|
I have received requests on how to try the model out: Search "Double Harvesting (TeamLiquid)" by ZeromuS as an Extension Mod in HotS Custom Games to try it out. Email your replays of your games on DH to: LegacyEconomyTest@gmail.com might have partnership with a replay website soon as well In Game Group: Double Harvest |
On April 12 2015 08:13 y0su wrote:Show nested quote +On April 12 2015 07:04 ZeromuS wrote:On April 12 2015 06:51 y0su wrote: The double harvest method just seems... "unnatural"
Just curious, what kind of impact would it have if all expansions (after natural) were gold? (6 patch). Worker pairing would mean you'd get optimal harvesting at 12 instead of 16 so your 4 base model would be 51 on minerals (24 on gas). 16+16+12+7 would be better? (perhaps the gold mineral return per trip and overall resources per gold patch could be modified as needed)
edit:maths fail You want to try and have income match the existing sc2 income as much as possible as the number of workers stabilise in both income models. 6 gold income is probably too much on too few workers, and its going to throw mineral:gas ratios out of whack. The other issue is it doesn't fundamentally attack the 2:1 worker ration in SC2. Sure you can have more bases but you effectively have the same issue. There is some optimal count of mineral patches to probes in a 2:1 ratio. So maybe instead of 3 bases, you cap out at 4. In the end you want to remove the cap as much as possible, so that players cannot just sit on some "optimal" base count. This is the same reason the 6 patch 1 gas FRB approach from before won't work. You cap out at 24 mineral patches if you want less than 80+ workers. If you plan to make between 65 and 75 workers there is ZERO benefit to ever having more than 24 mineral patches available to you at any given time in game. Ever. If you remove pairing, then the theoretical cap is 48 mineral patches. The theoretical cap for mineral income goes from 3 bases to 6. When the "im gonna sit and turtle up and have the same income as you" number of bases is 6 instead of 3, then pure turtling gets very very hard. But isn't the end result is still an "optimal base count" (whatever system you use - be it 3 base, 4 base or 10 base). Again, I reiterate that doing something like all gold after nat would probably require a tweak to the gold mining amount (and patch value) to mimic the current 16 worker count. I do see where this could cause timing issues where you're able to essentially fully saturate a 3rd quicker... Also, since both the double mining and double harvest methods increase overall mining (due to fewer worker trips) couldn't the mining time get a slight increase to keep income relatively unchanged? I was thinking the exact same thing. It would result in a graph more similar to the orignial, with the same outcome.
|
|
This is one of the best posts I've ever seen on TL. Nice job ZeromuS.
|
Read through the whole thing, really hope Blizzard take a look at this, great job! :D
|
Canada13372 Posts
On April 12 2015 08:13 y0su wrote:Show nested quote +On April 12 2015 07:04 ZeromuS wrote:On April 12 2015 06:51 y0su wrote: The double harvest method just seems... "unnatural"
Just curious, what kind of impact would it have if all expansions (after natural) were gold? (6 patch). Worker pairing would mean you'd get optimal harvesting at 12 instead of 16 so your 4 base model would be 51 on minerals (24 on gas). 16+16+12+7 would be better? (perhaps the gold mineral return per trip and overall resources per gold patch could be modified as needed)
edit:maths fail You want to try and have income match the existing sc2 income as much as possible as the number of workers stabilise in both income models. 6 gold income is probably too much on too few workers, and its going to throw mineral:gas ratios out of whack. The other issue is it doesn't fundamentally attack the 2:1 worker ration in SC2. Sure you can have more bases but you effectively have the same issue. There is some optimal count of mineral patches to probes in a 2:1 ratio. So maybe instead of 3 bases, you cap out at 4. In the end you want to remove the cap as much as possible, so that players cannot just sit on some "optimal" base count. This is the same reason the 6 patch 1 gas FRB approach from before won't work. You cap out at 24 mineral patches if you want less than 80+ workers. If you plan to make between 65 and 75 workers there is ZERO benefit to ever having more than 24 mineral patches available to you at any given time in game. Ever. If you remove pairing, then the theoretical cap is 48 mineral patches. The theoretical cap for mineral income goes from 3 bases to 6. When the "im gonna sit and turtle up and have the same income as you" number of bases is 6 instead of 3, then pure turtling gets very very hard. But isn't the end result is still an "optimal base count" (whatever system you use - be it 3 base, 4 base or 10 base). Again, I reiterate that doing something like all gold after nat would probably require a tweak to the gold mining amount (and patch value) to mimic the current 16 worker count. I do see where this could cause timing issues where you're able to essentially fully saturate a 3rd quicker... Also, since both the double mining and double harvest methods increase overall mining (due to fewer worker trips) couldn't the mining time get a slight increase to keep income relatively unchanged?
There is no way to avoid an "optimal" base count. When workers mine most efficiently alone you can only ever have a 1:1 ratio, its impossible to be smaller. So in a system where resources exist to be obtained the most efficient way to do that will always be a "cap".
The goal isn't to completely remove the theoretical cap since this is legitimately impossible. The goal is to make the cap difficult to obtain. The harder it is to reach the cap the more time (<3 theDWF) there is for players to interact on the map and for their strategies and approaches in the game to interact.
If a player wants 6 bases as quickly as possible. They are free to achieve it. But 6 bases take a long time to get in terms of building times alone on Nexus/CC/Hatch. In addition it takes a lot of map control, which is not free, and it takes a lot of units to obtain that map control which takes a lot of micro.
And this is only from the "expanding player" perspective. The not expanding player has the power to harass, slow down, attack, TECH with their gas income, convert that to a timing, upgrades etc.
Its fine to have a six base cap. So long as it takes a lot more time, and a lot more effort to get there than in the current economic model
As to harvest time:
We did increase the double mining by more than just 2 actually. We increased it by about 2.15ish or something. This still has the worker pull issue because it takes so long to interrupt the cycle.
The nice outcome of the higher income is twofold: players get more money early for faster build order development (instead of 12 worker money), and the overall higher income speeds up the slow "I have my natural I'm gonna wait for X and Y to happen before I leave my base". The overall higher income ALSO speeds up the need for another base just a little bit. Much less drastic than in LotV but I think just enough (coupled with 1400 mineral patches it gets quicker I *think* we didnt test 1400 patches in long games just 1500) to make the game move with a touch more pace.
Maybe not LotV pace, but definitely a bit faster than HotS. And not because we force players to expand quickly, but because we give players the tools to get those expansions. A few more minerals could be a nexus alongside a 4 gate zealot warp in timing. The minerals could be an extra round of zerglings to pressure the map or defend the third hatchery as it gets put down. The extra minerals could be a sooner third hatch after taking gas (instead of 3 hatch before gas). The extra minerals could be another barracks and more marines for a strong 2 base timing (at the detriment of a third for example).
The extra minerals unlock a lot of mineral based strategies while keeping the gas based ones almost the same. The relative timings are all off in LotV atm because 12 workers provide much more money and at a higher worker count, making tech based timings shift. The 12 workers are an extra larva round for Zerg, and this alone pushes any Protoss WG timing to be effectively 6 larva behind (relative to HotS). The discussion of the impact of a 12 workers start, as i mentioned in the OP is for another article we have started on.
But in the end tech related timings all get pushed into an odd space when you have 12 workers, when your minerals mine out so quick, and when you are forced to spend minerals expanding. You end up with mineral heavy armies from some races and skipping gas spending, and in other cases gas only armies with minerals feeding expansions.
I think this has a bigger impact on protoss than people think. A forge and 2 cannons is almost an entire Half patch. A forge and 2 cannons are 1/8th of your available income on one base. Ravagers kill buildings, making the investment into cannons less valuable compared to HotS etc, it all snowballs.
But I think ive just started rambling
|
I'm not very knowledgeable about the behavior of economies in SC2, so thank you for breaking it down in this article. My gut said that the LotV economy changes were positive but something was still off, and now I think I understand why much better.
Thanks again ZeromuS, very well done.
|
This is a fantastic article and I also hope everyone takes the time to read all of it.
I would like to further some of the thoughts mentioned on the effect of altering mineral income as compared to gas. Changes to the existing mineral mining alters the relative strengths of mineral heavy unit compositions when compared to gas heavy compositions, at which point the strengths costs and times associated with every unit in the game might then have to be re-examined.
Taking the blizzard quote: "The main goal here was to make a change that would keep the feel of resourcing rates similar to Heart of the Swarm ..."
I feel that another intention behind this is that by keeping close to the current system the majority of the units in the game can remain unchanged. The more the economic model is altered the more blizzard is essentially re-creating an entirely new game instead of an expansion where there is some kind of baseline for the relative strengths of units to bring everything else in line with.
It seems to make the most sense for legacy of the void to be released some time between Blizzcon and the start of Wcs in 2016. Is there enough time to test new economic models as well as all of the new and potentially adjusted units and bring the game up to a standard that everyone expects?
Just some thoughts, I may be wrong as I have put no work or research into it.
One more note. What about the possibility of further upgrades or abilities that affect resource rates, these could be universal and permanent or only apply within a certain radius of the building that built them or last for a limited amount of time. Just as a player can invest in tech to alter the strengths of their army, they could further invest in tech to alter the strengths of their economy, increasing the options that are available. This of course would have a clear visible effect so an opponent could tell when someone is planning to gain economic advantages by upgrading. (This could also have the potential to change the 'optimal' base count the further a game progressed)
|
Quick question:
If Blizzard were open to simply changing worker AI to remove pairing would you prefer that to the Double Harvest method?
|
Canada13372 Posts
On April 12 2015 08:33 KrazyTrumpet wrote: Quick question:
If Blizzard were open to simply changing worker AI to remove pairing would you prefer that to the Double Harvest method?
I think we would need to test it. there is no way to know how that would work, but if optimal is 42 min/s and it drops at 9 after that it might be worth TRYING.
It would slow the game down in the early game though, which is the opposite of their current design goals. Due to no worker pairing at that income rate, you wouldn't be able to 13 gate after a 9 pylon for example with always making workers. If anything you need to increase early mining a little bit - even in BW you spent far less time just making workers before your build began to develop.
I think that is one of the goals as well - make builds develop sooner. And you can only do this by increasing early game income. Be it through worker starts or income mining. I believe, once more, blizzard took the wrong approach. The worker income should be higher early game on lower worker counts since 12 worker counts has a lot of unintended consequences.
As i just said in an earlier post, it changes the dynamic of gas based tech strats vs early mineral strats. Relative to HotS protoss is also behind in LotV beginning at the 1st second of the game. Protoss is normally due to chronoboost ever so slightly in a better position probe wise early, and relative to when buildings begin for zerg in particular, I *think* that WG and other researches like blink are a touch behind (compared to the same relative position in hots).
We will be testing relative timings between the races and tech timings after a 12 worker start in our next big article
|
Hoping Blizzard takes a lot from this!
|
On April 12 2015 08:39 ZeromuS wrote:Show nested quote +On April 12 2015 08:33 KrazyTrumpet wrote: Quick question:
If Blizzard were open to simply changing worker AI to remove pairing would you prefer that to the Double Harvest method? I think we would need to test it. there is no way to know how that would work, but if optimal is 42 min/s and it drops at 9 after that it might be worth TRYING. It would slow the game down in the early game though, which is the opposite of their current design goals. Due to no worker pairing at that income rate, you wouldn't be able to 13 gate after a 9 pylon for example with always making workers. If anything you need to increase early mining a little bit - even in BW you spent far less time just making workers before your build began to develop. I think that is one of the goals as well - make builds develop sooner. And you can only do this by increasing early game income. Be it through worker starts or income mining. I believe, once more, blizzard took the wrong approach. The worker income should be higher early game on lower worker counts since 12 worker counts has a lot of unintended consequences. As i just said in an earlier post, it changes the dynamic of gas based tech strats vs early mineral strats. Relative to HotS protoss is also behind in LotV beginning at the 1st second of the game. Protoss is normally due to chronoboost ever so slightly in a better position probe wise early, and relative to when buildings begin for zerg in particular, I *think* that WG and other researches like blink are a touch behind (compared to the same relative position in hots). We will be testing relative timings between the races and tech timings after a 12 worker start in our next big article Ok, thank you for answering.
|
On April 12 2015 08:41 Espers wrote: Hoping Blizzard takes a lot from this! Blizz has disappointed in their ability to learn from the community. However they did promise to change that, and to try and communicate with the community more, I see this article as a test. SetGuitarsToKill has already linked the article to @Starcraft and Psione on Twitter, the almost certain reality is that we get no response. But perhaps they will announce they will look into the matter, we will see...
|
Extremely well presented. Great job.
|
Typo in the original post.
|
I think I know an even better economy system.
Sorry for my bad english. English is not my native language.
I probably have a better solution for the resource system. But first I explain where I think Blizzard see the problems and why many solutions are out of the question for Blizzard.
Problems that are probably seen by Blizzard
Positive effect that is probably seen by Blizzard I think Blizzard want to keep the possibility to switch workers from a lost base to a base with just about 16 workers in order to partially increase the mineral income and don't lose a full base income. They don't want that 16 or more workers are completely useless after losing an expansion. Therefore 8 additional workers per base can gather with an lower efficiency.
Arising problems The problem what arises from this fact is that you can employ too many workers per base with a too big efficiency. In the current version you can employ 30 workers per base. Since you also need an army, you just need 3 bases, which mean at most 90 workers. You need at most 90 workers. In most cases, even less. The point is you need at most 3 bases. Most of you know the resulting problems, so I will not describe them now.
The reason why Blizzard don't want to remove mineral patches in order to adjust the resulting problems is probably because of the positive effect that I have described above first. They want to keep this positive effect. That is the reason why they don't reduce the required number of workers per mineral patch. In addition they don't want to remove some mineral patches, because they don't want to decrease the effect size of this positive effect.
Probable solution The solution consist of 2 steps.
First step Optimize the mineral patch occupation time so that exact 2 workers can work with a 100 % efficiency and a third worker would not or almost not increase the income. All this under optimal circumstances. For example, the shortest distance between HQ and mineral patch.
Second step 2 or even 3 workers can simultaneously gather on one mineral patch, but if for example 2 workers simultaneously gather on one mineral patch then the occupation time is increased by x %. For example 25 % for 2 workers and 60 % for 3 workers. The numbers can be adjusted. It's about the rough idea. If you want you can leave gather only 2 workers at the same time.
Result: With such a system you can adjust the resource problem with more flexibility and without to remove the described positive effect. In addition you can increase this positive effect whilst 3 or as many workers as you like can simultaneously gather on one mineral patch and you can set the efficiency for each worker number, which simultaneously gather on one mineral patch. The portion of the travel time to the mineral patches and back from the mineral patches is very minor. Therefore even by mineral patches with a greater distance it is not enough time to employ an additional worker with a significant efficiently.
|
Canada13372 Posts
On April 12 2015 08:51 Odowan Paleolithic wrote:Typo in the original post.
Good catch thanks!
|
ZeromuS is a god amongst mere mortals. This is easily the best post I've read on TL.
|
Read the entire thing, nice job!
I hope that Blizzard gives this a fair chance and comprehends what is being said. As you mentioned, it's a long beta. So 2-3 weeks on "Double Harvest" might go a long way to figuring out a better economy system. But that's just my opinion.
Again, great job on this, truly a cool read :D
|
I didn't read the entire post, but I was glad i read 'expanding should be a strategic choice, not a requirement'. I think Falling (?) has elaborated on that principle before, and was disappointed that blizzard hadn't addressed the it yet, as far as i know.
|
I swear to God; If Blizzard won't test this in the beta, I won't even consider buying LOTV.
|
|
|
|