EPTThank you everyone who put time on this and let's hope Blizzard takes heed.
A Treatise on the Economy of SCII - Page 11
| Forum Index > Legacy of the Void |
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 | ||
|
2v2levels
United States88 Posts
Thank you everyone who put time on this and let's hope Blizzard takes heed. | ||
|
Ovid
United Kingdom948 Posts
On April 12 2015 22:45 Teoita wrote: That also impacts early game scouting, as well as defending cheese though, so it's not quite the ideal solution, unless you can find a sweet spot where mining is changed to your desired level, but the interactions with every unit in the game (particularly early game ones like marines/lings/zealots/queens, and harassment ones like hellions/dts/oracles etc) is unchanged. That sounds extremely trick though. I think a change of workers and how they attack is in order anyway. I would prefer them to have a slight increase in range (I mean tiny) so that interactions take place like this Overall speed isn't what I'm really looking to change since as you said it means the unit could either be run down or outrun a unit it can't currently. (Even though movement speed is slightly changed depending on the direction of the unit) if you changed the acceleration/deceleration you could create a moving shot of sorts that if an opponent has just a moved latched onto a unit you could micro it so you can win everytime. (I think I will try and make this work in the editor now so I can show you what I mean) | ||
|
Grovbolle
Denmark3804 Posts
On April 12 2015 07:14 Teoita wrote: Nowhere in the article do we claim that 4base income is 3 times higher than 3base or whatever...the bars are in fact relative to 2142 and 2235 respectively, the y axis just so happens to start from 2100 instead of 0 because, well, what you want to look at isn't two bigass bars that are slightly different from each other, you want to see exactly how different they are. I just don't understand how that could possibly be misleading. Not to mention, altering the scale on an axis to illustrate a point better is in fact extremely common in scientific papers. Ever heard of a log/log plane? Just because it happens all the time doesn't mean it is not a bad practice. Also log scales are a completely different thing. If you ever read a book on data visualization by an actual expert (Stephen Few for instance) you would know that the power of a bar chart is the abillity to encode the data points for fast glance decoding. When I glance over these charts some of them makes it look like one is 3-4 times more effective than the other. The author even chose to add the actual data point values to CLEAR UP THE CONFUSION his poor usage of bar charts is sure to make. There is no reason for him to use a bar chart if he is not gonna use it in a way that is representative of the underlying data. | ||
ZeromuS
Canada13378 Posts
On April 12 2015 22:23 Ovid wrote: Either lalush extension mod is bugged or there's a big flaw with this model. I tried force pairing workers and what I found was sometimes (after the first mining packet) if you force a worker to mine that patch the worker will force the one that was halfway through mining off of the patch. I then tried to replicate that without having clicked anything and I noticed very rarely but it does happen sometimes that the workers automatically override the others. I assume this isn't intentional? No it is intentional. If you could force them to pair the AI would eventually pair them The way we force no pairing and bouncing is by abusing the harvest ai. Remember each harvest is an action, they do it twice but they only leave and return to the next/cc/hatch. So of course they will be interrupted if another one starts (or if you force it). The econ management is shifted instead of early age worker pairing you need to make sure the startimgv6 dont avoid a close patch. You also need to make sure the first few workers go to the empty mineral patches and not disrupt others (for first two). Next there is way more econ management in mid and late game. Let's say you have 16 in main and 9 in Nat then take natb asses. If you pit the workers in the Nat in the gas and not Maynard from the main you are losing a LOT of potential income. This interaction applies at all times of the game. You always need to be checking and managing worker counts. You need to decide : do I put 8 guys on the new base or defend them better on the old ones? Do I replace the harassment killed workers with new ones or maynarding ones? More decisions than rally nexus make probe. | ||
|
ZeromuS
Canada13378 Posts
On April 12 2015 23:39 Grovbolle wrote: Just because it happens all the time doesn't mean it is not a bad practice. Also log scales are a completely different thing. If you ever read a book on data visualization by an actual expert (Stephen Few for instance) you would know that the power of a bar chart is the abillity to encode the data points for fast glance decoding, when I glance over these charts some of them makes it look like one is 3-4 times more effective than the other - the author even chose to add the actual data point values to CLEAR UP THE CONFUSION his poor usage of bar charts is sure to make. There is no reason for him to use a bar chart if he is not gonna use it in a way that is representative of the underlying data. This isn't being published in some major scientific paper. The graphs are a visual aid. Can we stop derailing the discussion because of a graph? | ||
|
Grovbolle
Denmark3804 Posts
On April 12 2015 23:46 ZeromuS wrote: This isn't being published in some major scientific paper. The graphs are a visual aid. Can we stop derailing the discussion because of a graph? It's not a visual aid if it isn't actually AIDING the visual perception. Whatever, the article was good. | ||
|
Umpteen
United Kingdom1570 Posts
![[image loading]](http://i.imgur.com/FIuNPub.png) The lines represent, from lowest to highest: 1. HotS mining rates 2. One base mining with 2 gold patches up to 16 workers 3. Two bases mining with a total of 4 gold patches up to 16 workers 4. Two bases mining with a total of 5 gold patches up to 16 workers (2 patches in main, 3 in natural) Headline numbers: One base with 2 gold patches
Two bases with 5 gold patches (2 in main, 3 in natural)
Overall, this seems to tick a lot of the same boxes as Double Harvesting. There's an initial economy boost, spreading workers across more bases is advantageous, and the cap is, in practice, all but unattainable (24 gold patches would be 7-8 bases). Add to that the other, previously mentioned advantages:
What am I missing here that's stopping anyone else caring about this idea? | ||
|
hewo
Norway119 Posts
On April 12 2015 23:46 ZeromuS wrote: This isn't being published in some major scientific paper. [...] Correct, but why not aspire to be the equivalent to a major scientific paper in the sc2 community? We are pretty close, you know. | ||
|
Ovid
United Kingdom948 Posts
On April 12 2015 23:44 ZeromuS wrote: No it is intentional. If you could force them to pair the AI would eventually pair them The way we force no pairing and bouncing is by abusing the harvest ai. Remember each harvest is an action, they do it twice but they only leave and return to the next/cc/hatch. So of course they will be interrupted if another one starts (or if you force it). The econ management is shifted instead of early age worker pairing you need to make sure the startimgv6 dont avoid a close patch. You also need to make sure the first few workers go to the empty mineral patches and not disrupt others (for first two). Next there is way more econ management in mid and late game. Let's say you have 16 in main and 9 in Nat then take natb asses. If you pit the workers in the Nat in the gas and not Maynard from the main you are losing a LOT of potential income. This interaction applies at all times of the game. You always need to be checking and managing worker counts. You need to decide : do I put 8 guys on the new base or defend them better on the old ones? Do I replace the harassment killed workers with new ones or maynarding ones? More decisions than rally nexus make probe. I didn't explain very well. The current drone is mining (named 1 second drone is name 2) 1 is mining 2 I'm controlling I tell 2 to go mine on the close patch that 1 is mining from 2 should bounce but if I click at a particular point usually when 1 is is just about before the second mining scan 1 then bounces and allows 2 to mine. Shouldn't it be that 1 still is forced to bounce rather than the one currently mining. | ||
|
5p4z3n3k0
Netherlands19 Posts
| ||
|
Randomaccount#77123
United States5003 Posts
| ||
Teoita
Italy12246 Posts
Of course gold minerals is also another way to explore the "change worker efficiency" dynamic that could be explored though. @Barrin: i agree that less mineral per trip might be a good idea and worth exploring; 5 per trip simply has the side effect of increasing income (therefore speeding up the early game) slightly, so it might be more in line with Blizzard's design intentions without changing the growth of economy vs the growth of tech, like the 12 worker start currently does. It's also the default value for mining, so it's a good starting point. | ||
|
ETisME
12249 Posts
I could try it out but without a proper opening etc, we won't really know what it will be like after the economy is more figured out. | ||
|
Umpteen
United Kingdom1570 Posts
On April 13 2015 00:19 Teoita wrote: @Umpteen: with gold minerals income spikes up for the first 8 workers, but when you go to 16 workers overall income is higher; we instead searched for something that would make 16 and above be close to sc2 currently (ie, hit a similar plateau in income), but make the first 8 workers a lot more efficient, and then next 8 less so. Of course gold minerals is also another way to explore the "change worker efficiency" dynamic that could be explored though. True, overall income with 16 workers is 10% higher, as opposed to 5% (IIRC) with double-harvesting. You inspired me to go back and try again, and I found something very interesting: Three gold patches per base (7 minerals per trip), blue minerals harvest 4 per trip instead of 5
| ||
|
Teoita
Italy12246 Posts
| ||
|
Jukado
805 Posts
| ||
|
Randomaccount#77123
United States5003 Posts
| ||
Plexa
Aotearoa39261 Posts
On April 13 2015 00:33 Umpteen wrote: True, overall income with 16 workers is 10% higher, as opposed to 5% (IIRC) with double-harvesting. You inspired me to go back and try again, and I found something very interesting: Three gold patches per base (7 minerals per trip), blue minerals harvest 4 per trip instead of 5
Despite what teo is saying, the advantage of DH isn't the early game mineral boost (thats just a numbers thing, can easily be manipulated in a variety of ways). The advantage is the shape of the minerals mined per minute (mmpm) vs number of workers graph, and in particular the fact after 8 workers there is less increase on the mmpm per worker added, whereas currently that threshold is set at 16. Your model of adding gold minerals doesn't change this curve, and neither does the LotV or FRB models. This is important because if you accept that 60 workers is optimal in terms of economy supply vs army supply, then the 48 workers mining minerals need 24 mineral patches to be optimal. The gold patch model doesn't change this, and hence doesn't change the 3 base paradigm. FRB changes the 24 patch cap to four bases which is better (and arguably 5 bases due to less workers mining gas). DH changes this to that you need 48 patches to be optimal which is 6 bases and in any normal game you're never going to have 6 bases mining simultaneously. The effect of this is that there is always an advantage to having more bases if you can defend it, meaning there is no effective base cap unlike every other model suggested. It's worth noting that FRB (under Barrins 6m1g implementation) gets really damn close to this, and why that model worked quite well in encouraging expanding. | ||
|
Teoita
Italy12246 Posts
| ||
|
Whitewing
United States7483 Posts
On April 12 2015 21:01 5p4z3n3k0 wrote: Nice post, for dirty protosses... ( joke ofc  ) After I read your post it got me thinking. How does the mule fit into the DH method ? Or would you consider that to be a balance issue, instead of an economic issue ? spaze, edit:Grammar It isn't relevant to the model, since it's existence is already a terran specific boost to income outside the function of scvs. If it's unchanged, it's relative value drops slightly because income rates in general are increased, but it's flat value wouldn't change at all, still 270 minerals returned over the length of its life. | ||
| ||
