|
Dakota_Fanning
Hungary2359 Posts
On September 28 2011 18:28 SnetteL wrote:Show nested quote +[B]On September 27 2011 21:05 Dakota_Fanning wrote:
- Too fast cancel (~0.84 sec): if a command (like train, research, upgrade, hatch) gets cancelled right away Noone is going to build/cancel on purpose for spam, building a wrong thing and correcting it is IMO meaningful apm. It requires buttons to be pressed and doesn't give you an edge that will inflate your APM. Even pro's make mistakes and it is important to be fast to correct them. Eventually, yes, you did do the same as the person who did it correctly but with more buttons pressed. You can't forget however that you also used more time and that if you didn't recongnize and correct your mistake you also would have about the same apm as in the other situations but you might have made a game-ending mistake (thrown down a wrong building or w/e). That's why I feel cancelling something fast should be rewarded.
Yes, I know one will not spam to make something and cancel it, but EAPM is not entirely about filtering spam but to show Effective APM. Making something and canceling it right away is not effective. If you make something but 1 minute later you realize you need the money for something else and therefore you cancel it... that's another story and in my opinion it's not ineffective.
Either way the final version will be, it will not change EAPM noticeably as these fast cancels are very rare.
|
Dakota_Fanning
Hungary2359 Posts
On September 28 2011 18:36 coL.drewbie wrote:
you should use the same things that bwrepinfo used to determine eapm :D although i don't know what it used and the site appears to be in russian so my advice is really bad sry T_T
i <3 sc2gears and use it every day though : )
SC2 and SC-BW are 2 different games, with some fundamental differences which makes the same "old" EAPM algorithm unsuitable (for example you can't in SC-BW but in SC2 you can select multiple buildings, rally multiple buildings, you can select much more units at once, you can queue researches and upgrades etc.).
|
Most ideally would be to have a submenu with all of these options for eAPM calculation, so that the user can configure to some extent what he considers effective in the calculation. In addition, I find it useful to have the control of vision (camera hotkeys, scrolling, unit following) as actions too.
|
I think fast repetition of move maybe shouldnt be excluded. There are many micro cases where you spam and issue a ton of move commands in a short period of time. For example splitting marines against banelings. Otherwise it sounds good.
|
Dakota_Fanning
Hungary2359 Posts
On September 28 2011 19:40 Kashll wrote:
I think fast repetition of move maybe shouldnt be excluded. There are many micro cases where you spam and issue a ton of move commands in a short period of time. For example splitting marines against banelings. Otherwise it sounds good.
Splitting marines involves Select-Move pairs which is considered effective.
If you select a group of marines and you just spam move commands, the marines will not split... And if you spam many move commands in a short period of time then some of them are not really necessary.
|
On September 28 2011 20:00 Dakota_Fanning wrote:Show nested quote +On September 28 2011 19:40 Kashll wrote:
I think fast repetition of move maybe shouldnt be excluded. There are many micro cases where you spam and issue a ton of move commands in a short period of time. For example splitting marines against banelings. Otherwise it sounds good. Splitting marines involves Select-Move pairs which is considered effective. If you select a group of marines and you just spam move commands, the marines will not split... And if you spam many move commands in a short period of time then some of them are not really necessary.
Fair enough
|
How do you have time to write algorithms and implement them in Sc2 gears all the dam time? In any case its much appreciated and keep up the good work.
|
One thing I consider important in APM and EAPM calculations is that it is transparent for the user how the calculations are done. For example the value of Blizzard's current APM implementation is low as we do not know what it's exactly based on (maybe should be called SAP43S = some actions per 43 seconds).
Currently in sc2gears one can review the action list easily. As micro & macro APMs are also separately calculated it would be nice to have filters for both micro & macro actions (yes we have plenty of filters already, but it's not necessarily transparent into which category each actions belongs). Regarding EAPM it would be helpful if reasons why certain actions are disregarded from the calculation would be shown directly in the action list. The following links show how bwrepinfo presented the disregarded actions: http://bwrepinfo.narod.ru/101a.html and http://bwrepinfo.narod.ru/shots.html
|
Awesome stuff. I still haven't actually started using SC2gears but it's always nice to see community members picking up where blizz fails so hard...
|
This is awesome. I can't really think of any faults with what you consider effective. Maybe camera movements but it seems debatable. Is there a possibility of adding SQ into sc2gears? A guy on reddit made some sort of program that pulls the info and calculates it for you. If that was added too sc2gears would be complete IMO
|
I think that queuing up move commands is effective APM. For example, sending lings/overlords on scouting patterns is definitely effective. Maybe queuing up a little circle to block a hatch is not. I also thing that hotkeying and un hotkeying units is effective.
|
I fully support this
my AMP is always the lowest in the games by at least 30 (lately i mostly play team games) but in 90% of the games I win and 25% of the games I lose, I have the highest score points wise (usually by a considerable margin)
I was wondering if that is just reflected by better decision making and use of my clicks
Cant wait to have this implemented!
|
Nice, just wanted to say thanks for your great work 
|
On September 28 2011 18:54 Dakota_Fanning wrote:Show nested quote +On September 28 2011 18:36 coL.drewbie wrote:
you should use the same things that bwrepinfo used to determine eapm :D although i don't know what it used and the site appears to be in russian so my advice is really bad sry T_T
i <3 sc2gears and use it every day though : ) SC2 and SC-BW are 2 different games, with some fundamental differences which makes the same "old" EAPM algorithm unsuitable (for example you can't in SC-BW but in SC2 you can select multiple buildings, rally multiple buildings, you can select much more units at once, you can queue researches and upgrades etc.).
I think in areas where the games compare it would be good to keep it similar to bwrepinfo to make cross-game comparisons possible.
On bwrepinfo's website is a list of things considered excessive apm:
-Building units or upgrades without resources
-Selecting different units without giving commands, if done too quickly (no timeframe mentioned)
-Pressing a hotkey of an already active group
-Useless controlgroup assigning (bind cc to 6, never press 6 in the game or reassign 6 to different unit/building)
-Repeditive Attack, Move, Rally commands issued to nearly the same spot
-Too fast repetition of commands like Attackmove or Move (no time mentioned)
-Senseless duplication of commands like Hold or Stop
There is also a thing called active-zone. Image every command creates a circle around it owned by the player that issued the command. The areas in which circles overlap(like battles) are called active zones and change dynamically throughout the game. In these areas the terms for excessive apm change.
|
On September 28 2011 18:49 Dakota_Fanning wrote:Show nested quote +On September 28 2011 18:28 SnetteL wrote:[B]On September 27 2011 21:05 Dakota_Fanning wrote:
- Too fast cancel (~0.84 sec): if a command (like train, research, upgrade, hatch) gets cancelled right away Noone is going to build/cancel on purpose for spam, building a wrong thing and correcting it is IMO meaningful apm. It requires buttons to be pressed and doesn't give you an edge that will inflate your APM. Even pro's make mistakes and it is important to be fast to correct them. Eventually, yes, you did do the same as the person who did it correctly but with more buttons pressed. You can't forget however that you also used more time and that if you didn't recongnize and correct your mistake you also would have about the same apm as in the other situations but you might have made a game-ending mistake (thrown down a wrong building or w/e). That's why I feel cancelling something fast should be rewarded. Yes, I know one will not spam to make something and cancel it, but EAPM is not entirely about filtering spam but to show Effective APM. Making something and canceling it right away is not effective. If you make something but 1 minute later you realize you need the money for something else and therefore you cancel it... that's another story and in my opinion it's not ineffective. Either way the final version will be, it will not change EAPM noticeably as these fast cancels are very rare.
I think this is the wrong way to look at it. If we start saying that certain actions were purposeful but "ineffective," then where does it end? Remove actions if your micro is bad?
In this example, canceling a building quickly is removed because its considered "ineffective," but a ton of canceled buildings are probably extractor tricks, which would be considered effective by the people doing it.
|
I think the definition seems pretty sound. For SkillCraft, we've been struggling with deciding on something more sensible (from a cognitive measurement perspective) than APM.
We're going to use a number of definitions, and see which seems to be the least noisy based on the sample of games we've got. We'll let you know what we find.
|
I'm looking forward to this. Will we be able to see EAPM and APM at the same time? I think it would be interesting to compare the graphs. For example, spammers will likely have a large difference between EAPM and APM in the beginning, but less difference in the late game. This would make the spamming immediately visible. Not that it really matters, but it would be interesting anyway. I like graphs. :-)
|
United Kingdom20323 Posts
|
United Kingdom20323 Posts
On September 28 2011 20:00 Dakota_Fanning wrote:Show nested quote +On September 28 2011 19:40 Kashll wrote:
I think fast repetition of move maybe shouldnt be excluded. There are many micro cases where you spam and issue a ton of move commands in a short period of time. For example splitting marines against banelings. Otherwise it sounds good. Splitting marines involves Select-Move pairs which is considered effective. If you select a group of marines and you just spam move commands, the marines will not split... And if you spam many move commands in a short period of time then some of them are not really necessary.
yea thats good, thanks
no spam
|
Don't get me wrong, but I don't think EAPM does serve any purpose, whether you do a well thought formula - like you - or some not so well thought one - like Blizzard. The spamming still contributes to the quantity of the overall actions you can do. You can see the quality of the actions from who has won the game anyways.
|
|
|
|
|
|
EPT
