I was wrong: Alacrity not so bad for Bodyguards

[DukSaber]

I know right?

 

It's pretty absurd that I would devote this much effort into trying to get a troll to provide useful feedback. Those 4 Steps aren't set in stone. I just made them up on the fly to try and teach you how to think rationally and provide constructively provide feedback, since you had clearly have little experience with either.

 

What can I say? I'm just a nice guy like that.

 

 

Im still dying to hear from a single person that gets anything from these graphs that are based on nothing.

 

Anyone?

 

You almost had the last post until i made this one.

[DomingoDarkrage]

Im still dying to hear from a single person that gets anything from these graphs that are based on nothing.

 

Anyone?

 

You almost had the last post until i made this one.

 

I got something out of it. Please leave the thread if you are not sure how to give constructive feedback.

[Frisemort]

Thanks for the good work, ignore the trolls.

[RuQu]

I should mention here that I made another mistake (oops).

 

As I was experimenting with ways to view the results of the simulations, before I settled on HPRCT, I was at one point generating HPCT results where I included the cast time of HS when you decided that it wasn't worth casting anything else.

 

That's what leads to the dips, the conscious choice to cast a low-HPCT ability because it is free and not much healing is really needed.

 

Well...I forgot to comment that line back out, so the initial graphs I put up were of HPCT, not HPRCT. I have the new Rakata ones up now, Columi are ready to go I just need to edit them in, and Tionese are being generated.

 

These curves are much, much smoother. They also show the drastic DR that Surge suffers from, and greatly increase the value of Alacrity which was held back by the instant nature of HS (RSh) before.

[DukSaber]

I got something out of it. Please leave the thread if you are not sure how to give constructive feedback.

 

 

Well what did you get? The question was tell me something u got out of it. Then ill tell you why its wrong.

[TempestasSilva]

Well what did you get? The question was tell me something u got out of it. Then ill tell you why its wrong.

 

I tell you what I got. I got a better understanding of how alacrity works and is affected by our choices. I also got that some people like to be jackholes instead of constructively helping. Congrats, you have made that point loud and clear.

 

RuQu has taken a stat which no one had a clue about and everyone initially said was absolutely worthless and added some actual science and math to it which has vastly helped us better understand it. And despite being insulted and belittled, he has taken constructive criticism and made at least two major revisions. He has done so while sharing the details of the calculations so others could effectively review and comment on his work. And the whole while he has been a lot more professional when dealing with people like you than I could ever be.

 

What the heck have you done to help the community?

[Brad_Pitt]

alacrity was never worthless, it was always something thats in the bh medic gear, paired with power. sometimes its surge and crit though depending on the tier.

 

alacrity should have been always viewed as, saving global cooldown and faster casting time.

 

global cooldown is 1.5 seconds on instant moves. by adding alacirty and getting your 1.5 second activation moves down to say 1.3, everytime you cast moves that are lower than the 1.5 global cooldown, you are saving up the time towards using an instant that normally wouldnt be there or healing or damaging moves.

 

so for instance, healing scan 1.3(normally1.5 with tree), rapid scan 1.7s(normally 2.0 with tree), total time savings of .5s so far.

 

then using other moves like powershot 1.3s(normally 1.5) will end up saving you some time as well. so imagine shooting 8 powershots, thats a savings of time of 8 * .2s, so 1.6 seconds. so thats a free power shot(considering time) and an extra .3 seconds added on to the next powershot. or using an instant move which takes 1.5s and an extra .1 left over to use towards something else.

 

This is all obviously post nerf of surge, that i am talking about.

 

 

 

but the pre nerf still applies in ways of shooting out tracer missles. because having a free tracer missle every 7 shots, is an extra 1800-3.2k dmg. so the gains can be directly correlated to say something along the line of

 

(1800-3.2kdmg)

1800/9.8(seconds required to cast 7 and thus ganing a free powershot)=183 min added to DPS

3.2k/9.8 =326 max added to DPS

 

so average gain to DPS would be about 256, assuming you hit. sorry if this math is kind of funky, i might not have factored in the time to use the free power shot, although im not sure i needed to. hmmm....

 

 

I do however agree only at the higher levels does this make the most sense, because you get the biggest return on multipliers and having maxxed our gear.

Edited March 11, 2012 by Brad_Pitt

[RuQu]

alacrity was never worthless...etc

 

The problem was, for quite a while there were two camps who were both operating on a combination of "feelings" and "napkin math." The one side said "things go faster." The other said "It doesn't do anything for half our abilities, and it makes others cost more."

 

The trick was to try and put numbers on it, and that takes more than just napkin math. But, of course, the more complicated the analysis, the greater room for error if an assumption is wrong.

 

This would all have been easier if I was a DPS Commando, since DPS is an obvious and easy metric. As a Combat Medic, I instead decided to address the Healer side first, which is far more complicated, to the extent that SimulationCraft, the very complex (and well respected) WoW simulator never fully realized a proper healing simulation. (Yes, I know they had a few, but I am actually in discussion with the author of said modules and he was never happy with the quality of those sims)

 

So yes, while the value of Alacrity to Surge was different pre-nerf, alacrity was, apparently, always worthwhile. And now we have more to back that up than just people yelling back and forth over differences in their feelings.

[Ewert]

The problem was, for quite a while there were two camps who were both operating on a combination of "feelings" and "napkin math." The one side said "things go faster." The other said "It doesn't do anything for half our abilities, and it makes others cost more."

Well tbh, calling something napkin math just because it's not a fancy spreadsheet of insertable values & spesific healbot healing cycles but just spreadsheeted comparison of Heal per sec or Heal per resource is a bit extreme. As well as that "makes others cost more" was and is still just a misinterpretation plain and simple, an increase of heal/time with no increase of heal/resource means an increase of resource/time usage but it's not a "costs more" thing ... using a powerincrease / cast was just the wrong value to look at instead of the powerincrease / time, in which alacrity actually made the powerincrease coefficient stronger (made power worth more).

 

So yes, while the value of Alacrity to Surge was different pre-nerf, alacrity was, apparently, always worthwhile. And now we have more to back that up than just people yelling back and forth over differences in their feelings.

Don't get me wrong, but I have to say that your spreadsheets did cause problems in the other direction, people calling alacrity worthless due to them. Especially as people bought the reasoning of "alacrity makes things cost more" part too. All the base stuff like coefficients, stat DR results for power/crit/surge were nice. The alacrity part however went awry.

 

A 95% confidence interval for results of burst healing during gas/cells with HS/RS for crit/surge would be a nice addition to stats, as that needs some statistical computing. You should have compared burst healing ability from the get-go anyways, that is what I personally did and overall healing ability from added rapid shots was just gravy on top (the rapid shots catching up quite close overall healing) after running the output numbers for alacrity for burst vs other stats.

[Ewert]

This is all obviously post nerf of surge, that i am talking about.

Surge never was that hot to begin with. Lure of the big numbers ... It used to be somewhat more comparable to power and crit, but not good for healing due to uncontrollability, so the weakest of them even before nerf. Now it's just plain and simple bad straight up mathematically too without even needing to discount the randomness of it.

 

I crafted adrenals in early game like this: surge rocks! Oh man look at that DR, surge sucks. Crit rocks! Oh man look at that DR, crit sucks. Power rocks! =P Alacrity rocks more, but I am pretty fast already, so I'll skip doing that adrenal.

 

For the record, I should run normal op for alacrity adrenal and fortitude stim. =P A bit more burst (for dps mostly, healing is not a problem with 2 healers) and bit more hp for nightmare mode useful in my experience so far even with nearly full Rakata level gear ...

[Brad_Pitt]

The problem was, for quite a while there were two camps who were both operating on a combination of "feelings" and "napkin math." The one side said "things go faster." The other said "It doesn't do anything for half our abilities, and it makes others cost more."

 

The trick was to try and put numbers on it, and that takes more than just napkin math. But, of course, the more complicated the analysis, the greater room for error if an assumption is wrong.

 

This would all have been easier if I was a DPS Commando, since DPS is an obvious and easy metric. As a Combat Medic, I instead decided to address the Healer side first, which is far more complicated, to the extent that SimulationCraft, the very complex (and well respected) WoW simulator never fully realized a proper healing simulation. (Yes, I know they had a few, but I am actually in discussion with the author of said modules and he was never happy with the quality of those sims)

 

So yes, while the value of Alacrity to Surge was different pre-nerf, alacrity was, apparently, always worthwhile. And now we have more to back that up than just people yelling back and forth over differences in their feelings.

 

yeah, sorry i forgot to quote the guy above, it was directed at him

[RuQu]

Well tbh, calling something napkin math just because it's not a fancy spreadsheet of insertable values & spesific healbot healing cycles but just spreadsheeted comparison of Heal per sec or Heal per resource is a bit extreme. As well as that "makes others cost more" was and is still just a misinterpretation plain and simple, an increase of heal/time with no increase of heal/resource means an increase of resource/time usage but it's not a "costs more" thing ... using a powerincrease / cast was just the wrong value to look at instead of the powerincrease / time, in which alacrity actually made the powerincrease coefficient stronger (made power worth more).

 

"Napkin Math" wasn't an insult, nor do I tend to include the spreadsheet work in that term. It was directed at a lot of the math thrown around in post. It includes my early posts as well, and any short, shallow analysis, no matter how valid. RSc costs 25 Heat and takes 2 seconds, with a net cost of 15 Heat in max regen. Speed it up to 1.5s, and you will only regen 7.5, for a net cost of 17.5 Heat. Napkin math, but no less true because of the simplicity of it. What it doesn't do, however, is capture the larger picture that a more complete simulation does.

 

Lots of people have/had spreadsheets. DR and coefficients were known by most people having the discussion, yet no one talked about what the proper metrics were to compare. Burst vs Sustained vs Chain-casting were mentioned, but the results were always subjective or vague. "RSh makes up for the increased cost." "You can get an extra SCG faster." Etc. Lots of words, not a lot of support.

 

Don't get me wrong, but I have to say that your spreadsheets did cause problems in the other direction, people calling alacrity worthless due to them. Especially as people bought the reasoning of "alacrity makes things cost more" part too. All the base stuff like coefficients, stat DR results for power/crit/surge were nice. The alacrity part however went awry.

 

Not quite. The wrong metric was used, and wrong assumption of activity. In a chain casting situation, such as damage that outpaces your healing, the old analysis still proves true. If you can't pause between heals, you will be less efficient with Alacrity than without, and Commandos will be less efficient than Mercs because of their window differences.

 

It just so happens that we almost never have to heal like that in current content. With the upgraded simulator, once I have logs of the fights I can simulate the damage coming in properly and see if this looks like a potential issue in the new content.

 

And while the old result may have been wrong, and swayed some people against alacrity for current content, I am now correcting that. That's how the process works. I do wish there were a few more people doing competing work so we could compare results and check each other.

 

A 95% confidence interval for results of burst healing during gas/cells with HS/RS for crit/surge would be a nice addition to stats, as that needs some statistical computing. You should have compared burst healing ability from the get-go anyways, that is what I personally did and overall healing ability from added rapid shots was just gravy on top (the rapid shots catching up quite close overall healing) after running the output numbers for alacrity for burst vs other stats.

 

There is no RNG in Commando/Merc healing, so repeated runs always produce the same result, and there is no confidence interval. I am planning to implement using a confidence interval as an early breakpoint for the proc-based classes like Scoundrel healing and Gunnery DPS, so if someone says do 10000 iterations, it will break early when the 2sigma or 3sigma point is passed and save computation time. Not a top priority at the moment though.

 

How do you get meaningful results from a burst calculation with no damage simulation? How long do you burst? How high do you allow heat to get? How long do you give yourself to recover before needing to burst again? Different values for any of the above will give different results.

Edited March 11, 2012 by RuQu

[Brad_Pitt]

Not only that^

 

but most healers have 1 or 2 instant heals, that we can do and kite enemies away and get around a corner.

 

like, 3 enemies are on you, and you run around the corner while instant healing with kolto missle, using cure to remove effects, and using emergency scan, while ur still waiting for kolto missle to come back u cant heal unless u stop moving or use the power surge thing to have ur next move cast instant.

 

so the whole thing of having like 1 person always on you usually doesnt happen, people give up really easy once they see u are a healer i feel like.

 

kiting away enemies usually only 1 will follow me and once they realize im a healer they fully turn around and run away, and i chase them with with every instant move i can cast.

 

so i feel like having someone say they need burst healing calculated, i just feel like you should never let yourself get that low, you shouldnt need to burst heal.

 

like, save your shield for when ur at 30% hp, pop medpac, instant heal, kolto missle. and maybe healing scan then rapid scan, and u should be at 100% easily. i feel like chain casting almost never happens unless ur keeping sumone alive while 2-3 ppl hit them.

 

but just putting a kolto shell on them with the healing scan HoT will basically negate 1k dmg every 3 seconds, which for most classes they only hit like 3-4k with crits every 1.5 seconds.

 

 

 

I fully support you doing any and all calculations though

I just hope you're having fun with it, sometimes I get all into figuring out something for a game, and other times I realize it took me 4 hours of calculations and spreadsheet data to figure out something I could have googled lol.

[RuQu]

To be fair, Brad, PvP is another beast entirely, and too dynamic to ever really model effectively. Although just because I can't imagine a way doesn't mean it's impossible, simply that it is beyond me.

 

This game is young enough there aren't a lot of answers via Google yet. That's half the fun. There are a few spreadsheets about. A lot of people tend to make their own for their own use, and, sadly, there isn't a lot of discussion of the decision logic, metrics, and assumptions that go into them which leaves newcomers starting from scratch and a lot of people arguing past each other about results without checking if they are measuring the same (or the right) things.

[Ewert]

"Napkin Math" wasn't an insult, nor do I tend to include the spreadsheet work in that term. It was directed at a lot of the math thrown around in post. It includes my early posts as well, and any short, shallow analysis, no matter how valid. RSc costs 25 Heat and takes 2 seconds, with a net cost of 15 Heat in max regen. Speed it up to 1.5s, and you will only regen 7.5, for a net cost of 17.5 Heat. Napkin math, but no less true because of the simplicity of it. What it doesn't do, however, is capture the larger picture that a more complete simulation does.

Except you do it the "wrong" way instantly again, calculating a "netheat" and saying it's more costly to expend resources faster while also healing faster. It's a semi-useful value (healing/netcost), yes, but not the primary one to compare things with, HPS and HPR (resource) still are the primary ones.

Not quite. The wrong metric was used, and wrong assumption of activity. In a chain casting situation, such as damage that outpaces your healing, the old analysis still proves true. If you can't pause between heals, you will be less efficient with Alacrity than without, and Commandos will be less efficient than Mercs because of their window differences.

 

It just so happens that we almost never have to heal like that in current content. With the upgraded simulator, once I have logs of the fights I can simulate the damage coming in properly and see if this looks like a potential issue in the new content.

Yep wrong metric, it used to be something valid in "Other-Games" since they tended to have a limited mana pool that had to last a whole fight that sort of was a "more dmg than can be sustainably healed", with that manapool being the resource expended. Here we have sustained healing ability ad infinitum, and instead must counteract against bursts and have enough to prevent death-by-attrition (which, with 2 or 4 healers per op type, does not happen really).

 

There is no RNG in Commando/Merc healing, so repeated runs always produce the same result, and there is no confidence interval. I am planning to implement using a confidence interval as an early breakpoint for the proc-based classes like Scoundrel healing and Gunnery DPS, so if someone says do 10000 iterations, it will break early when the 2sigma or 3sigma point is passed and save computation time. Not a top priority at the moment though.

Ahh you missed the point of the confidence interval thing:

 

You do a burst cycle of maximum HS+RS during a SCG. Having a 40% crit chance and 90% surge may give anywhere from 8 crits to 0 crits. Having 95% CI would be helpful in knowing how low and how high it can be compared to having the increases in power or alacrity instead, which are static (relative) % increases.

How do you get meaningful results from a burst calculation with no damage simulation? How long do you burst? How high do you allow heat to get? How long do you give yourself to recover before needing to burst again? Different values for any of the above will give different results.

Well there are two simple options:

Burst from zero to 60 heat with TSO, SCG and vent heat between

Burst for the duration of SCG, heat ignored (for max resource efficient max HPS metric)

 

The thing is, one should not get stuck in trying to get a perfect value, but a useful metric. For bursting there is arguably a max of "from zero heat to 100 with TSO, SCG and Vent heat used pushing for max HPS". That's the absolute maximum, IMHO. If anything bursts bigger than that, you are wiping no questions asked. You are pretty certainly wiped even after that, could be useful for enraged boss for example. Pure HPS aimed for, not sustainability.

 

But I'd go with the "infight" bursts mentioned above. Of note is that both are alacrity "heavy". You could mix it up with "tank burst" and "raid burst" so that the tank burst has a kolto missile both at start of SCG and final before SCG ends, whereas raid burst does not due to the dmg being spread around.

[TempestasSilva]

alacrity was never worthless, it was always something thats in the bh medic gear, paired with power. sometimes its surge and crit though depending on the tier.

 

alacrity should have been always viewed as, saving global cooldown and faster casting time.

 

global cooldown is 1.5 seconds on instant moves. by adding alacirty and getting your 1.5 second activation moves down to say 1.3, everytime you cast moves that are lower than the 1.5 global cooldown, you are saving up the time towards using an instant that normally wouldnt be there or healing or damaging moves.

 

so for instance, healing scan 1.3(normally1.5 with tree), rapid scan 1.7s(normally 2.0 with tree), total time savings of .5s so far.

 

Well, duh. But you missed the point. Until RuQu, the Merc forums were filled with:

http://www.swtor.com/community/showthread.php?t=141187&highlight=alacrity

http://www.swtor.com/community/showthread.php?t=110100&highlight=alacrity

http://www.swtor.com/community/showthread.php?t=152351&highlight=alacrity

Etc.

 

Once RuQu came out with his initial work on alacrity, there was at least something worth discussing. Prior to that it was all feelings and flawed logic.

 

Like it or not RuQu's work brought people out who said hey wait a second, that doesn't seem quite right (only usually in a much ruder way) and the discussion has continued to evolve.

[RuQu]

Except you do it the "wrong" way instantly again, calculating a "netheat" and saying it's more costly to expend resources faster while also healing faster. It's a semi-useful value (healing/netcost), yes, but not the primary one to compare things with, HPS and HPR (resource) still are the primary ones.

 

These were the metrics used earliest in this debate. Admittedly these threads were on the Commando forum since that is my side for this AC, and you may not have seen them.

 

What is the HPR of Emergency Scan?

 

0 Alacrity: RSc/HSc/RSh has a net rotation cost of 0 Heat (25-5*5s). What is its HPR? For simplicity (napkin math time!) let's say that rotation healed for 5k, and it takes 5s so that's 1k HPS with infinite HPR.

 

20% Alacrity: Rotation now takes 4.3 seconds, and the rotation cost is now 3.5 Heat (25-5*4.3). HPR is now 1428.57 (down considerably, infinity %, from infinity) and the HPS has risen to 1162 (up 16.2% because RSh is instant and didn't speed up).

 

Yep wrong metric, it used to be something valid in "Other-Games" since they tended to have a limited mana pool that had to last a whole fight that sort of was a "more dmg than can be sustainably healed", with that manapool being the resource expended. Here we have sustained healing ability ad infinitum, and instead must counteract against bursts and have enough to prevent death-by-attrition (which, with 2 or 4 healers per op type, does not happen really).

 

I'm glad we are discussing proper metrics, but clearly it is HPR that is outdated. Other games lacked free heals as fundamental to their rotations, and, exactly as you said, HPR was key to keeping a fixed pool lasting a full fight. With resource neutral rotations and essentially infinite resources if managed properly, HPR fails to be useful, or even produce rational numbers. If you finish a 5 minute fight at 11/12 Ammo, did you heal 2 million damage for 1 Ammo?

 

HPS and HPCT also have issues because there are times when you consciously decide that the extremely weak, but extremely efficient, RSh is enough. It is the correct choice for the situation, but it causes both of those numbers to decrease due to its low HPS and low HPCT. You have to use a rotation that will correctly choose to use it, and a metric that doesn't drop because of this correct choice. This is where HPRCT came from.

 

Ahh you missed the point of the confidence interval thing:

 

You do a burst cycle of maximum HS+RS during a SCG. Having a 40% crit chance and 90% surge may give anywhere from 8 crits to 0 crits. Having 95% CI would be helpful in knowing how low and how high it can be compared to having the increases in power or alacrity instead, which are static (relative) % increases.

 

Well there are two simple options:

Burst from zero to 60 heat with TSO, SCG and vent heat between

Burst for the duration of SCG, heat ignored (for max resource efficient max HPS metric)

 

The thing is, one should not get stuck in trying to get a perfect value, but a useful metric. For bursting there is arguably a max of "from zero heat to 100 with TSO, SCG and Vent heat used pushing for max HPS". That's the absolute maximum, IMHO. If anything bursts bigger than that, you are wiping no questions asked. You are pretty certainly wiped even after that, could be useful for enraged boss for example. Pure HPS aimed for, not sustainability.

 

As you say, for max HPS you are spamming HSc and RSc. At 0 Alacrity, the net cost per pair is 7.5 Heat in max, but the first cast will cost 25 because you don't regen while full. Above 40 it will cost 14.5 per pair, above 80 it will cost 18.

1: 25

2: 32.5

3: 40

4: 54.5

5: 69

6: 83.5

7: HSc gets you to 96.5 and you can't cast RSc

 

At 20% Alacrity, the net cost in each bracket is 11, 16.6, and 19.4

1: 25

2: 36

3: 47

4: 63.6

5: 80.2

6: 99.6

 

So both get in a full 6 combos, but 0 Alacrity sneaks in an extra HSc (assuming it can with cooldowns). The question becomes if a 20% increase in HPS (all have cast times) is greater than the value of that 1 HSc. I think we can see that it is, as RSc has the larger HPS value and there are 6 of HSc/RSc combos.

 

But....this kind of analysis is very simplistic, and you would have to sub some of the later combos out for full price RSc since SCG just doesn't last that long.

 

I also don't think there was ever too much debate on this particular point, since the argument that alacrity didn't do anything for our instants doesn't factor into a burst discussion.

 

The problem with basing all of the decisions around burst or enrage timer burst is that these conditions just didn't happen that often. We aren't arcane mages, blowing all of our resources in a burst-recover rotation tied to Vent Heat. If you ever had to do this more than once per 2 minutes, you would wipe. It's something we would only ever see at enrage timers.

 

But I'd go with the "infight" bursts mentioned above. Of note is that both are alacrity "heavy". You could mix it up with "tank burst" and "raid burst" so that the tank burst has a kolto missile both at start of SCG and final before SCG ends, whereas raid burst does not due to the dmg being spread around.

 

Unsurprisingly, I think the answer is "more complicated sims!" Currently I am averaging over a range of incoming boss damages. I can add in an option to give a "content average" where it averages over some range of damage you should expect to see, and also a "specific DPS" where you can set it to one value only. I can add in an enrage timer, an enrage multiplier, and a box for tank HP, with a break statement if the tank dies.

 

Of course, we would still need to agree on metrics. I think, in that more complicated new sim, the #1 metric becomes "did the tank die?" Beyond that, I think HPRCT best captures the value of stats on healing, as it scales with all stats and doesn't drop when RSh is the right choice, and I think HPR is a meaningless metric in context of our mechanics.

[Ewert]

What is the HPR of Emergency Scan?

Well infinite of course... Basic math.

 

That's why one compares both HPS and HPR. For resource costing abilities you can build a combined value with rapid shots for heat neutrality and get a HPS value for that combination (I used that in my BG post). That could be used for a death-by-attrition comparison value.

0 Alacrity: RSc/HSc/RSh has a net rotation cost of 0 Heat (25-5*5s). What is its HPR? For simplicity (napkin math time!) let's say that rotation healed for 5k, and it takes 5s so that's 1k HPS with infinite HPR.

Of course you don't count net heat for resource cost. 5000/25=200 is the HPR but what you could use with that is the 5000/5=1000 "combined" HPS value since it is heat neutral.

20% Alacrity: Rotation now takes 4.3 seconds, and the rotation cost is now 3.5 Heat (25-5*4.3). HPR is now 1428.57 (down considerably, infinity %, from infinity) and the HPS has risen to 1162 (up 16.2% because RSh is instant and didn't speed up).

No, the HPR is still 200. You just used it faster, you are too stuck in thinking on terms of netheat per ability use. You need to break it apart for correct comparisons. To compare the "netheat" or rather the "combined" value, we add a 0.7 seconds worth of rapid shots for zeroing out heat use, then divide by used time. As an aside, we should be comparing Hs+Rsc as a baseline for the real HPR, since RSh is a free ability and should be skipped for the pure HPR value. Anyways, we add that extra .7 of RSh, take fictive heal split of 2600, 1800 and 600 for RSc, HS and RSh respectively since you didn't specify, and we get a Heat neutral combined HPS of 1056, up 5.6% from non hasted.

I'm glad we are discussing proper metrics, but clearly it is HPR that is outdated. Other games lacked free heals as fundamental to their rotations, and, exactly as you said, HPR was key to keeping a fixed pool lasting a full fight. With resource neutral rotations and essentially infinite resources if managed properly, HPR fails to be useful, or even produce rational numbers. If you finish a 5 minute fight at 11/12 Ammo, did you heal 2 million damage for 1 Ammo?

5*60*5 heat used total if we take a merc and stayed under 40 at all times. Unlearn the fixation on netheat, resources ARE expended even if end result is 0, 12 or whatever happens to be the "full" value. HPR is just perfect for comparing the resource costing ability priority of use. Just skip the netheat fixation, use the "filled with RSh to heat neutrality" if you wish to examine the overall continuous HPR value.

HPS and HPCT also have issues because there are times when you consciously decide that the extremely weak, but extremely efficient, RSh is enough. It is the correct choice for the situation, but it causes both of those numbers to decrease due to its low HPS and low HPCT. You have to use a rotation that will correctly choose to use it, and a metric that doesn't drop because of this correct choice. This is where HPRCT came from.

I reckon that is something akin to my use of "heat neutral combined with RSh" value. My method is crude, including KS and ES would make the decrease smaller for the HPS, but too bored to do a 1 minute cycle for it. However for ingame results a full cycle is somewhat unnecessary. Knowing the relative values for each heal, using a heatneutral value to approximate non-burst situation usefulness and strength instead of pure HPS to factor in alacrity for overall healing, and using HPR for resource use priority and HPS for burst priority, that is already plenty enough info to learn healing skill use priorities.

As you say, for max HPS you are spamming HSc and RSc. At 0 Alacrity, the net cost per pair is 7.5 Heat in max, but the first cast will cost 25 because you don't regen while full. Above 40 it will cost 14.5 per pair, above 80 it will cost 18.

1: 25

2: 32.5

3: 40

4: 54.5

5: 69

6: 83.5

7: HSc gets you to 96.5 and you can't cast RSc

 

At 20% Alacrity, the net cost in each bracket is 11, 16.6, and 19.4

1: 25

2: 36

3: 47

4: 63.6

5: 80.2

6: 99.6

 

So both get in a full 6 combos, but 0 Alacrity sneaks in an extra HSc (assuming it can with cooldowns). The question becomes if a 20% increase in HPS (all have cast times) is greater than the value of that 1 HSc. I think we can see that it is, as RSc has the larger HPS value and there are 6 of HSc/RSc combos.

Well you didn't pick either of the example situations I gave... If we take the 13sec gas, you can get off 5 pairs with 0% alacrity with the last 3 casts not benefitting from gas and total time of 17.5secs. Also naturally you start regen after HS already. So 16 at 1.5, 15 at 3.5, and then another 7.5 pair at full regen to 7sec, still 2 more pairs at full regen, only the last will push you over 40. So definitely a doable recoverable burst, for 5.3RSc and 5.4HS.

 

For 20% alacrity, we get 6.4 RSc and 6.5 HS off in 16.8 seconds. Not only did we do 20.75% more RSc and 20.37% more HS healing, we did it in less time, we actually exceeded the expected 25% output increase slightly due to the extra scans hitting during gas. But approximately the inverse value meaning output value of alacrity shows the burst increase (1/.8=1.25). Yes we used more resources, but this was about death-by-burst. If those last heals are needed too, there is a high chance that the 0% alacrity one would even allow a death, or at the minimum he would have to throw unbuffed RScans and overheat badly.

I also don't think there was ever too much debate on this particular point, since the argument that alacrity didn't do anything for our instants doesn't factor into a burst discussion.

Hence the duality of examining healing results: heatneutral combined HPS for death-by-attrition prevention and pure burst for examining death-by-burst prevention.

The problem with basing all of the decisions around burst or enrage timer burst is that these conditions just didn't happen that often. We aren't arcane mages, blowing all of our resources in a burst-recover rotation tied to Vent Heat. If you ever had to do this more than once per 2 minutes, you would wipe. It's something we would only ever see at enrage timers.

Well actually it is quite common while learning encounters when group members do Stupid-Things.

Of course, we would still need to agree on metrics. I think, in that more complicated new sim, the #1 metric becomes "did the tank die?" Beyond that, I think HPRCT best captures the value of stats on healing, as it scales with all stats and doesn't drop when RSh is the right choice, and I think HPR is a meaningless metric in context of our mechanics.

On my tablet so bit slow (on a trip) so not easy to check your Hprct thingy-of-majick.

 

If I were to do some pro spreadsheeting, I'd do two burst value sheets (tank one with gas and KM at start and end of gas ending at under 60 heat, and raid one with no KM otherwise same), and then an infinite <40 cycle with gas used at 20-25 heat and KM used at start of gas. No TSO no Vent heat, leaving those out since while they change the metrics some, one should never use them on CD, nor can one ever count on them being up for a burst.

[RuQu]

 

Well actually it is quite common while learning encounters when group members do Stupid-Things.

 

And those people die, and wipes happen, until they stop doing stupid things. Once geared, you can heal through the stupid.

 

Lots of talking around each other.

 

Before we spend hours going back and forth past each other, we should probably settle on some initial common ground.

 

Clearly you use spreadsheets to calculate short rotations. Are you opposed to full sized simulation of incoming damage, tank death, on the fly ability selection? If so, why? (and note that this may be a fundamental philosophical difference which would largely render any further argument a bit useless).

 

Assuming you are open to the idea of sims:

 

If simulating encounters, do you agree that so long as resources are tracked properly, and a proper ability decision logic is in place, that the exact HPR cost of abilities is irrelevant?

 

Do you agree that the incoming boss damage will not increase as gear improves (and in fact will likely decrease as tank gear improves), and therefore, for a fixed time length, the Total Healing Needing is constant, and HPS is not an ideal metric for comparing change?

 

Do you agree that, for classes like ours, free heals are often used when it is not necessary to spend resources, and that this decreases the HPCT, but that the decrease is intentional and proper and should not be reflected in the results, and, therefore the cast time of these "luxury" free heals should not be counted? (HPRCT)

 

If we can agree on the first two questions, then we are simply discussing what the priority rotations should be, and debating if the models/sims are complex enough. The last two questions become about debating the metric for said sims.

[Brad_Pitt]

Well, duh. But you missed the point. Until RuQu, the Merc forums were filled with:

http://www.swtor.com/community/showthread.php?t=141187&highlight=alacrity

http://www.swtor.com/community/showthread.php?t=110100&highlight=alacrity

http://www.swtor.com/community/showthread.php?t=152351&highlight=alacrity

Etc.

 

Once RuQu came out with his initial work on alacrity, there was at least something worth discussing. Prior to that it was all feelings and flawed logic.

 

Like it or not RuQu's work brought people out who said hey wait a second, that doesn't seem quite right (only usually in a much ruder way) and the discussion has continued to evolve.

I agree RoQu has certainly helped. I didnt really post on forums int he beginning at all as i didnt know much. I find myself posting on these because I canceled my sub because i cant stand huttball anymore being 63 valor, i have everything i want/need, bought out the suppliers on the Market, gonna resub when 1.2 is out.

 

 

I told everyone on my server its all about power and alacrity because I made a chart a long time ago, basically 3 weeks after the game came out, i was level 40 ish, and i figured out you wanted alacrity if you wanna dish out extra dps as a certain thing, like adding power vs crit and surge.

 

adding crit and surge is adding a chance to deal more dmg, adding power and alacrity is a sure thing and the results are instant. you can really only add on to yourself about 16% crit i believe it is through wearing full crit armor set up.

 

taking a simple multiplication of an extra 16% critting with a multiplier of 70% says that

 

1800(tracer missle average hit for me).16*.7 = 201.6dmg

 

says you will do an extra on average with what you added in 201.6 dmg

 

and using power + alacrity would be a much higher number, as adding as much to power as it would take to get 16% extra crit is like adding on 550 power, which is a huge boost, about 120 extra dmg onto that tracer missle, plus the alacrity that stacks with it, having a 20% activation lowering will easily push you over the 201.6dmg that crit and surge would have provided. plus you are forgetting that you still also have your 50% base multiplier and your crit base and your skill tree crit. I havent met anyone yet that has only their 50% multiplier, usually i see 60% upwards with awesome alacrity builds.

I used to be 42% crit with 78% multiplier, pre nerf.

 

Now I am 37% crit with 70% multiplier, 15% activation, 300 ish power added on.

that will out hit 42% crit/78% multiplier with having no activation bonus and no power any day, even before nerf.

 

 

People always forget you're only adding on 16% ish crit with the full armor set, and the surge multiplier pre nerf was around 99% if you had the best gear.

 

now once you add more than 250 surge it hits a huge drop, its best to add about 200-220 if you can. then use your skill tree for the activation time and alacrity bonuses.

 

 

now for higher hitting moves, crit and surge obviously is better because a multiplier of .16*.7 *( a high hitting assassin attack that say is 2500 base) will be better because alacirty is wasted.

 

its just on activation and channeled stuff that alacrity is going to be better on, and it really only helps being level 40+.

 

when i am a fully specced dps merc, i easily get 500k dmg per round, and thats with dying like 10 times because im going in and fighting in the middle of it or kiting people around. if i always stood back like most ppl do and didntt ake dmg, which i sometimes do, i have achieved over 650k dmg and like 50k healing.

Edited March 12, 2012 by Brad_Pitt

[Ewert]

And those people die, and wipes happen, until they stop doing stupid things. Once geared, you can heal through the stupid.

No, those people cause burst healing needs, which you may be able to heal through if you have high burst capabilities, as in alacrity. Nearly all boss mechanics have survivable screw-up dmg. Worst are Boney and Soa, but those are also the ones with possibly the least overall healing required and thus the most time to cool off usually.

Before we spend hours going back and forth past each other, we should probably settle on some initial common ground.

 

Clearly you use spreadsheets to calculate short rotations. Are you opposed to full sized simulation of incoming damage, tank death, on the fly ability selection? If so, why? (and note that this may be a fundamental philosophical difference which would largely render any further argument a bit useless).

Have you stopped beating your wife yet? About as good a question... Do whatever you feel like doing. Is it useful, that would be a better question. Sure, for fun factor, but honestly a sub40 heat infinite cycle simulation (since alacrity will change the cycle a sim would ease the examination) would answer best the theoretical max sustained overall healing value question (for alacrity vs power and the rest mostly, to get a solid grip on the overall healing loss). And a max burst cycle is stupidly easy to compare and needs no simulation. Anything beyond those two will go into theoryfantasyland, giving no real concrete extra info.

Assuming you are open to the idea of sims:

 

If simulating encounters, do you agree that so long as resources are tracked properly, and a proper ability decision logic is in place, that the exact HPR cost of abilities is irrelevant?

You are logically backwards there. HPR and HPS are the relevant values for those "ability decision logic"s, lol. In any case, for the infinite cycle sim, as the whole premise is heat neutrality, of course HPR is relevant (for the ability choice logic) but after that logic is coded in the sim then it is all about HPS, and what results each stat gives unto it.

 

Do you agree that the incoming boss damage will not increase as gear improves (and in fact will likely decrease as tank gear improves), and therefore, for a fixed time length, the Total Healing Needing is constant, and HPS is not an ideal metric for comparing change?

If you do that infinite cycle sim then of course the HPS is the #1 thing to compare. Btw, one thing occurred to me by this: the better your raid gets, the better alacrity gets. As yes, the overall healing reqs go down, and all you really have to do is burst heal the occasional bad luck rng or inattention dmg or unavoidable burst dmg...

 

Do you agree that, for classes like ours, free heals are often used when it is not necessary to spend resources, and that this decreases the HPCT, but that the decrease is intentional and proper and should not be reflected in the results, and, therefore the cast time of these "luxury" free heals should not be counted? (HPRCT)

Honestly, those terms are silly. Yes I am blunt. As well as using them for stat value calculations would be doubly so, silly not blunt. Logical failures, I'd say. Two prefixes for HPS is all it takes to get explicit and useful metrics: burst HPS and sustained HPS. Examining anything less than max sustainable HPS is silly because you just do NOT choose your gear for maximizing your idle time. So a sim for that would be a waste of time. A sim for sustained HPS would be great though, as I said earlier in this post. Examining burst however is easy and needs no sim, as shown earlier in the part you so fast dismissed.

 

If we can agree on the first two questions, then we are simply discussing what the priority rotations should be, and debating if the models/sims are complex enough. The last two questions become about debating the metric for said sims.

While fun, a complex sim would not really be accurate and thus not the sim with which to examine stat choices with. It could give an inkling of extra info on where to compromise with alacrity vs power (+crit & surge), but as I earlier said, even that point of compromise changes depending on the gear level and skills of the raid.

 

Due to my gear and raid environment I kinda am past the point of usefulness for the overall sustained HPS value sim, I just want more alacrity and endurance, but for those still progressing it would show some true numbers for stat choices.

[RuQu]

Have you stopped beating your wife yet? About as good a question... Do whatever you feel like doing. Is it useful, that would be a better question. Sure, for fun factor, but honestly a sub40 heat infinite cycle simulation (since alacrity will change the cycle a sim would ease the examination) would answer best the theoretical max sustained overall healing value question (for alacrity vs power and the rest mostly, to get a solid grip on the overall healing loss). And a max burst cycle is stupidly easy to compare and needs no simulation. Anything beyond those two will go into theoryfantasyland, giving no real concrete extra info.

 

Simple models give simple answers to simple questions. If you have a simple question for yourself, build a simple model. If you want to explore more, they get complicated fast.

 

My work started out as a simple spreadsheet to answer a simple question. The answer raised more questions, which needed more complex models.

 

There are gains to be had there.

 

You are logically backwards there. HPR and HPS are the relevant values for those "ability decision logic"s, lol. In any case, for the infinite cycle sim, as the whole premise is heat neutrality, of course HPR is relevant (for the ability choice logic) but after that logic is coded in the sim then it is all about HPS, and what results each stat gives unto it.

 

You are stating the trivially obvious as though it was profound. The priority list is made from a combination of ability HPR and ability HPS.

 

What is the purpose of computing the HPR for the rotation, especially if you intentionally ignore the net cost. By ignoring net cost, which you casually dismiss with no valid reasoning that I've yet seen, you are massively biasing your results. You are holding the HPR constant for a given ability while increasing the HPS, when in fact the HPR is decreasing per ability with an offset for any healing per resource gained using RSh.

 

If you do that infinite cycle sim then of course the HPS is the #1 thing to compare. Btw, one thing occurred to me by this: the better your raid gets, the better alacrity gets. As yes, the overall healing reqs go down, and all you really have to do is burst heal the occasional bad luck rng or inattention dmg or unavoidable burst dmg...

 

Honestly, those terms are silly. Yes I am blunt. As well as using them for stat value calculations would be doubly so, silly not blunt. Logical failures, I'd say. Two prefixes for HPS is all it takes to get explicit and useful metrics: burst HPS and sustained HPS. Examining anything less than max sustainable HPS is silly because you just do NOT choose your gear for maximizing your idle time. So a sim for that would be a waste of time. A sim for sustained HPS would be great though, as I said earlier in this post. Examining burst however is easy and needs no sim, as shown earlier in the part you so fast dismissed.

 

My early models all used HPS. First sustained only, then sustained and burst. The answers did not come out well for Alacrity.

 

Before you say it, no I did not hard-code the Effective Cost change. I modeled the regen, and cast abilities. Effective Cost is a real effect, whether or not you choose to ignore it, and the results of that pure HPS rotation showed it. The effect was most noticeable for sustained, and mostly went away for burst, but any burst simulation requires that the player have enough downtime to recover between burst. This is not always true, so general recommendations have to factor in both sustained and burst...and again sustained showed alacrity as having strong negative trade-offs.

 

Of course, that's why I moved away from HPS.

 

While fun, a complex sim would not really be accurate and thus not the sim with which to examine stat choices with. It could give an inkling of extra info on where to compromise with alacrity vs power (+crit & surge), but as I earlier said, even that point of compromise changes depending on the gear level and skills of the raid.

 

Due to my gear and raid environment I kinda am past the point of usefulness for the overall sustained HPS value sim, I just want more alacrity and endurance, but for those still progressing it would show some true numbers for stat choices.

 

A complex sim is worse for examining stat choices than an overly simple one? Not sure I follow that logic inversion.

 

It seems to me that the problem lies in your denial of the existence of Effective Cost increases of abilities, compounded by the fact that at the level of content you are running the result, though arrived at incorrectly, is still correct. At your gear level, Alacrity is a great stat...just not for the reasons shown by your calculations. This leaves you in the position of having the right answer, backed up by it working well in-game, and a simple sim that shows the right answer, and then trying to justify the flawed assumptions of the sim based on the fact that it still arrived at the correct answer while missing how wrong that answer is for other people in other situations. A stopped clock might give the right time in NY, but that time is wrong in LA.

[Yazule]

The only thing I don't like about RuQu is he is a Republican... feel the hate flow through you, come to the dark side.

[RuQu]

The only thing I don't like about RuQu is he is a Republican... feel the hate flow through you, come to the dark side.

 

I'm planning to level an Operative next. Does that help?

[Yazule]

I'm planning to level an Operative next. Does that help?

 

i have a guild for you on Hedarr Soongh if interested