| Forum Index > SC2 General |
|
OrangeGarage
Korea (South)319 Posts
On December 19 2015 10:37 StarscreamG1 wrote: The correct solution to fix early game was to make MSC start with 0 energy. Cause with this patch the spell is weaker at mid-late game, a lot of units can destroy pylons fast.. whoa this was actually talked about in sgall | ||
|
StarscreamG1
Portugal1652 Posts
On December 19 2015 11:02 RCCar wrote: whoa this was actually talked about in sgall whats "sgall"? | ||
|
StatixEx
United Kingdom779 Posts
| ||
|
OrangeGarage
Korea (South)319 Posts
Its a Korean community for Starcraft. Interesting to see how two communities would think the same think | ||
|
StarscreamG1
Portugal1652 Posts
On December 19 2015 11:11 RCCar wrote: Its a Korean community for Starcraft. Interesting to see how two communities would think the same think Ok, nice then. The msc starting with 50 energy it can even cause stupid things like what parting did at dreamhack, destroying it to get another one eith the 50 energy. | ||
|
purakushi
United States3300 Posts
| ||
Excalibur_Z
United States12231 Posts
Wow, wow, wow. So much content to go over here. 1. 10 subdivisions per league is a lot. There's going to be a lot of tier changing per player. If you want the rankings to be accurate at a glance, if you want them to be meaningful, they need to also be current. There is one distinct advantage to no-demotions and that is being able to easily identify smurfs. When you see someone in Diamond playing against a true Bronze, you know that's a player who has intentionally tanked his rating (or is playing unranked, and those two things have a lot of potential overlap). However, this is sidestepped by the Leave League button. The solution here is simple: remove the Leave League button (why is it still there?), and clearly show on the score screen when a player is playing Unranked (change the color of the unranked player's MMR to match his race-specific unranked MMR: red for Terran, purple for Zerg, yellow for Protoss). Only by doing these two things will it still make sense to not have demotions. 2. Bonus pool is in a weird state, because there are still two parallel ranking systems that exist: points and MMR. As long as you're still using points, then it's okay to keep using bonus pool because the points effectively don't matter for accurate matchmaking, but they still do matter for ranking visibility within a tier. The bizarre part is that with tier buckets being so narrow, and because earned bonus pool points follow you across promotions, your current displayed position is even less of an indicator toward progress into the next tier than it currently is. For enthusiasts on Reddit and TL, it's not a big deal because eventually people will figure out the MMR breakpoints per tier, but it still feels wonky to have to consult external sources to figure out the age-old "okay, now how close am I?" 3. Separating Master and GM makes sense in a way, but feels bad at the same time. Assuming bonus pool sticks around, there is no other way to sort players than by points (since bonus pool influences points). GM would necessarily have to be separate because otherwise you would have weird inconsistencies where the #200 player has a lower MMR than the #201 player, but has more points, making them GM. The only way it makes sense to have one fluid league (which I advocate, by the way) is to eliminate bonus pool for Master+ and rank by MMR. In the event that you still need an activity measurement, require a set number of games per week for players in the top 200 (I'd even consider the return of MMR decay for this specific instance). If you're good enough to hit the top 200 but can't fit in 10 games for the week, you would decay some minor amount (or maybe your rating is set equal to the #250 or #300 player?). The other reason separating Master and GM feels weird is because the MMR threshold for GM is dynamic. Master players don't really have a way of answering "okay, now how close am I?" like they could for lower tiers, even when MMR is published. There's a lot of guess-and-check posting that could be avoided by making that more transparent, and the way you do that is by showing upfront who you have to overthrow in the #200 spot to get promoted. Creating periodic updates to GM league doesn't really address this. Overall, I'm really excited about these changes, and I'm expecting great things here. | ||
|
Aegwynn
Italy460 Posts
| ||
|
SheaR619
United States2399 Posts
| ||
|
ZAiNs
United Kingdom6525 Posts
On December 19 2015 11:14 StarscreamG1 wrote: Ok, nice then. The msc starting with 50 energy it can even cause stupid things like what parting did at dreamhack, destroying it to get another one eith the 50 energy. How is that stupid? It's a very rare and situational cool thing to see IMO  . | ||
|
Agh
United States896 Posts
| ||
|
TheoMikkelsen
Denmark196 Posts
| ||
|
p68
100 Posts
| ||
|
Horiken
Japan68 Posts
| ||
|
paralleluniverse
4065 Posts
The "ladder revamp" is terrible. Basically, you want to remove the GM lockout, but for everyone else you want to reuse the same inaccurate ranking system that is wrong and distorted by bonus pool just with more precision by splitting each league into 10, and claim "Oh it's OK that the ranking system is wrong, because we'll show the MMR, which is right". Why can't we have a ranking system that is right, and still show the MMR? I don't think you've thought through the PURPOSE of a ranking system and the PURPOSE of displaying MMR, and how they differ. Are you trying to create an accurate ranking system or a progression system? Because it's very clear that you're trying to create a progression system but calling it an accurate ranking system and telling us that it's OK because we'll display MMR. It's not OK. 1. Bonus Pool Must be Scrapped Bonus pool tuning We like how bonus pool is effective as a tool that rewards players for continuing to play. The current numbers for bonus pool point accrual don’t seem appropriate for each matchmaking format. We intend to explore our data about how often players/teams play in each format in order to get better numbers per game format. No. Bonus pool is the most distortionary factor that makes your rank wrong and it needs to be removed. By the end of the season, a extremely large majority of your points is bonus pool, meaning that ladder ranks are largely about massing games (up until bonus pool is spent at the end of the season), not skill. Bonus pool creates a treadmill effect where you are penalized by having your rank intrinsically decline every single hour, and you need to play to get back up to the same position. Bonus pool distorts ranks. When a player's MMR has actually decreased, bonus pool artificially props up his rank by increasing the time taken for points to self-correct. Bonus pool penalizes people who start the season later. If a player has 1500 MMR, 300 uncertainty about MMR and spent all his bonus pool, and another player has join the season later, requires 100 games to spend his bonus pool, but after 10 games has 1500 MMR, 300 uncertainty about MMR, then why should the second player be ranked significantly lower after these 10 games, when in fact both players are the exact same? Why should the ranks be wrong until the second player needlessly plays an additional 90 games? There is no reason. It's an attempt to make ranks wrong by obfuscating them. Bonus pool rewards activity, but there is no reason to reward activity when what we really want to know is uncertainty about MMR. There is no need to use bonus pool or activity as a proxy to uncertainty about MMR, when you can directly calculate uncertainty about MMR and factor it into ranks directly. There is a need to stop inactive players from clogging up the ranks, but that can be dealt with by having shorter seasons and a play X (X = 20 sounds about right) games in the season requirement. When players are on multiple ladders (e.g. 1v1, 2v2, multiple arrange teams, multiple archon partners), they are penalized by the bonus pool unless they spend the bonus pool on every ladder. If a player is on 10 ladders, they will need to play 10 times as many games as an exclusively 1v1 player or else the bonus pool will heavily penalize them. The ranking system will NEVER BE ACCURATE unless you SCRAP the bonus pool penalty. If you disregard all these facts showing the bonus pool is distortionary and makes ranks wrong and choose to keep the bonus pool, at least use a more accurate name so that you are not deceiving your players because it's NOT a bonus, rename it to "penalty pool". 2. MMR vs League/Ranks/Points Showing MMR We’re thinking we can show MMR as a tool for players to use so they can accurately tell exactly what someone’s skill is. Tier system We still believe it’ll be the most fun for players to focus on which tier of which league he or she belongs to. We’re leaning towards keeping mid season demotions disabled here. The main reason for this is that your current skill is portrayed by MMR already, so it will be cooler to see your highest rank that season through the League/tier system. Keep loading screen information simple, but have more transparency on the score screen (for example, showing the MMR of both players on there) This is mostly based on your feedback saying that too many details about an opponent before a match might introduce weird factors that could disrupt your ability to play well. We’re thinking we can focus on providing that information after the match has ended. Displaying MMR is a great idea. But having MMR and points being different and no mid-season demotions makes ranks inaccurate. You are sending mixed and confusing signals to the player. On one hand you claim that MMR is skill (despite the name "matchmaking rating" doesn't exactly say "skill rating"), on the other hand you claim that league/ranks is skill. But rank and MMR are different. There will be people with high MMR, low league/rank and people with lower MMR, higher league/rank. Which one is really skill? Which is correct? Two metrics, which is the "true" skill rating? Points should be changed to MMR or moving average of MMR or X% quantile of MMR (X = 5 or 10 or 20, this is what accounts for uncertainty about MMR so that you don't need to have a bonus pool as a terribly bad proxy of uncertainty about MMR) or moving average of the X% quantile of MMR. Or have points track one of these 4 metrics. Track means to have points start at 0 after placement and have points quickly converge to the metric after several games, or to set it equal to the metric after placement and then set change in points equal, up to a cap, to change in the metric. Finally, have leagues as a percentile of points, this is a fluid and continuous scale with no sticky league boundaries. Under your "display both MMR and league/ranks" plan you are sending mixed signals about which is the "true" rank. Under this suggestion, it is clear: MMR is the current point estimate of skill rating. Points, depending on which of the above metrics you choose, is (the trend of) your skill rating (after accounting for uncertainty in your skill rating). And league/rank is the percentile of points out of all active players. Distorting ranks by having no mid-season demotions just so that it can be a cosmetic display of your highest league per season is the opposite of what an accurate ranking system should be. If you want to show the highest league in the season, then just display that on the profile, or have the portrait border reflect that, don't wreck the ranking system for it. Allowing player's to be demoted from their league tier, but not their league would arbitrarily make the league boundaries more special than league tier boundaries, privileging those who just barely scrape into a higher league and creating a bulge of players at the lowest league tier. 3. Season Length Ladder Revamp We’ve been discussing the details of our ladder revamp and wanted to give you an update so we can get your feedback on the current direction. A good place to begin is to discuss season duration. With all the things that are changing, this is something we can entirely revisit. In Heart of the Swarm, Ladder Seasons aligned perfectly with the 3 seasons of WCS. However, we’re considering whether Ladder Seasons could be made shorter, allowing you to work towards achieving your maximum potential more frequently. There are definitely pros and cons that we see in each method, but we wanted to hear your thoughts to help decide how we handle Ladder Seasons next year. Please give us your thoughts on this. We’ve got quite a few other ideas to discuss with you though. Let’s go in-depth and take a look at our current thoughts about the Ladder: At Blizzcon, you mentioned shorter seasons to reduce the effect of bonus pool, i.e. the effect of massing games, has on increasing ranks. Now you claim that it's "allowing you to work towards achieving your maximum potential more frequently". This statement has no meaning. I've read it again and again, I cannot work out what the meaning of "allowing you to work towards achieving your maximum potential more frequently" is. It is a completely vacuous statement. Given that MMR, and hence league, is unchanged by seasons, what is the PURPOSE of having seasons if in a perfectly accurate ranking system, everyone would be put back exactly at the same skill rating (and if they're not it wouldn't be a perfectly accurate ranking system because it would mean the previous skill rating was wrong)? Blizzard probably thinks of seasons of having 2 purposes that are not really valid: 1. To snapshot your ladder rank at a point in time (if so, then why not just snapshot without resetting). 2. A progression system to re-begin the ladder grind (this is antithetical to an accurate ranking system because your skill didn't reset to 0 like your rank at the instant that the ladder reset, in fact it's almost certainly the same, meaning a ladder reset should have zero effect). But there's another reason to have seasons: 3. To kick inactive players off the ladder. This is the only legitimate purpose to having seasons and for this purpose, it make sense to have shorter seasons (say, a season every 1.5 to 2 months). The problem, however, is that you want to use bonus pool to penalize inactivity, and bonus pool makes ranks wrong. To deal with inactivity, it makes more sense to remove bonus pool, have shorter 1.5 to 2 month seasons, and have a play X games in the season requirement (X = 20) in order to have your rank snap-shotted at the end of every season. 4. GM and League Promotions/Demotions We are exploring to see if it’s better to have separate leagues for GM and Master The main reason for this is that getting promoted to GM is such an awesome experience. If we were to combine these leagues like we discussed at BlizzCon, we would lose this cool factor. The most important issue to solve here is to still have accuracy even if we split the leagues. We definitely don’t want the situation that we currently have, where a player is clearly not at the GM level anymore but he can’t get demoted out of it as long as he’s active. The current discussion point here is to update GM at set, specific intervals. This way, players can expect a change will happen at a set point of time that is known to everyone, and play the game with this information. Oh, so you think the ladder system is going to be short enough to reduce the penalty of not massing games that is dished out by the bonus pool for the 99.999% of players not in GM, but it isn't going to be short enough for GM players? Your dilemma here is nonsensical. If GM is the top 200 of Masters, then how do you "lose this cool factor" of being promoted to GM when you still actually have to get into the top 200 to get promoted into GM. So no, you don't lose the cool factor of being promoted. A promotion is still needed at some point in time. On the other hand, if you make GM a league, i.e. with sticky boundaries, then you're deliberately distorting ranks to keep people who don't deserve to be in the top 200 in GM, hence undermining your claim that you want accurate ranks and for people who don't deserve to be in GM anymore to be kicked out. Ideally, ALL leagues should be fluid like the GM/Masters boundary, i.e. there are no sticky league boundaries. Your idea of having GM promotions at fixed time intervals (once a week, perhaps) is not a bad one, and it's something I've suggested previously for all leagues, it is better than the distortion of sticky boundaries, because it can be interpreted by players, this is the percentile at a specific point in time, whereas players can not derive any such meaning from failing to be promoted/demoted due to a sticky boundary despite meeting the MMR requirement. Sticky league boundaries must be removed. Whatever you decide (fluid league boundaries or league updates at fixed times, the former is preferable, the latter is acceptable, but sticky league boundaries must be rejected), it should apply to ALL leagues, not just GM. So if GM updates once a week, every league, including Bronze, should update once a week, there is no reason why only GM should be privileged with this less distortionary way of calculating ranks. 5. Summary of Suggested Changes - Display MMR. - Remove bonus pool. - Reduce season length to 1.5 to 2 months. - Introduce a play X (X = 20, say) games per season requirement for a player to be considered active. Inactive players are not ranked at the end-of-season snap-shot. - Choose 1 of the following: Change points to MMR or moving average of MMR or X% quantile of MMR (X = 5 or 10 or 20) or moving average of the X% quantile of MMR, or to track one of these 4 metrics. Be transparent on which you choose, i.e. what points are. - Determine league and league tiers based on a percentile of points out of active players. - Add back mid-season demotions and remove all sticky league boundaries. - Choose 1 of the following: Fluid league boundaries that are updated in real time, or league updates at fixed times such as once a week. Apply this to all leagues, not just GM. | ||
|
NKexquisite
United States911 Posts
Not sure what impact the Disruptor change will have, but it did seem rather silly watching Protoss just throw balls at each other just hoping they would hit for 10 minutes. | ||
|
BronzeKnee
United States5214 Posts
On December 19 2015 09:21 jinjin5000 wrote: Reducing parasitic bomb damage isn't gonna do much other than being worse vs capital ships. Most air units have low health abyway and will get cleaned up by corruptors, which are decent at cleaning up The real problem is the insta-cast and stacking damage. Parasitic bomb should be a skill shot like Fungal then, easier to dodge with fast flying units. | ||
|
jinjin5000
United States1395 Posts
On December 19 2015 12:51 BronzeKnee wrote: Parasitic bomb should be a skill shot like Fungal then, easier to dodge with fast flying units. just make unit glow green for 1-2 second before it affectrs the unit so opponent have chance to counter it | ||
|
HeroMystic
United States1217 Posts
On December 19 2015 12:36 NKexquisite wrote: It's an overall good patch. I don't think the Thor change matters, Ravagers did seem to hit the field a bit early and parasitic bomb was obviously too strong. Not sure what impact the Disruptor change will have, but it did seem rather silly watching Protoss just throw balls at each other just hoping they would hit for 10 minutes. Air units that'll care most about the Thor buff the most: -Vikings -Liberators -Ravens -Corruptors -Vipers -Void Ray -Warp Prism In TvT this'll most likely invalidate Vikings by a severe amount, and Thors will now win vs Liberators. In TvZ, Corruptors will get zoned out pretty hard and this'll keep Liberators safe. The PB nerf means more Vipers are needed and Thors will kill Vipers in three volleys. Nothing really changes in TvP. Kim said this should be a mech buff but it's actually just make Thors a good support unit for Liberators in TvZ, which in turn will make Liberators a good support unit for Bio, which in turn is a buff to Bio, not Mech. | ||
|
jinjin5000
United States1395 Posts
On December 19 2015 13:24 HeroMystic wrote: Air units that'll care most about the Thor buff the most: -Vikings -Liberators -Ravens -Corruptors -Vipers -Void Ray -Warp Prism In TvT this'll most likely invalidate Vikings by a severe amount, and Thors will now win vs Liberators. In TvZ, Corruptors will get zoned out pretty hard and this'll keep Liberators safe. The PB nerf means more Vipers are needed and Thors will kill Vipers in three volleys. Nothing really changes in TvP. Kim said this should be a mech buff but it's actually just make Thors a good support unit for Liberators in TvZ, which in turn will make Liberators a good support unit for Bio, which in turn is a buff to Bio, not Mech. eh? Nah. its a buff to mech. No way its straight up buff to bio tankivac you can argue its a buff to bio but thor? no way. | ||
| ||
StarCraft 2 StarCraft: Brood War Dota 2 League of Legends Counter-Strike Other Games |
|
Sparkling Tuna Cup
AllThingsProtoss
WardiTV Spring Champion…
Online Event
BSL
Afreeca Starleague
Soulkey vs Scan
Barracks vs Light
Wardi Open
Monday Night Weeklies
PiGosaur Monday
Afreeca Starleague
Sharp vs Speed
Jaedong vs BeSt
[ Show More ] |
|
|
EPT
