|
I'm curious to see how much impact the maps and positioning of SCVs affects this as well. Most maps have different mineral lay out and angle to the CC and have very minor differences in mining speed. Also focusing your SCVs on closest mineral has an impact on the numbers. All of those are quite small, but when you are comparing such tiny differences already, it seems like they can skew the results somewhat.
If you could be more specific on your testing method and process, it would be interesting.
|
On December 31 2011 14:25 Sadistx wrote:Show nested quote +On December 31 2011 14:05 darthcaesar wrote:On December 31 2011 13:38 nttea wrote:how did you test this? might need confirmation cause of possible variances still awesome. I touched SCVs only once - when I did my 3/3 split. Then I just kept the CC rallied on the same spot. I rallied my 9th scv to make a depot, then a rax, then another depot, then back to mining. Other than that, I just made SCVs and let them do whatever - no mineral stacking, etc. Unfortunately, most builds do not work like that until 7 minutes.
The most important thing is for the two tests to be consistent to one another. Anything you do to one you do do the other and so on. That's the philosophy I used. As long as the two tests are consistent with one another, then you can still tell which one is better relative to the other.
|
I tried it out a little bit. Another thing that comes into the equation is the actual build you are doing. While this might open up an option for newer and different builds, it also delays tech builds like banshee, which a 10-20 second delay can be very devastating to, regardless of how many extra mineral you might get.
At least this is trying to do the same standard builds, just delaying the OC. There might even be better ways to do existing builds with this, just a question of which ones. For example Gas first cloak banshee has always been pretty standard 16 OC and works out very well as such (even before the 5 extra seconds to rax build time).
|
On December 31 2011 15:04 MrCash wrote: I tried it out a little bit. Another thing that comes into the equation is the actual build you are doing. While this might open up an option for newer and different builds, it also delays tech builds like banshee, which a 10-20 second delay can be very devastating to, regardless of how many extra mineral you might get.
At least this is trying to do the same standard builds, just delaying the OC. There might even be better ways to do existing builds with this, just a question of which ones. For example Gas first cloak banshee has always been pretty standard 16 OC and works out very well as such (even before the 5 extra seconds to rax build time).
well any gas first build you're gonna get a 13 rax and 16 OC. actually the rax change time smoothed this out a bit. As for the OP, the results are intuitive, like he said its quicker Mule vs greater income. I may keep doing 15 OC in a high pressure bunkering 2rax.. those 30 mineral returns make all the difference and getting each one of them 22 seconds sooner is huge.
|
On December 31 2011 15:08 gogogadgetflow wrote:Show nested quote +On December 31 2011 15:04 MrCash wrote: I tried it out a little bit. Another thing that comes into the equation is the actual build you are doing. While this might open up an option for newer and different builds, it also delays tech builds like banshee, which a 10-20 second delay can be very devastating to, regardless of how many extra mineral you might get.
At least this is trying to do the same standard builds, just delaying the OC. There might even be better ways to do existing builds with this, just a question of which ones. For example Gas first cloak banshee has always been pretty standard 16 OC and works out very well as such (even before the 5 extra seconds to rax build time). well any gas first build you're gonna get a 13 rax and 16 OC. actually the rax change time smoothed this out a bit. As for the OP, the results are intuitive, like he said its quicker Mule vs greater income. I may keep doing 15 OC in a high pressure bunkering 2rax.. those 30 mineral returns make all the difference and getting each one of them 22 seconds sooner is huge.
In that case, this would be my explanation to the OP as to why it never became standardized to either 15 or 16 OC, which is, the slightly better income did not justify delaying the tech or expo for some builds.
|
On December 31 2011 15:13 MrCash wrote:Show nested quote +On December 31 2011 15:08 gogogadgetflow wrote:On December 31 2011 15:04 MrCash wrote: I tried it out a little bit. Another thing that comes into the equation is the actual build you are doing. While this might open up an option for newer and different builds, it also delays tech builds like banshee, which a 10-20 second delay can be very devastating to, regardless of how many extra mineral you might get.
At least this is trying to do the same standard builds, just delaying the OC. There might even be better ways to do existing builds with this, just a question of which ones. For example Gas first cloak banshee has always been pretty standard 16 OC and works out very well as such (even before the 5 extra seconds to rax build time). well any gas first build you're gonna get a 13 rax and 16 OC. actually the rax change time smoothed this out a bit. As for the OP, the results are intuitive, like he said its quicker Mule vs greater income. I may keep doing 15 OC in a high pressure bunkering 2rax.. those 30 mineral returns make all the difference and getting each one of them 22 seconds sooner is huge. In that case, this would be my explanation to the OP as to why it never became standardized to either 15 or 16 OC, which is, the slightly better income did not justify delaying the tech or expo for some builds.
I've tried to see if players do the build for a certain reason, but I've only seen players sticking to the same timing in all of their builds. I don't think anyone actually plans to use it one way or another yet. It seems that players just prefer one style over the other.
|
15 for cheese 16 for standard game. Got it.
|
|
bout time someone explored this. Thanks!
|
what explains the sharp drop around 280? is that when the 16OC mule dies but the 15OC mule is still going?
|
Can you include a second graph where you normalize the data in difference against the total number of minerals mined at that time? Basically take the mineral difference and divide it by the total number of minerals collected by whichever method has collected the most minerals in total at that time.
This will give a graph which is more representative of the game impact of the two different methods at each time, as obviously 20 extra minerals at 1 minute is worth more than 50 extra minerals at 5 minutes.
If you want to be more accurate, you can normalize it instead against the number of "free" minerals at that time, taking into consideration only the mandatory buildings each build must produce (orbital, depot, rax). Technically there is a difference between different builds but that's probably too minor to consider.
|
whoa, that's cool. I need to keep this in mind then.
|
Seeker
Where dat snitch at?36888 Posts
16 OC it is. Awesome blog!!!
Why did they extend Barracks timing by 5 seconds though? Seems like an unnecessary change....
|
On December 31 2011 16:45 Seeker wrote: 16 OC it is. Awesome blog!!!
Why did they extend Barracks timing by 5 seconds though? Seems like an unnecessary change.... They change bunker rushes somehow in every patch because they still think it's too strong.
|
On December 31 2011 17:11 xmShake wrote:Show nested quote +On December 31 2011 16:45 Seeker wrote: 16 OC it is. Awesome blog!!!
Why did they extend Barracks timing by 5 seconds though? Seems like an unnecessary change.... They change bunker rushes somehow in every patch because they still think it's too strong.
No one would notice the difference if we changed back, well except the terran players who'd have perfectly lined rax and cc timings :p
|
*Puts on statistician hat* What methods did you use? Did you get a bot to replicate the results over and over? Did you do the build yourself over and over? Did you do it with and without optimal mining? I see the graphs, but I don't know what information produced the graphs, you didn't say how you got it or why I should trust it to hold up when I do it.
There is more I could say, but I'm not convinced that this debate is well settled based on these graphs with the accompanying explanation. There should be more information describing why your conclusion is valid based on your data gathering methods because I'm not convinced it's thoroughly analyzed. Especially not to the point to say unilaterally that 16OC is the best and should be the standard.
<3 Mip (MS Statistics)
|
16 OC might be slightly better in the long run but you're not looking at the complete picture. The 15 OC is probably better for defending rushes or putting on early pressure as you can get those first marines and hellions out a little faster. Also you might be able to get your FE up sooner with 15 OC, so its not clear cut.
|
Nice, but the change won't happen overnight.
Only in the most recent season of GSL has Nestea/Losira/Korean Zergs started to go 14 hatch instead of 15 and 16 hatch, and Nestea was extractor tricking every single time even well into this year (Losira learned 9 overlord was better earlier than he did, it seems, but he used to extractor trick too).
So I suppose that means you are saying that you know better than pro players?!? Post reported!
|
Thanks, I will do this.
But is the main reason for the long-term investment because of the timed-life of the mule? I think a small decision like this should only really be considered for the first 10~ minutes of the game... 15OC is better, right?
|
Nicely done! Just a thought though on supply timings for various builds. I have to test this but say you go 2rax with reactor after the first marine. Here the second supply depot is built after the second barracks and you just barely don't get supply blocket at 19. With an extra SCV I think you do, but again, this needs testing.
|
|
|
|