Could /Regen use something else?

The issue isn't with 180 and 1800 second calculations being meaningless. They are valid within specific contexts, but their importance is overvalued outside of that context. You can claim that the 1800 seconds of constant fighting is representative of running a mission, but it doesn't consider the fact that, with mitigation sets, you are fully capable of stopping when you are at low health in order to recover and functionally reset your timer. You even made the point that, if a player isn't stopping to use Rest every time it is recharged, they aren't being as efficient as they are capable of being (which is the entire point of running through a mission as a single combat rather than turning it into a series of separate combats). Since mitigation sets are not forced to run through continuous attrition combat, you can't say that a long constant period of fighting is valid for those sets because it isn't representative.

Unless you are running an explicit farm mission (i.e. one designed to minimize time between combats to maximize rewards), you're going to have, at a minimum, 5-10 seconds between each fight simply due to navigating the mission itself, which does virtually nothing for */Regen (since you're pretty much guaranteed to be at full health at the end of any fight) while it appreciably increases the survivability of sets like */SR and */Invuln because they are assumed to enter into subsequent combats at below full resources. Similarly, because each discrete fight exists in shorter time frames, the relative risk for mitigation based sets is lower than it is for recovery based sets because they have substantially higher performance within those short time frames. Combine the two issues by using longer time frames that assume a constant state of combat, and */Regen looks a lot better than it should and mitigation based sets look a lot worse.

The entire question is what the defined playstyle is assumed to be: are players assumed to run directly from one fight to the next with an absolute minimum downtime between each (creating a situation wherein combat is virtually ongoing for the entire duration of the mission) or is the mission separated into a series of discrete shorter periods of combat punctuated by variable periods of recovery (from travel time from one combat to the next or simply using Rest)? Unless you're going to assume that farming (downtime negligible constant combat) is now the standard of solo mission running (which wouldn't mesh with the forced requirements of only using average player performance the standard of balance), a long period of constant combat isn't going to be a legitimate model of actual mission performance. It would be more accurate to look at missions as a series of short combat periods rather than a single long period explicitly because of this.

Even on non-farm PUGs, most teams barely stop. Sure 5 or 10 seconds often exists between fights, but I completely disagree that those 5 or 10 seconds are less valuable to Regen than to Invuln or SR. IME, those 5 or 10 seconds are more valuable to a Regen. Sure, not all the time, but more often than not. It is not my experience that I am always at full by the end of a fight, even on any of my regens and in that time between spawns the regens easily heal back more than my non-regen scrappers.

Of course, I can choose to stop and rest when my Invuln gets to 20% health. But how can you think that is not a disadvantage? If I must stop and another scrapper can go on, that is an advantage for them. Sure, I likely can take more punishment in the short period that exists in a fight (that would be an advantage for Invuln), but if we can both survive, then a Regen is going to need to slow down less. Only if we reach a point where the Regen will fall over (or fall back), does the Invuln advantage matter and even then, we still have to count the resting time against the Invuln vs. the kiting time of the Regen (or the time to get back up and XP lost to debt).

And teams complicate everything. When talking about this, it works best to simply assume that the individual in question is soloing because otherwise you have to deal with the vast plethora of buffs and the question of how valuable survivability is when you're not being attacked by everything in question.

As to the usefulness of the recovery time frame, numerically, a mitigation based set is going to get more real, applicable benefit out of a period of non-combat than a damage recovery based set will simply because any specific quantity of hit points is going to be worth more to the mitigation set (thanks explicitly to how mitigation functionally multiplies remaining hit points) than it would be to the recovery set. It's not a question of how many hit points you heal during the downtime. It's a question of how much that downtime increases your comparative survivability for the next fight. For */Regen, it's a virtually negligible improvement because the set itself spends a vast majority of its time at full health thanks to all of the damage recovery it packs in.

Well, then, you must assume you are simply running along from spawn to spawn. Any downtime solo is immediately no XP gained. If you are going to be adding 10 seconds for Invuln to rest, then they are going to start losing right away.

Sure, any set can slow down and handle individual fights great. I mean I run in and kill all five enemies on my FF defender and get taken to half health, then wait 20 seconds before engaging the next spawn. Obviously, a regen scrapper is no better off than me, because my ability to avoid damage is actually far better over those 20 seconds than a regen scrapper could hop to achieve. Those same 5 enemies will easily deal more damage to my regen scrapper. The fact that the damage will be healed back is, I guess, not very relevant to you. It is relevant to my experiences and to how I see most people want to play. Isn't this type of thing why Fitness is being made inherent?

I am pretty sure this paragraph is pretty silly, but I am tired and about to go to bed. I'll try to wrap my head around it tomorrow.

Darn you Arcana !!

Now you have given me more to consider for my Defender primary comparisons. LOL

My current work is geared toward a 30 second comparison, which I have always viewed as a limit rather than a sustainable value. But I also have gone ahead and done the same work for 180 seconds as (what I felt initially) being a more sustainable amount of incoming damage. But mostly because there are alot of powers which do not work very well with a 30 second window.

Now, thanks to you, I can clearly see that the whole time-sustainable damage picture will be a curve with zero seconds being the true absolute limit, but changing across each arbitrary time-frame chosen. I instinctively knew this, but looks like I am gonna need to rework some things and actually create graphs. I work in electronics and we do some testing of circuit boards using oscillascopes, and what we look for in these models are plataeus on the graphical representation of the data. I "suspect" that creating these same types of graphs using models like your 30sec/60sec time-frames, but across every interval from 1 to 180 secs will not have very distinctive "plataeus", so what are your feelings about how to analyse "slope" changes over time (in this context) ?

Thank goodness for Excel, cause this is fast approaching my math limits, hehe.

P.S. I take my statement back. It looks like there will definitely be a distinctive slope change (plataeu) when you graph each time-interval as its own point. The following are two representations of a hero on baseline regen 0.42hps shown. The first chart shows Y as DPS (%total health/sec) and X being the time to defeat in 10 sec intervals. As you approach 180 secs, you see that it is getting close to the immortality value for baseline regen, while at the front end it would approach 100% damage/sec for defeat (pretty obvious that if you sustain 100% of your hps before even the first regen tick, you will be defeated).

http://img43.imageshack.us/img43/5013/rxclp5.png

http://img829.imageshack.us/img829/3572/rxclp6.png

The second chart is just a representation of the change in each interval relative to the previous point.

OK, with fresh eyes, I still think the above is silly.

You start off pontificating that sets other than Regen are better off not fighting stuff. More seriously, your claim is that other sets can better use downtime, because the hit points they regen are more valuable layered with their defense and resistance. You then claim that Regen does not benefit from downtime, because they do not need it.

1) Regen needing less downtime is an argument in Regen's favor.
2) Regen does not need as much downtime, but I disagree with your premise that they are likely at full health by the end of most fights. I think the travel time between spawns is actually quite helpful, quite often. Regen would suffer a lot more if you just kept dumping spawns on them without that occasional 5 to 10 seconds of zero incoming damage.
3) Those extra hitpoints meaning more layered with defense and resistance is only valuable if it is likely any singue spawn is going to kill you. I don't think most people play at a point where one spawn is likely to be deadly. Sometimes adds happen, but usually not very often. Regen is well built to handle adds thanks to multiple clicks.

When the first sentence of your paragraph states that other sets get great benefit from downtime and the last sentence of your paragraph states that Regen almost never has downtime because it is almost always at full health, this does not seem like a paragraph that supports the premise that Regen is lacking compared to other sets.

Not only is it silly, its inexplicable. Everyone who plays regen, even people who think it sucks, knows time is on regen's side. Its the basis of the phrase "you look good until you're dead" that regen has been identified with since practically release.

To put a more specific numerical spin on that statement, normalized recovery is the key here. If the mitigation set has 50% mitigation and 100% recovery and the recovery set has 300% recovery, its recovery will, point for point, be better. Conversely, if the recovery set has 180% recovery, the mitigation set's recovery will be, point for point, better.

As regen's regeneration + healing recovery rate with just recon, FH, and Int hovers around 800%, it basically always wins that race against other melee defense sets at standard SO slotting levels. The only time it loses that fight is when you start to consider things with slotted health and above 78% mitigation, like soft-capped SR scrappers with slotted health or aid self.

In order for the numbers to have meaning, so your analyses don't go flying off the rails, your analysis has to be foundationally solid and you have to be able to make contact with reality, which means it helps to actually know what happens in the game so you know what reality actually is. Its usually pretty obvious when an analysis lacks one or the other. Self-contradiction is usually a smoking gun for both.

The limiting case is to look at individual attacks. To survive a fight, at every moment in time you need to have two things:

1. Enough health + damage mitigation to survive the next attack that lands.

2. Enough recovery to get enough health back to make #1 true when the next attack lands.

That's the ultimate interplay between mitigation (which is really just a way to amplify the value of your health bar, which is the true "burst damage" buffer) and recovery. All of our other calculations are ways to simplify that into approximations. We could look at a fight with one spawn as in effect being approximated by having all the damage of that fight happening at the very beginning, followed by a period of recovery that lasts as long as the fight lasts, to see if you would have enough health at the end of the fight to fight another one, or if you would need to rest (which would, in effect, make the fight effectively longer even though you aren't technically attacking during the last part). This saves us from looking at individual swings.

Or we could look at a fight as all the damage being averaged out over its duration along with recovery. If we *know* we're looking at cases where the burst damage won't kill the target we're looking at, this is a safe assumption. If we're not sure because we're looking at high damage cases where that assumption is unsafe, then this may generate false results. Conversely, the assumption above is far harsher than reality.

One method to attempt to split the difference is to assume the incoming damage is a triangle, high to start and drops to zero during the fight - presumably as you defeat things. It approximates reality a little better if you're looking at single-target dominated situations. Conversely, in heavy AoE situations, the average damage case might actually reflect reality better.

Bottom line, though, is that combat is about taking damage without dying, and then recovering that damage so we can take another. The "perfect" mathematical solution is to look at individual attacks and healing pulses in something like a Monte Carlo Markov. Too much work for most people. I constructed one back when another poster suggested it, but it ended up being not worth it for the computational head ache. The alternative is to look for simpler proxies. Short time window + Long time window does a decent job of that, and absolute Monte Carlo simulations back them up, and thus its what I use.

The curves you describe are, however, the true survivability "value" of the mitigation being discussed. High mitigation sets of comparable value will drop below a mitigation set to the right of your curve, but jump higher to the left of your curve. Its kind of a hard thing to communicate in text though, and its hard to come up with an intuitive way of judging two such curves superimposed to decide which one is the "better" one. That gets into the question of what player(s) think are important.

From a balance perspective, those curves usually tell you less who is stronger than who, and more who is not unambiguously better than who. If one curve is always higher than another at all points its obviously better. If its lower always its obviously worse. If they cross somewhere in the middle, its more ambiguous, and suggests balancing them requires a more qualitative approach.

Thanks for the detailed response Arcana,

The charts I gave were just thrown together to show the shape of the data.
One thing that I am currently working on is the creation of counter curves that represent something more like your "sawtooth" idea. I want to vary the amounts based on a unit of incoming damage we can call "One Minion". These counter arcs would intersect the mitigation curves at different points. By seeing how each set performs against 3/6/9 minion arcs, we attach a time value to each powerset.

I have been messing around in game to try and average what damage and recharge values to assign for a "One Minion" unit. Recharge is fairly variable (4-12 seconds) across villain types, but I feel pretty good about using an 8 second recharge. Damage between Ranged and Melee is just as wild, but I feel pretty good about a Ranged/Melee ratio of 66%/100%.

Any thoughts on what a "One Minion" metric would be ?
I am having trouble finding where you broke this down in your comparisons.

EDIT: Ok I found it. Could have sworn you had more info explaining how you arrived at 4.47%/sec for three minions. Is it safe to assume this value is entirely related to scrapper hp values?

Yeah, its based on an estimate of about 20 dps for a standard minion at level 50. That itself was based on a number of tests of various minions to try to come up with an average, and the average settled somewhere close to the damage output of a malta swinging punches in melee range, which ended up being very roughly 20dps before accuracy is counted (~100 damage attacks with 5 second cycle time) at level 50. Since then, however, critter groups have appeared that have higher damage output than that, in particular Cimerorans, Vanguard, and Praetorians, that likely shift the average upward. And some critters, like Carnies, are likely shooting slightly faster in melee range than they used to, because of a circa I17 change to default AI that makes them more likely to use all their attacks rather than sit on some for a few seconds at a time. At high levels, especially at level 50, the average might be as high as 25-30dps now. I would need to investigate further to come up with a new average, but that will have to wait a bit until I finish a different project I'm working on now.

Yeah my gut tells me that 20 dps sounds like a low figure. But we are talking even con minions here. LoL

Most people wouldn't believe this, but in I5 (post GDN) three even Malta minions using nothing but that melee brawl attack could kill an SR scrapper with fully slotted defenses if you parked in front of them on a time scale of about a minute or two. You'd think a scrapper would face zero threat from three even minions at level 50, but in fact its the fact that offenses defeats them so fast that is as much, if not more responsible for scrapper survivability under those circumstances (that, and mez protection).

They aren't a threat because we kill them fast, and because there's enough rest in between spawns even when you're rushing through a mission to make what little damage they deal irrelevant. But their damage level is not negligible. We're just used to scrappers facing off against so much higher odds, it seems counter-intuitive.