|
On December 06 2010 10:17 Skrag wrote:Show nested quote +On December 06 2010 10:07 jacobman wrote:
Eh, I'll respect that. I still tend to think the results are going to be better than through replays. I've had bad replay comparison experiences with so many differences between different replays that I pretty much gave up on it.
One last point though. The affect of drones searching for mineral patches is very minimal. It doesn't happen a ton in the beginning of the game, and later on in the game it simply does not matter since you have a surplus of minerals at all times and no build timings are affected. Yes you may end up with less minerals but all builds will suffer the exact same mineral deficit due to this. This affect only affects replays in the later (after 20ish supply) game because each replay is usually affected differently by it. The effect is not "very minimal". Two minutes is worth about 80 minerals. Granted, that was the most extreme case I saw, adding the last possible worker on a base where 22 was the max, but 20-30 second times were not unheard of, even for earlier workers. It will typically affect different builds in similar ways, but the simulator doesn't take it into account *at all*, and will again have the build doing things before it could actually do them. That's exactly the sort of thing that can have the simulator saying "build X is better than build Y" when it's not actually true, because it executes things in a way that is actually impossible in-game. It's true that most build have the minerals they need most of the time, but there are very key points in all of them where that's not true, and the build is in fact waiting for minerals. Typically this will happen around big spending points, such as the spawning pool, possibly the first queen, and *always* the hatch.
I think you missed my point. My point was that after about 20 supply when you always have an excess in minerals, as far as timings go, it doesn't matter if you're missing the few minerals from a lost scv because you already have enough minerals anyways. Before the 20 supply mark there isn't too much of this problem anyways, because most of that time is spent under 16 workers. That's why I said I think the affect on doing comparisons is minimal. Yes, you may lose 80 minerals for one worker, but it's much more likely to happen after the 20 worker mark, when it doesn't affect timings anymore and when it is a bigger problem for replays rather than simulations.
|
I think you're underestimating the settle-in effect by quite a lot.
I just looked at your 13pool/15hatch replay from the other thread, trying to measure how much mining time was lost due to bouncing workers (which turns out to be *really* hard), and it came up with almost 30 seconds of wasted time by 2:32. That's a 20 mineral difference just in the first 14 workers.
Some of the differences in builds we've been measuring appears to be less than 20 minerals by the 6 minute mark.
|
On December 06 2010 10:44 Skrag wrote:
I think you're underestimating the settle-in effect by quite a lot.
I just looked at your 13pool/15hatch replay from the other thread, trying to measure how much mining time was lost due to bouncing workers (which turns out to be *really* hard), and it came up with almost 30 seconds of wasted time by 2:32. That's a 20 mineral difference just in the first 14 workers.
Some of the differences in builds we've been measuring appears to be less than 20 minerals by the 6 minute mark.
Yeah, but in the beginning that can be pretty much avoided. The rule was that we were going to just rally all the workers to the middle in that replay test. Microing the workers was not allowed.
If I am actually underestimating the final affect that bouncing will have, then honestly any builds where the bouncing will change, through slightly altering building times, which build is better are so close that they can't really be separated.
|
What I've been doing is only getting the economy of my build down. Workers, supply, and Town Halls. Then I have a chart of money/time and I decide how to best spend this. You can then theorycraft forever what to do with whatever money isn't your economy. Then when I've theorycrafted something good I take it into the game and try it out and see what tactics will be good with it. I find this is much faster than trying to refine every build in game. I only have to play through it once per expansion timing, and a few times to get used to the timings.
|
On December 06 2010 10:51 GoldenH wrote:
What I've been doing is only getting the economy of my build down. Workers, supply, and Town Halls. Then I have a chart of money/time and I decide how to best spend this. You can then theorycraft forever what to do with whatever money isn't your economy. Then when I've theorycrafted something good I take it into the game and try it out and see what tactics will be good with it. I find this is much faster than trying to refine every build in game. I only have to play through it once per expansion timing, and a few times to get used to the timings.
haha, my friends were actually just playing age of empires II today 
Are you trying to say that you theorycraft in order to figure out the defendability of a build? I was a little unsure exactly what you were saying.
|
On December 06 2010 10:46 jacobman wrote:Show nested quote +On December 06 2010 10:44 Skrag wrote:
I think you're underestimating the settle-in effect by quite a lot.
I just looked at your 13pool/15hatch replay from the other thread, trying to measure how much mining time was lost due to bouncing workers (which turns out to be *really* hard), and it came up with almost 30 seconds of wasted time by 2:32. That's a 20 mineral difference just in the first 14 workers.
Some of the differences in builds we've been measuring appears to be less than 20 minerals by the 6 minute mark. Yeah, but in the beginning that can be pretty much avoided. The rule was that we were going to just rally all the workers to the middle in that replay test. Microing the workers was not allowed. If I am actually underestimating the final affect that bouncing will have, then honestly any builds where the bouncing will change, through slightly altering building times, which build is better are so close that they can't really be separated.
20 of those 30 seconds came from the last 2 workers added though, which are the ones that are going to be the most difficult to micro perfectly.
Maybe we're thinking about this all wrong though. It is very difficult to get the AI to execute *perfectly*, but it's actually pretty easy to get it to execute build orders imperfectly, and to be perfectly honest, outside of the very top level players, nobody is executing anywhere near perfect anyway.
Maybe setting the AI up to execute the various builds, and measuring that way, would be "good enough".
It would certainly be better than relying on the results of the simulators IMO.
|
On December 06 2010 11:02 Skrag wrote:Show nested quote +On December 06 2010 10:46 jacobman wrote:On December 06 2010 10:44 Skrag wrote:
I think you're underestimating the settle-in effect by quite a lot.
I just looked at your 13pool/15hatch replay from the other thread, trying to measure how much mining time was lost due to bouncing workers (which turns out to be *really* hard), and it came up with almost 30 seconds of wasted time by 2:32. That's a 20 mineral difference just in the first 14 workers.
Some of the differences in builds we've been measuring appears to be less than 20 minerals by the 6 minute mark. Yeah, but in the beginning that can be pretty much avoided. The rule was that we were going to just rally all the workers to the middle in that replay test. Microing the workers was not allowed. If I am actually underestimating the final affect that bouncing will have, then honestly any builds where the bouncing will change, through slightly altering building times, which build is better are so close that they can't really be separated. 20 of those 30 seconds came from the last 2 workers added though, which are the ones that are going to be the most difficult to micro perfectly. Maybe we're thinking about this all wrong though. It is very difficult to get the AI to execute *perfectly*, but it's actually pretty easy to get it to execute build orders imperfectly, and to be perfectly honest, outside of the very top level players, nobody is executing anywhere near perfect anyway. Maybe setting the AI up to execute the various builds, and measuring that way, would be "good enough". It would certainly be better than relying on the results of the simulators IMO.
I'm totally up for that actually. The biggest problem I have with replays is they are inconsistent. That's why I want the computer to do it, except I chose to have the computer do it through simulation. Having the computer actually play through the builds seems like it would work pretty well.
How hard is it to set the computer up to play a build?
|
You could, you know, do what every other researcher anywhere does and aggregate a collection of results, instead of relying on rather flawed simulators...
|
On December 06 2010 13:43 Dominator1370 wrote:
You could, you know, do what every other researcher anywhere does and aggregate a collection of results, instead of relying on rather flawed simulators...
I'm guessing you're willing to do all the very slow games it will take to get a reliable picture of each build too. Researches also use models and simplifications to analyze complex problems that would normally take up too much time and/or resources to be practical.
The difference here is that you suspect the simplifications are not acceptable.
|
Anyone else think that the earlier queen from 11 pool helps a great deal more than you'd expect? Helps make up for the slow arse larvae mechanic Zerg suffers from.
|
"Finding the most defendable economical build" is a bit too general IMO.
First of all, all economic builds are more or less equivalent in term of income. You can't really nitpick a ~100 mineral difference between builds.
Then it comes down to pool vs hatch timings. Faster pool allows you to have zerglings faster. Faster hatch makes you ahead of other builds as far as defending your expo goes... as long as the pool is up fast enough too. Most timings are give or take 15-30sec build from the pool or the hatch.
My personnal preference goes to pool first. While it makes you a bit more vulnerable to pylon/bunker block crap, you're also in a far less dangerous position if it happens. You'll have lings in time to make a not-too-delayed hatch, whereas if you went hatch first and got blocked, you have to put a delayed pool down while using ~4 drones to kill whatever he put at your nat.
What I'm trying to say is that pretty much all the BOs in the OP are viable. In order to find the "most optimal defendable economical build" you need to add more variables, such as what you're trying to defend against. Ex: you'll need early spine at your nat if you want to hold a protoss 4gates but you'll need ASAP gas if you want to have any chance against a double hellion harass.
Also cannot assume that most optimal minerals income will get you the most optimal mineral/gas income. Extractor timing can have a pretty big impact here because you're toying with supplies as well as resources.
|
On December 06 2010 10:57 jacobman wrote:Show nested quote +On December 06 2010 10:51 GoldenH wrote:
What I've been doing is only getting the economy of my build down. Workers, supply, and Town Halls. Then I have a chart of money/time and I decide how to best spend this. You can then theorycraft forever what to do with whatever money isn't your economy. Then when I've theorycrafted something good I take it into the game and try it out and see what tactics will be good with it. I find this is much faster than trying to refine every build in game. I only have to play through it once per expansion timing, and a few times to get used to the timings. haha, my friends were actually just playing age of empires II today Are you trying to say that you theorycraft in order to figure out the defendability of a build? I was a little unsure exactly what you were saying.
No I just use theorycrafting to perfect the order I should get stuff in. It turns it into an optimization problem which is easy for any math student, and then I play the build on the ladder to see if I have enough stuff or too much stuff. I can just say "I want X units for a push at Y food, what is the best way of getting that?" and see.
For instance lets say you're playing a game and you are doing fine with a build but at some point you notice you are getting too much money. With my technique you can mathematically find the best time to build a new production facility, instead of just guessing.
Then you play it on the ladder to see if you need more units sooner or later, its good.
|
Good news! After doing some research on sc2mapster, I found the set of commands that will let me set up the AI to execute perfectly at every point, including exact worker gather micro and transfer logic. It'll be a little bit of work to get things set up to execute various build orders, but now that I know where to look, it's pretty easy to get the AI to follow exact directions.
|
Can you link me to it?
I think a patch broke it since then, but maybe there is a new way?
|
On December 07 2010 04:56 Skrag wrote:
Good news! After doing some research on sc2mapster, I found the set of commands that will let me set up the AI to execute perfectly at every point, including exact worker gather micro and transfer logic. It'll be a little bit of work to get things set up to execute various build orders, but now that I know where to look, it's pretty easy to get the AI to follow exact directions.
I found those commands too. I was looking around. It would be ten times easier if we could just alter the AI scripts. The ones that the game uses are so simple. Using triggers to me appears to be much more complicated. Unfortunately I couldn't figure out how to get the game to follow an altered AI script.
What were the commands you found? They're probably the same one's I was thinking of. They might not be though, so I might as well ask.
|
On December 07 2010 04:56 Skrag wrote:
Good news! After doing some research on sc2mapster, I found the set of commands that will let me set up the AI to execute perfectly at every point, including exact worker gather micro and transfer logic. It'll be a little bit of work to get things set up to execute various build orders, but now that I know where to look, it's pretty easy to get the AI to follow exact directions.
Do you know how to set up a drone transfer and a rally point change in the galaxy scripts? I'm trying to set the computer up to run the builds in a game, but currently I don't know how to do those two parts.
|
On December 05 2010 19:06 jacobman wrote:That thread essentially stopped updating a while ago. All focus moved to the 11 pool. I've found builds better economically than the leading build in that post. Also, I've come to the conclusion that the method used to look at the economy, replays, is too unreliable. It's just too hard to get any replay to be perfect. Like I said in the post, imperceptible differences in the play of the build led to totally different numbers, making it near impossible to figure out how good a build actually is. Also, I want to try testing builds in actual games in addition to this to figure out which builds are actually viable. I'm hoping some people can figure out ways to pull of some builds that appear really risky at first.
it is not impossible to get any replay perfect you just set your game speed to as low as you need it and when it comes out in the replay it looks like you were playing normal speed
|
On December 07 2010 20:24 SlapMySalami wrote:Show nested quote +On December 05 2010 19:06 jacobman wrote:That thread essentially stopped updating a while ago. All focus moved to the 11 pool. I've found builds better economically than the leading build in that post. Also, I've come to the conclusion that the method used to look at the economy, replays, is too unreliable. It's just too hard to get any replay to be perfect. Like I said in the post, imperceptible differences in the play of the build led to totally different numbers, making it near impossible to figure out how good a build actually is. Also, I want to try testing builds in actual games in addition to this to figure out which builds are actually viable. I'm hoping some people can figure out ways to pull of some builds that appear really risky at first. it is not impossible to get any replay perfect you just set your game speed to as low as you need it and when it comes out in the replay it looks like you were playing normal speed
The results still vary by a quite a bit. I've tried replays a lot myself and seen other peoples attempts.
|
On December 07 2010 20:14 jacobman wrote:Show nested quote +On December 07 2010 04:56 Skrag wrote:
Good news! After doing some research on sc2mapster, I found the set of commands that will let me set up the AI to execute perfectly at every point, including exact worker gather micro and transfer logic. It'll be a little bit of work to get things set up to execute various build orders, but now that I know where to look, it's pretty easy to get the AI to follow exact directions. Do you know how to set up a drone transfer and a rally point change in the galaxy scripts? I'm trying to set the computer up to run the builds in a game, but currently I don't know how to do those two parts.
Rally points can be changed by issuing an order to the unit/building you want to change the rally for.
Transfering could be achieved by giving the drones you want to transfer a harvest order targeting the mineral patches at the expansion. (That was the key order I couldn't figure out how to do before btw, sending a specific drone to a specific mineral patch)
|
On December 08 2010 02:08 Skrag wrote:Show nested quote +On December 07 2010 20:14 jacobman wrote:On December 07 2010 04:56 Skrag wrote:
Good news! After doing some research on sc2mapster, I found the set of commands that will let me set up the AI to execute perfectly at every point, including exact worker gather micro and transfer logic. It'll be a little bit of work to get things set up to execute various build orders, but now that I know where to look, it's pretty easy to get the AI to follow exact directions. Do you know how to set up a drone transfer and a rally point change in the galaxy scripts? I'm trying to set the computer up to run the builds in a game, but currently I don't know how to do those two parts. Rally points can be changed by issuing an order to the unit/building you want to change the rally for. Transfering could be achieved by giving the drones you want to transfer a harvest order targeting the mineral patches at the expansion. (That was the key order I couldn't figure out how to do before btw, sending a specific drone to a specific mineral patch)
Do you know what the syntax is for that?
It's really hard to find the names for abilities since there isn't complete documentation yet. Here are my guesses so far.
Mineral Patch: either C_MineralField or C_CU_MineralField
Gather Minerals: C_AB_DroneHarvest
Move: C_AB_Move
Can you confirm any of those so I might know which terms aren't causing syntax errors?
|
|
|
|
|
|
EPT
