|
your Country52796 Posts
On May 05 2015 10:52 Soularion wrote:Show nested quote +On May 05 2015 08:01 Plexa wrote:We'd also be interested in hearing what other participants thought about the mod  Thanks beasty/lalush I dumpstered a masters player with proxies, 10/10 mod Seriously though, rushes /did/ seem a bit strong. But then again, I'm way, way too awful to really say much, but big thanks for putting it together and even bigger thanks for letting me shit on people with your model. <3 forevers
Honestly though, if rushes are a bit stronger that's probably a good thing. There is way too much greed in non-mirrors at the moment
|
Legit question. In this mod, is 10 hatch actually viable? Someone did it in one of the games I casted and it seemed alright.
|
On May 05 2015 12:27 The_Templar wrote:Show nested quote +On May 05 2015 10:52 Soularion wrote:On May 05 2015 08:01 Plexa wrote:We'd also be interested in hearing what other participants thought about the mod Thanks beasty/lalush I dumpstered a masters player with proxies, 10/10 mod Seriously though, rushes /did/ seem a bit strong. But then again, I'm way, way too awful to really say much, but big thanks for putting it together and even bigger thanks for letting me shit on people with your model. <3 forevers Honestly though, if rushes are a bit stronger that's probably a good thing. There is way too much greed in non-mirrors at the moment
There is nothing left in SC2 but greed.
|
United States4883 Posts
On May 05 2015 12:49 HeyImFinn wrote:
Legit question. In this mod, is 10 hatch actually viable? Someone did it in one of the games I casted and it seemed alright.
I noticed that too. Someone put down an insanely early hatch, and I was like WHAT, but the other community casters didn't seem to notice, and it didn't have any negative impact on the early game in comparison to the pool first of his opponent.
Retracted statement: + Show Spoiler +I can't say I watched the entire thing, but I watched enough to say that I think this mod definitely isn't quite at the right spot yet. In one game, we didn't even notice that the game wasn't loaded with mod until ~8:00 into the game, and only because the minerals were going up in intervals of 5. So yeah, I think that Beastyqt's words are right on, especially regarding some of the racial imbalances; cheeses are stronger, all-ins are a bit more powerful, and having one more base than your opponent doesn't translate into an economic lead very quickly. Most likely, we'll need to look at tweaking the numbers a bit and dropping the early game income some, and we may end up at DH8 in the end. Also, we might try experimenting with Blizzard's method of reducing the amount of overall minerals per base to 1350 or lower and see how that turns out.
I think we're on the right track, but I echo the sentiments that this feels very much like HotS with very marginal benefits for expanding. Our goal is to discourage 3-base turtling and ultimately give more strength to mass expanding, but this was not *always* the case in the games that were played. At the same time, though, much of the economy hasn't been playtested thoroughly, and more substantial data is necessary to make any informed decision on whether we've actually created a better, dynamic environment or just put together a quality of life fix.
A note about MULEs: a lot of people have addressed the role of the MULE in the DH economy, and I think a lot of that is due to misunderstanding about the nature of worker pairing and the interaction of MULEs with workers. First off, MULEs always stack on top of SCVs and mine no matter what, so in the DH model, MULEs still mine the same amount of minerals in the same amount of time as they did in HotS. The only difference is the efficiency of MULEs vs workers, especially after mineral lines reach optimal saturation at 8 workers; at this point, the efficiency (and thus income) of each worker gets progressively less while the MULE stays constant. What this ends up meaning is that MULEs have comparatively decreased value on mineral lines that have 8-12 workers on them (compared to HotS) and increased value on mineral lines that have 12+ workers, meaning that Terran is unlikely to mass expand and spread workers, and sitting on 3-4 bases with MULEs is ideal. I don't have any hard math on this, but these are my observations, and a lot of the people in Bacon_Infinity's chat were bringing this up when we saw a 3-base Terran keep fairly even with a 6-base Zerg.
I'm not sure that this is a bad thing as much as it creates an asymmetrical balance (i.e. in BW, Zerg was supposed to mass expand and chip away at the Terran while Terran was generally encouraged to take bases more slowly, take cost-effective trade, and eventually build up an ultimate army that couldn't be stopped). We'll have to test more, but I wanted to bring attention to the MULE and how it appears to work in the DH model.
|
I fear the constant adjusting of the model weakens the point you are trying to make.
All those conclusions drawn about viable builds without a metagame (or to a less degree fleshed-out balance) to support DH9 sound very premature to me.
|
On May 05 2015 16:12 Big J wrote:
I fear the constant adjusting of the model weakens the point you are trying to make.
All those conclusions drawn about viable builds without a metagame (or to a less degree fleshed-out balance) to support DH9 sound very premature to me.
Alas, It seems this will suffer the same fate as most projects that thinks in terms of solutions instead of identifying and specifying problems first, you end forcing your solution to a not well described environement and you lack the tools and indicators to correctly validate the said solution.
It can work out of sheer luck and godly inspiration, but most of the time you end up trying to put a screw in a wall with hammer.
Threfore, such tournaments can't be conclusive cause you don't know exactly what you are testing nor how you measure success or failure
It's a good public relation thing, though... and who knows, if you hit hard enough with the hammer, the screw might end up stuck in the wall...
|
Aotearoa39261 Posts
Actually, I'm pretty pleased with the tournament overall. The model seems to be working as intended, will post a follow up post in the near future.
|
United States4883 Posts
Actually, in retrospect to some of the comments Plexa made to me in Skype chat, I retract statements about uncertainty. In a lot of ways, this tournament was a huge success of the model, but a lot of it was hidden in the non-surprise factor of its effects. In other words, the changes definitely affect the way the game works after 3 bases (which is what we were aiming for) without making it feel different from HotS.
I still believe we need to get more research done, but I think we're on the right track.
|
Although i only played two matches (ZvT and ZvZ) i think it was definetly noticable.
I'm not quite sure how many workers you want to have per base (on top of the 8/9 "optimal" workers).
The ZvT was quite cheesy, but in one of the games (iron fortress, i believe). I definetly noticed, that i had much more larvae (for army) on 3 bases with a similar, or better, economy on two bases. With an easy to hold 3rd base, i feel like the model allows for much more army production in early/mid game, which feels good vs hellion play.
The ZvZ surprised me as i usually do 2 base roach timings (They definetly do not work the same way in DH). My opponent took an early third base and his army just skyrocketed from there.
Overall i liked the model, but one thing i noticed (which i'm not sure i like):
When a worker completed its first round of mining, another worker can "interrupt" him and mine his first round, instead of the first one doing his second. Sometimes i got really confused why I didn't get any minerals, just to get a whole bunch in a very short time.
|
ZvZ definitely is very different, you kind of have to take the 3rd base much much faster and you are going to be punished harder for staying on 2 bases for too long.
|
United States4883 Posts
On May 05 2015 16:38 Rasias wrote:
Although i only played two matches (ZvT and ZvZ) i think it was definetly noticable.
I'm not quite sure how many workers you want to have per base (on top of the 8/9 "optimal" workers).
The ZvT was quite cheesy, but in one of the games (iron fortress, i believe). I definetly noticed, that i had much more larvae (for army) on 3 bases with a similar, or better, economy on two bases. With an easy to hold 3rd base, i feel like the model allows for much more army production in early/mid game, which feels good vs hellion play.
The ZvZ surprised me as i usually do 2 base roach timings (They definetly do not work the same way in DH). My opponent took an early third base and his army just skyrocketed from there.
Overall i liked the model, but one thing i noticed (which i'm not sure i like):
When a worker completed its first round of mining, another worker can "interrupt" him and mine his first round, instead of the first one doing his second. Sometimes i got really confused why I didn't get any minerals, just to get a whole bunch in a very short time.
In other words, the stream of income didn't feel steady? May I ask, is this an issue all the time, or only on certain worker counts (i.e. "especially noticeable after 12 workers")?
We did notice that some of the workers would occasionally pair on a mineral node and trade turns with harvesting cycles, but generally the worker pairing would break after one full mining cycle (i.e. when the workers returned cargo). Any kind of anomalies or kinks in the worker behavior is always interesting to look at, especially because our goal with DH is to make the workers unpredictable :p.
|
I am looking forward the replay pack to make a more educated statement regarding the mod.
However, here are my current thoughts after watching the bacon_infinity stream, and then switching to Catz for the finals.
New Strategies in DH9
Contrary to some believes, I think DH9 is close to optimal. We have viable early agression play which is not an all-in. We have a viable tactic of denying/delaying natural and 3rd. We have a viable containment strategies. And of course we have a viable 4-base play.
Of course, if you put each of those strategies into an extreme, you get weird games. e.g. going greedy fast 6-base play can get you killed, because you lack an actual army. People who expected it to be a viable, non-cheesy strategy are probably dissapointed; but having an expo give instant major benefit - that would be overpowered and would kill almost any other style of play.
Let me stress it: DH9 is to encourage expanding. But it is not here to kill any other style, such as low-base aggressive play, or even turtle play. Every option should remain viable.
There are of course some concerns:
Concern 1: Early all-in
Early agressive game which is an all-in. 8 workers mine at 100% efficiency, while 15 do not. As a result, the latter has a bit less of extra money (compared to Vanilla HotS) to defend against an all-in on 8 workers. Once players become aware of this phenomena, they may adjust scouting and defensive play. It may or may not be enough....
Concern 2: Mule strength
Secondly, the topic of MULEs. As I said repetedely in the chat during the stream, and SC2John pointed it out over here. The workers are generally a bit more efficient at mining, compared to Standard. The Mule hasn't changed though. As a result - in terms of raw resources - MULE is a bit weaker. Only when base goes past 14 workers, DH9 income falls behind Standard and bonus resources coming from a MULE become stronger.
But the topic of MULEs can be seen from another perspective. Each race has a macro mechanic. Protoss has chronoboost-on-Nexus, Zerg has inject-larvae-into-drones, Terran has Mules. Out of those three, Protoss and Zerg get an indirect nerf, because the faster worker production gives diminishing returns. Terran mule does not change.
For that reason, I am inclined to change my original statement, and agree that Mules got an indirect buff and are mildly overpowered. We may need to nerf their income, for example - by 5% - to match the mining rate of a 16-worker base.
On May 05 2015 15:48 SC2John wrote:
In one game, we didn't even notice that the game wasn't loaded with mod until ~8:00 into the game, and only because the minerals were going up in intervals of 5.
It is because the economy change is never visible instantly. You notice things only after a while when things do not add up. I started suspecting that something is fishy a bit later, but was looking for a more direct proof.
|
On May 05 2015 16:38 Rasias wrote:
Overall i liked the model, but one thing i noticed (which i'm not sure i like):
When a worker completed its first round of mining, another worker can "interrupt" him and mine his first round, instead of the first one doing his second. Sometimes i got really confused why I didn't get any minerals, just to get a whole bunch in a very short time.
On May 05 2015 16:45 SC2John wrote:
We did notice that some of the workers would occasionally pair on a mineral node and trade turns with harvesting cycles, but generally the worker pairing would break after one full mining cycle (i.e. when the workers returned cargo). Any kind of anomalies or kinks in the worker behavior is always interesting to look at, especially because our goal with DH is to make the workers unpredictable :p.
Double Harvest does not break worker pairing. They can still pair for a long period of times, but the income is getting lower nevertheless.
What you describe, is what I call "an interleaved mining". You have two workers A and B, harvesting in the order ABABABRR (R standing for "return cargo"). Other option is "sequential mining" where you have AAARBBBR. In both cases the drop of performance is similar. I did some math and then extensive testing in the past to ensure that[1].
However, you are right that another side effect of interleaved mining is that you wait for about 6 seconds having no income, and then suddenly two workers return giving you 18 minerals total. It's not ideal, but it is most prominent when you are at 9-12 workers in a single base. As the number of bases and workers grow, it is not noticeable (right?)
Btw, SC2John - DH9 is not trying to make workers unpredictable. That's why I didn't like Starbow approach in the first place. We want predictability and reproducibility. What is important is the diminishing return effect.
----
Reference
[1] http://www.teamliquid.net/forum/sc2-maps/471776-mod-double-harvesting-better-saturation-curve?page=2#24
|
|
|
On May 05 2015 17:04 BlackLilium wrote:
What you describe, is what I call "an interleaved mining". You have two workers A and B, harvesting in the order ABABABRR (R standing for "return cargo"). Other option is "sequential mining" where you have AAARBBBR. In both cases the drop of performance is similar. I did some math and then extensive testing in the past to ensure that[1].
However, you are right that another side effect of interleaved mining is that you wait for about 6 seconds having no income, and then suddenly two workers return giving you 18 minerals total. It's not ideal, but it is most prominent when you are at 9-12 workers in a single base. As the number of bases and workers grow, it is not noticeable (right?)
yup, that's what i meant. I noticed it in the early game, at like 13 supply, i waited for minerals for pool, and it kinda "jumped" from 190 to 220 or something. I don't really know how important that is, but i could imagine that it could be confusing that your mineral counts jump differently depending on the map + AI (?).
On more bases i didn't notice it directly, but sometimes i felt like having small bumps in the mineral stream (although i think it's irrelevant later)
Edit:
What do you guys think is the max worker count on a base that's kind of useful? I tried 14 and i'm not sure if that's optimal.
|
You can avoid interleaved mining by sending workers to minerals at the right time (when the other worker is about to end mining). That's something you can do if you wish early game, since you have some "APM budget" anyway at that time.
Max useful worker count? I would say, it is still 16. Workers till 16 have 50% efficiency or higher. Above, it trops to 30% and goes down gradually...
|
Just for the record: watching the DH mod in various games really felt great. I'm noone to heavily read into statistics and go through graphs and curves which is why I mostly didn't say anything about that topic and lurked a lot. However, just from a consuming viewer's point of view - it looks much better than the current sc2 economic model. I really hope Blizzard looks into it and adapts it into LotV without too many tweaks.
Thanks for all the work to everyone involved in creating DH and everything associated with it!
|
On May 05 2015 16:28 Gwavajuice wrote:Show nested quote +On May 05 2015 16:12 Big J wrote:
I fear the constant adjusting of the model weakens the point you are trying to make.
All those conclusions drawn about viable builds without a metagame (or to a less degree fleshed-out balance) to support DH9 sound very premature to me. Alas, It seems this will suffer the same fate as most projects that thinks in terms of solutions instead of identifying and specifying problems first, you end forcing your solution to a not well described environement and you lack the tools and indicators to correctly validate the said solution. It can work out of sheer luck and godly inspiration, but most of the time you end up trying to put a screw in a wall with hammer. Threfore, such tournaments can't be conclusive cause you don't know exactly what you are testing nor how you measure success or failure It's a good public relation thing, though... and who knows, if you hit hard enough with the hammer, the screw might end up stuck in the wall...
The problem identified is the lack of diminishing returns on worker efficiency, and that's what DH is trying to solve. The mod is supposed to highlight precisely what happens when you implement this concept and therefore I think it's worthwhile to tweak the model so that the best possible version exists which is capable of showcasing this idea. If the mod completely destabilizes the early game you'll never see the intended effects because games will end too quickly.
But of course DH is always going to fail since any change to the game will destabilize it in some way and the authors don't have the power to add the necessary compensations such as tweaks to macro mechanics and so on. It's not supposed to be TL's pro mod after all; nevertheless I'm sure the TL strategy team is aware of this and will simply try to analyze the available data with these limitations in mind.
|
Any insight on how the tournament when and how economy changed with this model ? was it impactfull ? visible in the game ? did it produce better games ?
|
|
|
|
|
|
|
|
EPT
