|
On October 29 2012 06:38 Pimpmuckl wrote:NICE Read, holy cow that's some research you did there! Show nested quote +On October 29 2012 05:28 Pholon wrote:
Someone give this man a BA in Starcraft This certainly is a Bachelor of Science.
It has to be a Bachelor of Science. It's using a lot of maths. Lots of maths gets you a BS degree. Whereas little to no maths gets you a BA. Absolutely no maths might get you a BFA - but you can probably tell me the progression of chromatic chords, the proper minors, and color coordinate like no one else.
Still, that is certainly a stack of numbers worthy of a BS.
Also, its times like this I am okay with being lower league, so I don't have to hurt my head wondering which refinery is optimal.
|
I'm glad to see such a well made post but after reviewing it more close I've come up with some constructive criticisms:
1. In HotS the efficiency will be falsely labeled as 3/3 (100%) at the extractors etc. for the E, F and in some case D geysers.
2. Isn't gpm (gas per minute) instead of percentages a more useful measurement as it easier substitute gas to units and other costs? Numerically speaking you won't have to round off gpm as you have to percentages.
3. A sense of modesty is missing as geyser with 3 workers [E,0-45] only produces 2 less stalkers (104 gas) over 21 minutes and 31 seconds compared to a [G,90] geyser. Will those 2 missing potential stalkers make an impact?
4. With the addition of a 4th worker isn't the maximum mineral saturation or any other mineral benchmark reached later than the opponent. That mineral discrepancy, perhaps minuscule, is a trade off for maximum gas saturation.
|
24 hours anniversary post by OP yey! edit: +1 minute anniversary 
Current total viewers in 24 hours = 62,135 is even higher than the viewer count that my most popular thread before this has accumulated since originally posted 2 months ago. I kinda knew that this thread would attract more viewers than some other threads of mine, but I never imagined it woud gain this many this fast. Thank you all for reading and supporting.
Many people said that they read a similar research a long time ago, and butter kindly provided the link
+ Show Spoiler +On October 28 2012 22:18 butter wrote:Wow, I thought for sure mapmakers had fixed this issue in their maps by now since the analysis was published in 2010.
Now, I also heard that PsyStarcraft discussed this gas imbalance before. Does anyone have a link? I didn't follow him back then and I couldn't find the material by myself. I'd appreciate it if someone can find it. Or, any link to related research/content/video published before my work would help. Maybe we can learn more from wise predecessors of mine.
I feel a few people are getting it wrong about this thread. It's not about who found this gas imbalance first as I have never claimed to be the one, but rather about disclosing/explaining/quantifying this hidden feature of SC2 that only the pivileged knew about. It's nice if someone is one of those privileged who has known this fact for years, but that doesn't mean everyone should have known this feature that not many have openly discussed, that no caster today talks about, or that Blizzard has not commented on. "More material = good," not "more material = nothing is new because I saw this before."
On October 28 2012 22:50 MinimalistSC2 wrote:
Excellent OP. I actually read everything but the spoilers below the map pictures haha. good job Yeah, spending 20% of my effort for something only 1% of readers pay attention...
On October 28 2012 23:07 snively wrote:wowow so crazy thanks so much this is great! it must have sucked to do so much testing On October 29 2012 00:15 vanTuni wrote:The ... amount ... of .... work *mindblown* Pros better say thanks  Alt+tab and/or a 15min timer into real life chores does the trick 
On October 29 2012 08:22 GhandiEAGLE wrote:
I wonder if you even feel this much research was worth the end result? Rewarded by positive feedbacks even if it were not.
On October 29 2012 10:29 Gatesleeper wrote:+ Show Spoiler +
lol.
A couple of weeks ago, I wanted to see if adding a 4th miner to gas has any effect at all on gas income. So I made a custom game and went up to 5 bases, had 3 workers in each gas, then added a 4th worker to each gas. In the replay I watched the income tab to see if my gas income went up at all. It didn't, and I was disappointed.
Do you know which spawn location I was at on which map? 5 oclock on Ohana. According to this post, all 5 bases on the lower half of Ohana are optimal gas locations. -______-
lol. You couldn't have picked a worse location to test it on. That is insanely unlucky of you...
On October 29 2012 04:26 fezvez wrote:+ Show Spoiler +Show nested quote +On October 28 2012 17:24 fezvez wrote:+ Show Spoiler +As a side note on the "2 gas 4 workers" part : When (waiting time for the 3rd worker) < (additional travel time for farther gas), it should be the case. I did some modeling, and I agree with your conclusion (as long as (additional travel time for farther gas) means (additional travel time for farther gas back and from) ) This looks horrible to experiment though I thought no one would give much thought there. Just like you probably did, I made a mathematical model, but I gave up there because experiment looked horrible as you said. Very keen eye on the detail on your part. Thanks. OK, so I went through the other analysis from 2010. I noticed a number of differences, for example, lack of asymmetry, and slightly different values (the worst spot is at 89% mining rate, as opposed to 92.5% in your case) I don't believe that the game itself has changed (except perhaps for the "worker mineral trick" patch by Blizzard), so I'd be more inclined to think that the analysis in 2010 was not as thorough. Anyway, it still looks pretty serious (the discrepancy can be explained by the fact that he counted gas mining for only 300 seconds) The good thing is that he studied what happened when you put 2 workers in a gas (thus, deriving the gas mining of one worker at the same time), and he got the results that the mining rate was between 64 and 79% of the max mining rate (i.e : 1291). Analysing the fact that the worst gas gives you 92.5%, I tend to think that in fact, the mining rate for 2 workers is between 61.6% and 79%. Anyway, it's not going to change the conclusion, which holds for any spawning position for any map in the current ladder pool : The important conclusion is that it is always better to split your 4 workers so as to have 2 in each gas. Putting 3 in the closest one, and 1 in the farthest one is inefficient. Always.At best, by putting 3 in the 1291, and 1 in the worst possible gas, you obtain 130.8% mining rate of a single good gas. By putting 2 in each, you get 79+61.6=140.6%. Very nice to see you give a lot of thoughts into this. You are partially right and partially wrong in the quoted post. What's wrong is that being a 1291 gas doesn't guarantee having 79% with 2 workers, so the last equation "79+61.6=140.6%" is not necessarily true. Being a 1291 gas only guarantees that it has 66.6%+ with 2 workers. Therefore, "66.7+61.6=128.3%<130.8%" for 2 workers each is still a possibility, mathematically speaking based on spinesheath's 2010 research results. "Always" is too strong of a word at this point without enough accurate data and mathematical backup.
That said, you are right in that in the current ladder pool, 2 workers in each gas is highly likely to be better. At this point, I would conclude that about 95%+ of the time (% is just a wild guess), 2+2 is more efficient than 3+1 based on spinesheath's 2010 research. I would also conclude that even in rare cases where 3+1 is better, the gain is bound to be small due to relatively high efficiency even for far gases like E or F. In order to absolutely conclude this 2+2 vs 3+1 efficiency issue, more accurate data for 2 worker mining than 2010 data, which I haven't provided yet, is necessary as margin of error in 2010 data seems to overlap the border line. Come to think of it, "When you are not sure, 2 each is probably the safer bet anyways." in the OP sounds appropriate. Anyways fezvez, both you and spinesheath are awesome in shedding some light to 2 worker mining situation.
|
This is probably the sexiest post I have seen in a quite a while. I was watching Destiny coach someone on stream a long time ago and he mentioned to his student that there were certain geysers that he should be taking first because they were more efficient than others. Since then I had always tried to guess which would be the best on each map but thanks to you I can just load this post on my second monitor and slowly memorize which gasses are better to make sure I'm getting that sick gas income.
Thanks for all your hard work on this issue and for sharing it.
P.S. you have saved many of my greater spires from zealot drops, and probably helped me win a bunch of games, and I cannot thank you enough for this.
|
You, Sir, are the purest definition of a handsome nerd.
|
I think you should be working for blizzard, the way you analysed this is great.
|
Where is the "Like"-button?
|
On October 29 2012 14:47 archonOOid wrote:
I'm glad to see such a well made post but after reviewing it more close I've come up with some constructive criticisms:
1. In HotS the efficiency will be falsely labeled as 3/3 (100%) at the extractors etc. for the E, F and in some case D geysers.
2. Isn't gpm (gas per minute) instead of percentages a more useful measurement as it easier substitute gas to units and other costs? Numerically speaking you won't have to round off gpm as you have to percentages.
3. A sense of modesty is missing as geyser with 3 workers [E,0-45] only produces 2 less stalkers (104 gas) over 21 minutes and 31 seconds compared to a [G,90] geyser. Will those 2 missing potential stalkers make an impact?
4. With the addition of a 4th worker isn't the maximum mineral saturation or any other mineral benchmark reached later than the opponent. That mineral discrepancy, perhaps minuscule, is a trade off for maximum gas saturation.
1. The efficiency is rightly labeled. It is measured how efficiently you are mining the gas geyser by workers for that geyser. It is neither useful not even remotely desirable to know how that particular geyser compares to the best ones on the map AS WELL AS telling you how many workers are currently mining from it.
2. Rounding error can be accounted for. GPM is pretty uninteresting because that affects other topics, like how builds can spend gas, how many gases to take to spend for units, and the like. Saying 1396 versus 1498 is like "Okay, whatever" where saying one is 7% more than the other actually has meaning.
3. If you're reading this interested in an 8% change in gas, you're already knowing from the title that the change is very modest.
4. No, if I'm reading the question correctly. You would only consider adding a 4th worker to gas if mineral saturation was the furthest thing from your mind. You're gas strapped with plenty of minerals; the benefits of 4th on gas instead of the same guy on minerals outweighs it by an order of magnitude if not more.
Let me just say ... when my opponents point out that I have 4 guys in that gas geyser ... it'll be fun
|
Wow this is really cool. I actually think this adds something interesting to map design. I don't think all gasses needs to be efficient, as long as the same gasses are inefficient equally for all spawning locations. This makes some strategies vary depending on map when this is thought of.
|
Thanks for sharing this! I've been observing this discrepancy in the gas mining distance to CC with the naked eye, and i tried to always place the nearest rafinery first. Still, sometimes when i was involved in micro/worker harass i would forget and just place one extractor hastelly, not necesarelly in the optimal spot. There were times when i observed later the mistake, but still, in my mind this wasn't a top priority, as i underestimated the importance of this placement. Seeing the calculations you made opened my eyes a lot, and about the 4th worker, i would have never considered adding it... Great guide and i am sure this took you a lot of time to make!
|
It was pretty popular on metalopolis back when it was played, to take closer gas because 2nd one was needing 4th worker for full saturation. Later most kinda stop giving a damn :D
|
Excellent post. Taking more gas outside of main with E or F geysers in the main. Just another reason never to play entombed. Thanks OP.
|
Great read, well done putting it all together.. I generally just choose a random gas and that probably won't change but in professional games this could definitely make a difference!
|
|
|
OP, I have a question: How does the 3 worker gas production at a bad location compare to the 2 worker gas production at a good location? Further, I'm sure that the good locations will narrow considerably at 2 worker saturation as opposed to 3 worker.
|
Really scientific study I will never gas up a far vespene before a close one.
|
Nice effort, already knew it since the same imbalance was on Metalopolis! Doesn't really affect builds that much unless you are inefficient with them in the first place, then it becomes another hiderance to resource collection.
|
Another solid post Orek. MOAR DRAMA!
|
On October 29 2012 14:47 archonOOid wrote:
I'm glad to see such a well made post but after reviewing it more close I've come up with some constructive criticisms:
1. In HotS the efficiency will be falsely labeled as 3/3 (100%) at the extractors etc. for the E, F and in some case D geysers.
2. Isn't gpm (gas per minute) instead of percentages a more useful measurement as it easier substitute gas to units and other costs? Numerically speaking you won't have to round off gpm as you have to percentages.
3. A sense of modesty is missing as geyser with 3 workers [E,0-45] only produces 2 less stalkers (104 gas) over 21 minutes and 31 seconds compared to a [G,90] geyser. Will those 2 missing potential stalkers make an impact?
4. With the addition of a 4th worker isn't the maximum mineral saturation or any other mineral benchmark reached later than the opponent. That mineral discrepancy, perhaps minuscule, is a trade off for maximum gas saturation. 1. Give me a beta key
2. Great suggestion. Added two [gas/min] tables in the OP.
+ Show Spoiler +
As for round off, it happens no matter what due to significant numbers. % comparison is more useful for people to get the general idea, [gas/min] is more useful for those build engineers who want to dig deeper into the topic like yourself. Thank you for your suggestion.
3. Being able to warp-in a sentry or produce an observer a few secconds earlier is by no means a small thing at top level in an extreme example. I don't think this gas imbalance is the end all-be all, but every little bit counts not in my own scrubby games, but at least in pro games.
4. This is only my personal opinion, but 4th worker should be added only upon floating tons of minerals as I talked in the OP. When mineral is still tight, I don't think it's worth much. However, some people seem to have concluded that they should add 4th worker from the get-go. It's their choice. I merely provided the facts and numbers about 4th worker, so how to use them is up to individual players ultimately. Just because I personally think 4th worker isn't much worth early game, some wouldn't stop using it early based on what they read in this thread.
Overall, I appreciate your input upon reviewing the content closer than most other readers.
On October 29 2012 22:59 revy wrote:
OP, I have a question: How does the 3 worker gas production at a bad location compare to the 2 worker gas production at a good location? Further, I'm sure that the good locations will narrow considerably at 2 worker saturation as opposed to 3 worker. Assuming you read the Miscellaneous section, please also read my reply to fezvez in the 3rd post of this page and how we two discussed. In short, "2010 research by spinesheath exists, but more accurate data by me does not because I'm too lazy." It seems more people are interested in this 2 workers scenario than I originally thought. I will give some more time on this.
|
I have a Bsc(hons) degree and i put less effort in to that than you did in to this. Fair play!
|
|
|
|
|
|
EPT![[image loading]](http://i.imgur.com/C4O7x.png)
![[image loading]](http://i.imgur.com/GF6dW.png)
