EPTIn Response to David Kim re: SC2 Economy - Page 10
| Forum Index > Legacy of the Void |
In response to: http://us.battle.net/sc2/en/forum/topic/17085919227 | ||
|
rigginssc2
18 Posts
| ||
|
velvex
Germany226 Posts
On April 23 2015 03:55 rigginssc2 wrote: Reading through a bunch of plexia and zeromous posts I have to say, defining optimal for DH10 as anything other than 8 workers is a misuse of the word "optimal". The graphs all show efficiency per worker goes down after 8, so that means 8 workers is optimal. Similarly, the optimal number of workers in Hots/Lotv model is 16. Yet, with your use of the word "optimal", you're suggesting that people choose their worker count in order to maximise (= optimise) worker efficiency, which is not very realistic. People want to maximise their chances of winning – they may care for income per worker, but not in a way that they necessarily want it to be as high as possible. Essentially, what worker number is optimal depends on the strategy choice, the map, the opponent etc. It cannot simply be deduced from a table or a graph. | ||
|
BlackLilium
Poland426 Posts
On April 23 2015 03:29 rigginssc2 wrote: This means, in DH10 the 2 base player would have 16 workers over 2 bases. Thats an income of about 900mm (450x2). The 4 base player would have 32 workers over 4 bases. Thats an income of about 1800mm (450x4). Double the income. Of course, the same would be true for Hots, if you could have ever get to that many workers. The two base 32 and the four base 64. But the point is, you can't. At some point you have to cut rope and start making an army. You will still double the income on 4 bases with 32 workers, compared to 2 bases with 16 workers, regardless if Standard or DH 2x5. That is rather pointless comparison, since you simply have doubled everything (workers and bases). What matters, is when you actually "fix" the number of workers and try varying the number of bases. So, let's say you have 32 workers, either in 2 or 4 bases. In Standard that doesn't matter, you are at maximum anyway. In DH2x5 4 bases will mine faster by about 30%. There is no "double" anything. Now, that 30% is an incentive worth expanding, but it is not gamebraking. It is not a "must have". Raising and then defending those 4 bases when you have only 32 workers may be hard. On April 23 2015 03:29 rigginssc2 wrote: I think you misunderstood what David meant by "saturated" and then redefined it your own definition. Your definition made him, and the devs, really look as though they did not understand. In normal parlance, an SC2 base is considered "saturated" when you get to 16 mining workers -- not 24. Let's use that as our definition and evaluate. Why 16 is considered a saturated base? Is it because worker efficiency drops? Defining "saturation" as a point when workers start losing maximum efficiency is misleading. The whole point of DH is to make en efficiency drop earlier but by a smaller factor. Just because efficiency is a bit less than 100%, it does not mean that mining is already saturated. To put it at extreme: what if I made a system when every worker, starting from the first, drops the efficiency by 1%. Would you then call a one-worker base "saturated", beacuse second one mines only at 99% efficiency? As a compromise, we could define saturation when a worker efficiency drops below - say - 50%. But then, by this definition, DH 2x5 saturation point is still at 16 workers  For reference: In Standard 8-16 workers mine at almost 100% efficiency, and then workers 17-24 at only 30% efficiency. In DH 2x5 workers 8-16 mine at around 60% efficiency and then workers 17-24 at 30%. | ||
|
weikor
Austria580 Posts
That beeing said, Id love a longer beta, and some more drastic changes. Isnt that the point of a beta. Why not try a few weeks of this idea. More drastic changes could also be tried (removal of the MSC or removal of its 1x limit etc) and just see how the progamers figure out ways to deal with the changes. If theyre terrible, just revert them | ||
|
fancyClown
65 Posts
The only meaningful way to do that is to plot minerals earned as a function of time for both DH10 and LotV. This allows you to see when the first base gets mined out in both models. Anything else is just baseless conjecture. | ||
|
DoubleReed
United States4130 Posts
The Double Harvesting is about spreading out many bases all at once, so that there is essentially always a reason to expand. The reduced mineral nodes is about reduced effectiveness of bases over time. They just seem totally independent to me. | ||
Plexa
Aotearoa39261 Posts
On April 23 2015 04:47 fancyClown wrote: If you want Blizzard to abandon their current economy model for LotV, you need to clearly show what effect DH10 has on the available money vs. LotV. The only meaningful way to do that is to plot minerals earned as a function of time for both DH10 and LotV. This allows you to see when the first base gets mined out in both models. Anything else is just baseless conjecture. LotV and HotS have the same curves as in the OP. The difference is that half the patches run out sooner than the other half. That's not really something we can easily model since the timing on when those patches are depleted varies considerably depending on when the expansion was taken. Our preference is for a DH model with reduced minerals per patch (say 1350) as opposed to non-uniform mineral distributions. | ||
|
AkashSky
United States257 Posts
Of course this could be prevented by me playing aggressive early on, but still, the income scaling for zerg is exponential, whereas it is linear for Protoss. I just think that DK's statement isn't entirely inaccurate. | ||
EsportsJohn
United States4883 Posts
On April 23 2015 04:47 fancyClown wrote: If you want Blizzard to abandon their current economy model for LotV, you need to clearly show what effect DH10 has on the available money vs. LotV. The only meaningful way to do that is to plot minerals earned as a function of time for both DH10 and LotV. This allows you to see when the first base gets mined out in both models. Anything else is just baseless conjecture. It's the same time as HotS because the workers don't mine any faster; once the half patches are gone, the full patches still mine out at the same rate. The only difference is that your income drops by 50% per base every ~7:00 with 16 worker saturation. In Zeromus's original article, he posted this graph, which clearly already states this: http://www.teamliquid.net/staff/ZeromuS/Economy/relativemineouttime.png | ||
|
rigginssc2
18 Posts
On April 23 2015 04:12 velvex wrote: Yet, with your use of the word "optimal", you're suggesting that people choose their worker count in order to maximise (= optimise) worker efficiency, which is not very realistic. I think is pretty realistic. When you are playing currently you always want to saturate your base before expanding to another. You always want to make sure your bases are saturated. This is not unrealistic at all, this is just common knowledge and how you play the game. You over-saturate (go over 16) as you are expanding and then move workers to the new base. (saturate meaning 16 mining workers per base) On April 23 2015 04:13 BlackLilium wrote: You will still double the income on 4 bases with 32 workers, compared to 2 bases with 16 workers, regardless if Standard or DH 2x5. That is rather pointless comparison, since you simply have doubled everything (workers and bases). First, in Hots/Lotv you wouldn't need 4 bases if you had 32 workers. You would stop at 2. And that would not be double the income from 16-32 in standard it would be identical income. That is the whole point of the 2:1 argument in this article. The point is, in DH10 you actually can get 4 bases saturated (4x8=32), but in Hots/Lotv you cannot (4x16=64). The premise here is that if you were trying to power out economy or tech. A player in the Hots/Lotv version would not get the economy advantage that a player in DH10 would IF he tried to fast expand to 4 bases over the 2 base player. He simply could not get an efficiency advantage until he got over 48 workers (3x16). That's a lot and in Hots that would take a terran nearly 10 minutes to accomplish. On April 23 2015 04:13 BlackLilium wrote: Why 16 is considered a saturated base? Is it because worker efficiency drops? Defining "saturation" as a point when workers start losing maximum efficiency is misleading. I am not defining it to be anything. I am simply saying that the word "saturation" is already used to mean one thing -- 16 workers on a base -- so "defining it" to mean something else in this article/thread is confusing. I would imagine David simply used the word "saturation" in the normal SC2 way, which is understandable since his job/life is SC2, and it shouldn't be taken as his necessarily misunderstanding the original article. Personally, I think a few of the terms are a bit loose in this thread (like "optimal" as I mentioned), but the most important thing is just everyone being on the same page. | ||
|
EatThePath
United States3943 Posts
On April 23 2015 04:59 rigginssc2 wrote: I think is pretty realistic. When you are playing currently you always want to saturate your base before expanding to another. You always want to make sure your bases are saturated. This is not unrealistic at all, this is just common knowledge and how you play the game. You over-saturate (go over 16) as you are expanding and then move workers to the new base. (saturate meaning 16 mining workers per base) First, in Hots/Lotv you wouldn't need 4 bases if you had 32 workers. You would stop at 2. And that would not be double the income from 16-32 in standard it would be identical income. That is the whole point of the 2:1 argument in this article. The point is, in DH10 you actually can get 4 bases saturated (4x8=32), but in Hots/Lotv you cannot (4x16=64). The premise here is that if you were trying to power out economy or tech. A player in the Hots/Lotv version would not get the economy advantage that a player in DH10 would IF he tried to fast expand to 4 bases over the 2 base player. He simply could not get an efficiency advantage until he got over 48 workers (3x16). That's a lot and in Hots that would take a terran nearly 10 minutes to accomplish. I am not defining it to be anything. I am simply saying that the word "saturation" is already used to mean one thing -- 16 workers on a base -- so "defining it" to mean something else in this article/thread is confusing. I would imagine David simply used the word "saturation" in the normal SC2 way, which is understandable since his job/life is SC2, and it shouldn't be taken as his necessarily misunderstanding the original article. Personally, I think a few of the terms are a bit loose in this thread (like "optimal" as I mentioned), but the most important thing is just everyone being on the same page. Saturation is already a well-understood term meaning "adding workers to the patch achieves no income gain". AKA "The patch is saturated." Which you can extend to a base being saturated meaning all its patches are saturated. Saturation itself has nothing to do with efficiency but they often get conflated in discussions about mining and economy. | ||
|
EsportsJohn
United States4883 Posts
On April 23 2015 04:57 AkashSky wrote: Current hots economy favors protoss, because that race is slow to expand. However, David Kim's example of 2 base vs 4 base is actually kind of accurate in this model in the protoss vs zerg matchup. Zerg can expand to the 4 bases really quickly, and produce more workers than protoss. Masters zerg players I face have up to 60 workers by 8minutes, at this time, the most probes I have ever had is ~48. The 12 worker advantage at the 8 minute mark means that they could actually have about 1.7~my mineral income. Of course this could be prevented by me playing aggressive early on, but still, the income scaling for zerg is exponential, whereas it is linear for Protoss. I just think that DK's statement isn't entirely inaccurate. I'm not sure you're using the terms "exponential" and "linear" correctly here. I understand that you feel Zerg can burst economy a lot faster than Protoss, but that doesn't necessarily make it "exponential" growth. In a HotS economy, both are essentially on a linear curve, just that the Zerg curve extends out toward 4 bases while the Protoss income would plateau at 2. In the DH model, both are on an inverted curve in which the Zerg curve extends slightly higher than the Protoss curve, and would start plateauing sooner. This isn't taking into account gas, player interactions, or how maps function to spread out bases. So saying that Zerg will have 1.7x your mineral income ("nearly double") by 8:00 in a DH10 model is purely theoretical and perhaps faulty logic. We have to remember that the economy is not a "magic solution" to all of the problems in SC2, so trying to picture DH10 in current gameplay won't actually yield us proper results. Our goal is to take a step in the right direction by solving economy first, and then hopefully it will create a better environment for balancing units and costs. | ||
|
fancyClown
65 Posts
On April 23 2015 04:52 Plexa wrote: LotV and HotS have the same curves as in the OP. The difference is that half the patches run out sooner than the other half. That's not really something we can easily model since the timing on when those patches are depleted varies considerably depending on when the expansion was taken. Our preference is for a DH model with reduced minerals per patch (say 1350) as opposed to non-uniform mineral distributions. The curves of LotV and HotS are the same, but the big difference is the initial worker count. This makes it really hard to compare the 3 models in terms of actual effect on money earned over time. All you need to model is one base that produces workers until saturation. When saturated, worker production stops and no further expansions are taken. You are just modeling this one base and then ask: At which point is this base mined out? How does income increase over time? How does this compare between LotV and DH10? This is the only thing you are really interested in to get the point across. | ||
|
BlackLilium
Poland426 Posts
On April 23 2015 04:59 rigginssc2 wrote: First, in Hots/Lotv you wouldn't need 4 bases if you had 32 workers. You would stop at 2. And that would not be double the income from 16-32 in standard it would be identical income. That is the whole point of the 2:1 argument in this article. In standard you don't need 4 bases with 32 workers, but you can have them. You don't need 4 bases with DH 2x5 either, unless you absolutely want to maximalize worker efficiency. But maximizing worker efficiency is not always the best strategy, you know... You argumentation contains a sentence that "4 bases with 32 workers mine twice as fast as 2 bases with 16 workers" - which is true for both Standard and DH 2x5. It would double the income in any other sane resource system too, because you simply have twice as many things! As such it brings nothing to the discussion and I don't really see why David Kim would come up with something like this to justify his "double income". | ||
|
EsportsJohn
United States4883 Posts
On April 23 2015 05:21 fancyClown wrote: The curves of LotV and HotS are the same, but the big difference is the initial worker count. This makes it really hard to compare the 3 models in terms of actual effect on money earned over time. All you need to model is one base that produces workers until saturation. When saturated, worker production stops and no further expansions are taken. You are just modeling this one base and then ask: At which point is this base mined out? How does income increase over time? How does this compare between LotV and DH10? This is the only thing you are really interested in to get the point across. The difference is that you don't lose 50% of your income at 50% of HotS mineout time (~7:00) -_- At 16 workers, DH10 and HotS have almost the same efficiency and income. Literally the only difference in graphs would be that at 7:00, LotV economy would chunk off like a step function. | ||
|
Big J
Austria16289 Posts
On April 23 2015 05:10 SC2John wrote: I'm not sure you're using the terms "exponential" and "linear" correctly here. I understand that you feel Zerg can burst economy a lot faster than Protoss, but that doesn't necessarily make it "exponential" growth. In a HotS economy, both are essentially on a linear curve, just that the Zerg curve extends out toward 4 bases while the Protoss income would plateau at 2. In the DH model, both are on an inverted curve in which the Zerg curve extends slightly higher than the Protoss curve, and would start plateauing sooner. This isn't taking into account gas, player interactions, or how maps function to spread out bases. So saying that Zerg will have 1.7x your mineral income ("nearly double") by 8:00 in a DH10 model is purely theoretical and perhaps faulty logic. We have to remember that the economy is not a "magic solution" to all of the problems in SC2, so trying to picture DH10 in current gameplay won't actually yield us proper results. Our goal is to take a step in the right direction by solving economy first, and then hopefully it will create a better environment for balancing units and costs. The income growth of all races is "exponential" for as long as you expand fast enough. The linear/logarithmic part is true for saturating one base, but if you start saturating a second or even a third base at the same time, you get an exponential growth until certain saturation points are being reached. On top of that, the way the larva/inject mechanic works you also get an exponential worker growth until the income a new base gives you has caught up to the (very high) worker production. | ||
|
BlackLilium
Poland426 Posts
I think you are both right actually: - Your income growth is linear with the amount of bases and workers - Your income growth is exponential in time | ||
|
EsportsJohn
United States4883 Posts
On April 23 2015 05:30 BlackLilium wrote: The linear vs exponential economy... I think you are both right actually: - Your income growth is linear with the amount of bases and workers - Your income growth is exponential in time Fair enough, I'll take it. | ||
|
fancyClown
65 Posts
On April 23 2015 05:25 SC2John wrote: The difference is that you don't lose 50% of your income at 50% of HotS mineout time (~7:00) -_- At 16 workers, DH10 and HotS have almost the same efficiency and income. Literally the only difference in graphs would be that at 7:00, LotV economy would chunk off like a step function. We are talking about plotting minerals earned as a function of time, not as a function of worker count. Since DH10 and LotV start with different worker counts, the graphs would most certainly look different. | ||
|
Plexa
Aotearoa39261 Posts
On April 23 2015 05:33 fancyClown wrote: We are talking about plotting minerals earned as a function of time, not as a function of worker count. Since DH10 and LotV start with different worker counts, the graphs would most certainly look different. DH10 =/= 12 worker start. The number of workers you start with is irrelevant, and in LotV we'd expect initial tests with DH10 to be done with 12 workers. SC2john is right when he says income/time for a normal game in lotv would look identical to hots except when you mine out the 4 patches (~7:00) when the graph jumps down rapidly (step function behavior). EDIT: our comparison to HotS is for two reasons 1) People generally have greater access to HotS and we can put extension mods online there 2) The efficiency curves for HotS and LotV are identical assuming equal number of nodes available | ||
| ||
