ~20 workers more than Flash @9:00 from the same build ?! ?! ?!" This statement leads me to think that the only way that this is possible is that CoachAI was doing non-stop worker production over all else so it would have to be adapted to be useful for builds that involve worker cuts. I don't think it's impossible though, and maybe he could make the AI do, for example, an idealized Flash Build that accounts for worker cuts and also tries to build the right amount of army units at different times (so that it's not just mindlessly prioritizing worker production over all else), Even though it obviously wouldn't be a perfect model, it would give an decent enough picture of how optimal macro would look like. My main point is that, with regard to its original purpose, it probably isn't super useful at the moment but it might have potential.
- Major improvments to the TimedBo feature including TTS capabilities (time to say goodby to human coaches!). - New hotkeys. - Replay Bo recorder can now show more stuff by pressing CTRL, Upgrade/Tech start time is also logged.
I hadn't seen this thread until today, and when I read the title in the side-bar, I was like "I bet that's a Moataz post" jokingly. Then I was right. lol
I appreciate that you put forth the effort to perform exercises in critical thinking.
On June 18 2019 16:13 ninazerg wrote: I hadn't seen this thread until today, and when I read the title in the side-bar, I was like "I bet that's a Moataz post" jokingly. Then I was right. lol
I appreciate that you put forth the effort to perform exercises in critical thinking.
Thank you nina, I appreciate if you give it a shot and state your experience (what you like/dislike). In the past, I didn't think Nexus gives supply, but nobody is going to be a newbie forever.
Nobody is always right. Moataz included.
Edit: I only came again to this thread because the main CoachAI thread is closed (mods refused to re-open it), I just believe this thing shouldn't be buried for whatever reasons.