|
Btw, I have run my simulation with a few modifications again:
3tournament system:
Code B: 48random players from the playerbase + bottom16 of previous Code A
Code A: top16 of Code B + bot16 of Code S
Code S: top16 of Code A + top16 of previous Code S
I recorded the winrates for Code A and Code S, they show again 50:50, 50:50, 40:60 winrates for a 50:50, 50:50, 40:60 balance.
Also I recorded winrates for Code A in the old system (without Code B and top A being 48rand+bot16 of CodeS) with the same results.
But increased winrate effects do occur when you put stronger players of the UP race in a weaker enviroment.
One more thing that might be interesting (but some work to implement), would be create stronger players for all races (which I have already done to no effect), but not match players in a tournament randomly, but match them 32vs1, 31vs2...17vs16 in terms of their balance-weighted strength or their previously reached rank.
Maybe like that it I can make sure that stronger players cannot get knocked out by other strong players (Maru cannot get knocked out in a group with soO and Classic early in the GSL) and maybe it changes something? (I kind of doubt it, but it would be interesting regardless).
|
+ Show Spoiler +On July 14 2014 03:33 ZenithM wrote:Show nested quote +On July 14 2014 02:57 Big J wrote:On July 14 2014 02:25 ZenithM wrote:
I actually don't believe in the 50% effect in your models either. If you assume that players are equally skilled and game balance in skewed toward an OP race, all other things being equal, then top players from the OP race will continue to beat down the top players of the UP race, no matter what. That makes sense.
I however think there are possibilities for the 50% effect to happen in more complex models.
Such problems can arise:
- Players of one race are more skilled: I like to think we can dismiss this hypothesis and continue onto more interesting (but less snarky ;D) ones.
- You have a notion of race per-skill effectiveness, which represents the power output you can get from a race relatively to the skill level of the player. If you assume that one race allows for more power in return as you put more skill in it, compared to others, maybe you'll see the 50% thing. It's essentially equivalent to skewing the skill distribution for one race with a transform function (greater variance for a Gaussian distribution, for example), if you want to test that quickly. The simulation will behave as if players of one race are more skilled, but technically it's just because players from the other races are less able to shine through. Depending of the function you choose to apply to the distribution, it may compensate completely the imbalance (you get 50% winrates all across, but player base stay even), but I believe you can find functions that get both the reduced player base AND the 50% effect.
Do you think that makes sense or am I missing something? Doesn't the second argument kind of come down to "top players of the UP race are more skilled" though? Like, the outcome of creating a Gaussian with a higher standard deviation for the UP race, just means that there are more better and more worse players and less average players. [...] Yes, that's what I meant by: Show nested quote +The simulation will behave as if players of one race are more skilled, but technically it's just because players from the other races are less able to shine through. It behaves as the "top players of the UP race are more skilled" fallacy, but the causes are hopefully to be found elsewhere. My personal opinion about the game is that especially top Protosses have less to distinguish themselves with, compared to Zergs and Terrans. It's not that their best players are less skilled, it's simply that the game doesn't let them show their real advantage in skill. I think that's an important distinction, and it's not that unreasonable to think that the races are not equal in the difference in skill they're designed to allow. Edit: Also one thing I wanted to note: Show nested quote +The more I think about it, the more I believe the 50% effect does not exist. This is my current explanation:
On first glance, it is right that only the better Terrans get to play when they are underpowered against, say Zerg. Now what happens is that Zergs start to replace Terrans in racial distribution. However, they are not overtaking by a degree that would allow the top Terrans to brutalize them and even out the winrates. Why? Because Protoss prevents those Zergs from entering the competition at that level, since that matchup is still balanced. A patchzerg beating a Terran still also has to be able to beat Protoss and Zerg players to reach the tournaments, which he simply does not. Even more since the amount of Terrans is low, his ability to beat Terrans is marginally important!
This is indeed fine if we assume that one race isn't inferior to both the others at the same time, or one race isn't superior to both at the same time. I think that at the moments in Starcraft 2 where the "patch[race]" phenomenon occurred, it's precisely that one race was dominating the other 2. And at the moments of near complete extinction in GSL, that one race was kinda bad against the other 2. Edit 2: And I recommend reading that translation. Makes me want to leave the game untouched to see what Flash can bring through practice ((if anything :D).
I read the article, that was great stuff! In short I say: TheMarine for new head of SCII design!
|
't get those Snipe obsessions. Over and over and over again people keep on suggesting this buff. Why exactly that one? If you want to buff the ghost, there are a hundred ways to do it. If you want to buff Terran, there are a million ways to do it.
But Snipe is a shit spell. It's not fun to play with or against it. It's a spam-click competition.
Because Ghosts are underutilized vs Terran and Zerg. Snipe can make the Ghost a more stable unit, instead of purely a counter unit.
For me personally Snipe isn't that interesting of an ability, so would rather see the spell replaced with another cooler ability, but for now and untill someone comes up with a sick idea, we must settle with Snipe to make the recurrence of one of the coolest units in the game, the Ghost.
|
On July 15 2014 20:35 ejozl wrote:Show nested quote +'t get those Snipe obsessions. Over and over and over again people keep on suggesting this buff. Why exactly that one? If you want to buff the ghost, there are a hundred ways to do it. If you want to buff Terran, there are a million ways to do it.
But Snipe is a shit spell. It's not fun to play with or against it. It's a spam-click competition. Because Ghosts are underutilized vs Terran and Zerg. Snipe can make the Ghost a more stable unit, instead of purely a counter unit.For me personally Snipe isn't that interesting of an ability, so would rather see the spell replaced with another cooler ability, but for now and untill someone comes up with a sick idea, we must settle with Snipe to make the recurrence of one of the coolest units in the game, the Ghost.
Many things can do that though, which is what I was referring to with "a hundred ways to do it". There is surely something more interesting than buffing snipe to a status that you can kill stuff with and then limiting its usage to only things that you don't really need ghosts for to begin with so that it's not broken.
A few Ideas:
- Change the +10vs light to +10vs bio, my personal favorite. Puts it nearly on a level with marines (for higher costs) in terms of damageoutput/supply against hellbats, MM, everything Zerg. Stays the same against Zealots/Templar/DT.
Only units against which it is weaker would be sentries and a few air units (phoenix, raven, oracle) - mostly irrelevant stuff.
- plainly more damage. Instead of 10+10vs light, make it 15+5vs light, or plain 20damage. Could be too strong when sniping observers and then killing Protoss armies, but that's a very rare endgame case that could still be solved with proxy canons at the warp in points (easily affordable and often seen in those scenarios) to fall back on.
- Lower cost (they are amongst the most expensive units per supply in the game).
- a little higher HP. Yeah, they are more tanky than MM, they are still very squishy for a 200/100 unit (only 2/3 of a zealot or roach).
- give it a form of stim to move at the speed of MM. I think that's often a major hinderance for ghost play. Without a higher speed they are hard to use with MMM against units like banelings, zerglings or roaches. when falling back you always lose them...
|
On July 15 2014 19:01 Big J wrote:
Btw, I have run my simulation with a few modifications again:
3tournament system:
Code B: 48random players from the playerbase + bottom16 of previous Code A
Code A: top16 of Code B + bot16 of Code S
Code S: top16 of Code A + top16 of previous Code S
I recorded the winrates for Code A and Code S, they show again 50:50, 50:50, 40:60 winrates for a 50:50, 50:50, 40:60 balance.
Also I recorded winrates for Code A in the old system (without Code B and top A being 48rand+bot16 of CodeS) with the same results.
But increased winrate effects do occur when you put stronger players of the UP race in a weaker enviroment.
One more thing that might be interesting (but some work to implement), would be create stronger players for all races (which I have already done to no effect), but not match players in a tournament randomly, but match them 32vs1, 31vs2...17vs16 in terms of their balance-weighted strength or their previously reached rank.
Maybe like that it I can make sure that stronger players cannot get knocked out by other strong players (Maru cannot get knocked out in a group with soO and Classic early in the GSL) and maybe it changes something? (I kind of doubt it, but it would be interesting regardless).
This is fascinating, keep it up!
|
So, this just popped in my head, but is there any reason why the mothership core requirement can't be shifted from the cybercore to the stargate? this should make the window for early aggresion slightly wider whilst also throwing in a tech requirement for MScore related rushes/all-ins.
|
On July 17 2014 16:51 19Meavis93 wrote:
So, this just popped in my head, but is there any reason why the mothership core requirement can't be shifted from the cybercore to the stargate? this should make the window for early aggresion slightly wider whilst also throwing in a tech requirement for MScore related rushes/all-ins.
The entire point of MSC was to give Protoss reliable early game scouting.
|
On July 17 2014 16:53 pure.Wasted wrote:Show nested quote +On July 17 2014 16:51 19Meavis93 wrote:
So, this just popped in my head, but is there any reason why the mothership core requirement can't be shifted from the cybercore to the stargate? this should make the window for early aggresion slightly wider whilst also throwing in a tech requirement for MScore related rushes/all-ins. The entire point of MSC was to give Protoss reliable early game scouting.
At this point I call troll since the first mothership core was "stuck" on a nexus.
|
Free Hallucination gives plenty of scouting as it is, don't need the MSC to play an additional scouting role.
Rather than moving it to the Stargate though, I'd propose slowing the MSC down to the old Overlord speed (around 0.88) and adding a MSC speed upgrade to the Cybernetics Core at 50/50/110.
It would delay any straight up MSC attack on the mineral line by quite a bit and by putting the upgrade at the Cybernetics Core you cannot have Warp Gate and MSC Speed at the same time early on, weakening a Blink all-in.
It wouldn't weaken any defensive play with the MSC as you can still park it between the main and the natural to cast Photon Overcharge.
|
On July 17 2014 16:53 pure.Wasted wrote:Show nested quote +On July 17 2014 16:51 19Meavis93 wrote:
So, this just popped in my head, but is there any reason why the mothership core requirement can't be shifted from the cybercore to the stargate? this should make the window for early aggresion slightly wider whilst also throwing in a tech requirement for MScore related rushes/all-ins. The entire point of MSC was to give Protoss reliable early game scouting.
Your previous admittance that you don't actually play the game, and thus lack even basic understanding of how it works and how it has changed over time really starts to show itself when you make statements like this.
The original mothership core was molasses levels of slow. In the first part of beta it had a speed of 0.47, around 4 times slower than it currently is. It seldom left the Protoss player's base outside of certain situations because it was so easy to catch and kill, not to mention it it was so slow that it could not really render any assistance (well that and it didn't actually have an attack, only spells). Recall was typically used by having the mothership core follow behind your army, and you would retreat to the mothership and then recall or wait for it to catch up before recalling, but this was dangerous because it was too slow to micro away. To put into perspective how slow it was, it was 0.013 faster than the pre-buff overlord, as in the slowest version of the overlord from Wings of Liberty that was frequently regarded as too slow to use for scouting reliably.
The mothership core's original purpose was to help Protoss in the early game with defense, since that was where it was weakest in Wings of Liberty and most in need of help. It had all defensive-oriented abilities, including a way to provide detection and a way to give energy to units or buildings. It was only later on that it was transformed into a faster moving, more offensive-oriented unit. It didn't have an attack of its own, some iterations of it actually attached to the nexus for the cannon. It was a completely different unit than the current rendition.
|
I am trying really hard to stay calm about the current state of balance and don't want to whine all the time, but that Taeja v Jim game just sent me over the edge. The advantage of Protoss mechanics is just so insurmountable when in the hands of a high level player, that even God like micro (massive consecutive snipes and drops) and Protoss errors (lost a colossus vs SCVs) cannot save a Terran player. I am just utterly disgusted at the state of this game.
|
On July 17 2014 23:53 Thezzy wrote:
Free Hallucination gives plenty of scouting as it is, don't need the MSC to play an additional scouting role.
Rather than moving it to the Stargate though, I'd propose slowing the MSC down to the old Overlord speed (around 0.88) and adding a MSC speed upgrade to the Cybernetics Core at 50/50/110.
It would delay any straight up MSC attack on the mineral line by quite a bit and by putting the upgrade at the Cybernetics Core you cannot have Warp Gate and MSC Speed at the same time early on, weakening a Blink all-in.
It wouldn't weaken any defensive play with the MSC as you can still park it between the main and the natural to cast Photon Overcharge.
maybe a bit more like 100/100 or 100/150, 50/50 may not be enough to reduce the power of all-ins that bring a mscore with it.
|
ZvT will be stupid to play and watch again while the lategame weaknesses of terran wont be at all addressed with this (if zergs anyhow survive the 20-25 minute survival game).
I wonder why blizzard does not realize that balancing TvZ on a single stale style of play that allows/forces one race to constantly attack and the other race to only defend for half an hour is nothing but bad. Even hellbats wont have a role in this anymore as they require some commitment from the terran (which includes risks) while going straight into 4m macro rally has it all and can't be countered in early-mid.
Instead of developing the game further into a direction of different openings for both races that kind of counter each other and allow players to dynamically adapt to each other and create interesting variable games, they choose the solution that really makes one thing obvious: They don't at all know how to do it better.
But there are alot of possible approaches. The fact that T is weak in the lategame against both P and Z and is more or less stuck on bio in both mups makes it quite easy to do this better: The Raven missile could be changed so that it always hits in combination with a fix of PDD. The mine should never be the core unit of terran gameplay but the tank! Therefore tank mechanics and counter mechanics could be tweaked as mentioned hundred times in these threads (e.g. immortal shield mechanics, +dmg vs light/massive for tanks, etc.). Thors could be changed to complement this.
I can't imagine a single zerg out there that enjoys this always repeating kind of ZvT meta which does not allow any creativity but is in fact a devensive do or die situation for about 20-25 minutes in every game without anything to fall back on like MSC, mines, wall-ins with bunkers, with banelings taking that long to replenish, and then ends up in the exact same bemoaned situation that we have now. T and Z have reversed roles in TvZ when looking at the race mechanics and available units and this should be some long term goal to fix for blizzard.
Time to put SC2 aside until expansion2 release I guess. These changes wont improve SC2 a bit but only the winrates of terran.
|
On July 18 2014 18:25 LSN wrote:
ZvT will be stupid to play and watch again while the lategame weaknesses of terran wont be at all addressed with this (if zergs anyhow survive the 20-25 minute survival game).
I wonder why blizzard does not realize that balancing TvZ on a single stale style of play that allows/forces one race to constantly attack and the other race to only defend for half an hour is nothing but bad. Even hellbats wont have a role in this anymore as they require some commitment from the terran (which includes risks) while going straight into 4m macro rally has it all and can't be countered in early-mid.
Instead of developing the game further into a direction of different openings for both races that kind of counter each other and allow players to dynamically adapt to each other and create interesting variable games, they choose the solution that really makes one thing obvious: They don't at all know how to do it better.
But there are alot of possible approaches. The fact that T is weak in the lategame against both P and Z and is more or less stuck on bio in both mups makes it quite easy to do this better: The Raven missile could be changed so that it always hits in combination with a fix of PDD. The mine should never be the core unit of terran gameplay but the tank! Therefore tank mechanics and counter mechanics could be tweaked as mentioned hundred times in these threads (e.g. immortal shield mechanics, +dmg vs light/massive for tanks, etc.). Thors could be changed to complement this.
I can't imagine a single zerg out there that enjoys this always repeating kind of ZvT meta which does not allow any creativity but is in fact a devensive do or die situation for about 20-25 minutes in every game without anything to fall back on like MSC, mines, wall-ins with bunkers, with banelings taking that long to replenish, and then ends up in the exact same bemoaned situation that we have now. T and Z have reversed roles in TvZ when looking at the race mechanics and available units and this should be some long term goal to fix for blizzard.
Time to put SC2 aside until expansion2 release I guess. These changes wont improve SC2 a bit but only the winrates of terran.
I think TvZ will be more diversified then only 4M-plays. Hellbat transformation and the banshee-cloak buff are still in place so I expect a lot of aggressive hellion-banshee into hellbat-banshee openings alternating with more standard 3-base rally-pushes for an overall more varied matchup. The need for Zergs to prepare for standard play opens new all-in timings and terran becomes less predictable. However all-ins are still possible for Zerg.
I have high hopes that TvZ will become an amazing matchup again with varied openings by both races and tense micro battles when it comes down to 4M vs ling-bling-muta.
If however your prediction comes true and it will be the same stale game every single time I hope that Blizzard decides to buff hydras and/or infestors to make Roach-Hydra-Infestor viable instead of nerfing Mines like they did last time.
|
On July 18 2014 01:12 Ben... wrote:Show nested quote +On July 17 2014 16:53 pure.Wasted wrote:On July 17 2014 16:51 19Meavis93 wrote:
So, this just popped in my head, but is there any reason why the mothership core requirement can't be shifted from the cybercore to the stargate? this should make the window for early aggresion slightly wider whilst also throwing in a tech requirement for MScore related rushes/all-ins. The entire point of MSC was to give Protoss reliable early game scouting. Your previous admittance that you don't actually play the game, and thus lack even basic understanding of how it works and how it has changed over time really starts to show itself when you make statements like this. The original mothership core was molasses levels of slow. In the first part of beta it had a speed of 0.47, around 4 times slower than it currently is. It seldom left the Protoss player's base outside of certain situations because it was so easy to catch and kill, not to mention it it was so slow that it could not really render any assistance (well that and it didn't actually have an attack, only spells). Recall was typically used by having the mothership core follow behind your army, and you would retreat to the mothership and then recall or wait for it to catch up before recalling, but this was dangerous because it was too slow to micro away. To put into perspective how slow it was, it was 0.013 faster than the pre-buff overlord, as in the slowest version of the overlord from Wings of Liberty that was frequently regarded as too slow to use for scouting reliably. The mothership core's original purpose was to help Protoss in the early game with defense, since that was where it was weakest in Wings of Liberty and most in need of help. It had all defensive-oriented abilities, including a way to provide detection and a way to give energy to units or buildings. It was only later on that it was transformed into a faster moving, more offensive-oriented unit. It didn't have an attack of its own, some iterations of it actually attached to the nexus for the cannon. It was a completely different unit than the current rendition.
You're right, I have admitted that I don't play the game. (Although I've started up again recently) That's why you're not likely to find me discussing numbers or particular strategies. I don't pretend to have any answers, I just try to ask the right questions.
I remembered that lack of scouting options was a huge problem for Protoss in WOL, and given how the MSC has evolved it's a pretty natural mistake to misremember a connection here, don't you think? Obviously I was wrong, thanks for clearing that up. To think, I was actually defending the MSC's current design when I didn't need to be!
|
What if we gave mechanical units of Terran (Hellion/Hellbat, Tank, Viking, BattleCruiser and Thor) an upgrade similar to stim? A toggling "Overdrive" that while active give some percent move/attack speed/transform/deploy increase and deals damage over time? Builds on top of the current Terran skill set and gives mech more viability in late game.
|
On July 22 2014 12:03 Foreverkul wrote:
What if we gave mechanical units of Terran (Hellion/Hellbat, Tank, Viking, BattleCruiser and Thor) an upgrade similar to stim? A toggling "Overdrive" that while active give some percent move/attack speed/transform/deploy increase and deals damage over time? Builds on top of the current Terran skill set and gives mech more viability in late game.
Because the goal is differentiation, not amalgamation.
|
banshee-cloak buff : extremely small compare to debuffs they had.
1) one more detection unit to protos, oracle with speed buff.. So, protos has mobile detection on every tech path. Oracle WAY faster then banshee...
2) Overseer speed buff, making it mach easier to counter banshee as zerg.
Compare to that detection buffs, slight one time cost reduction is nothing.
|
On July 22 2014 12:14 Thieving Magpie wrote:Show nested quote +On July 22 2014 12:03 Foreverkul wrote:
What if we gave mechanical units of Terran (Hellion/Hellbat, Tank, Viking, BattleCruiser and Thor) an upgrade similar to stim? A toggling "Overdrive" that while active give some percent move/attack speed/transform/deploy increase and deals damage over time? Builds on top of the current Terran skill set and gives mech more viability in late game. Because the goal is differentiation, not amalgamation.
Except mech is not viable because it doesn't supplement bio well, and is too costly to go only mech. They need something that can create a mixed composition and this ability would allow it. If you can't micro the unit like bio its not worth the investment when bio is so cost efficient.
|
On July 22 2014 12:17 Mutineer wrote:
banshee-cloak buff : extremely small compare to debuffs they had.
1) one more detection unit to protos, oracle with speed buff.. So, protos has mobile detection on every tech path. Oracle WAY faster then banshee...
2) Overseer speed buff, making it mach easier to counter banshee as zerg.
Compare to that detection buffs, slight one time cost reduction is nothing.
Banshee is the only unit that has never been nerfed since the beta.
|
|
|
|
|
|
EPT
