Arcanaville

I'm aware of several out of whatever, but I'm not outing anyone on my own under any conditions whatsoever.

In fact I should say I'm only kind of making fun of BaB in my previous post because I've heard from him since his departure and he personally knows my sense of humor. I would otherwise never joke about that (also, I know BaB is following along but can't publicly respond, which means after three years I win, har har, hardy har har).

But while I know I can't (and frankly don't want to) eliminate speculation on who is and is no longer at PS, I would recommend keeping the speculation on the why to a minimum out of respect for the people no longer there. If they want us to know why, in 2010 they all have ways of informing us.

On the other hand, he kept telling me how the animation system made perfect sense to him, which suggests it might have been good to get out while he still had a functioning cerebral cortex. I pity the new animation guy, really I do.

Oh my god: I'm actually technically the most veteran person left with knowledge of the animation system.



BhhaaaaaaaaaB!

Short answer: yes.

Long answer: read the original arcanatime analysis which explains the thought process behind these measurements.

To be honest, I hadn't heard from BaB in a while. I had assumed he was working on something he couldn't chat about and respected that decision. So to be honest this is all news to me as well, and I'm not going to speculate aloud beyond that.

Actually, this is the one aspect of "ArcanaTime" that none of us ever fully resolved and I didn't do extensive enough testing to be able to conclude one way or the other. So maybe its time we just put it to bed once and for all.

I've thought up a series of tests to perform, and I'm going to try to do them a few at a time before I go to bed each night, which is the only time I have to do them right now (work's keeping me pretty busy these days). I will try to post what I figure out as I figure it out in this thread (as opposed to starting a new one). If I come up with any startling revelations, I'll write up a big wordy article eventually. But if I'm just confirming one of the prevailing theories, here's as good a place as any.

Going to try to set up a specific set of tests to determine this once and for all. Going to start by looking for worst-case scenario cast time powers and see what their cycle times are, and look at attack chaining from there (worst-case scenario cast time powers are powers that have the largest discrepancy between cast time and arcanatime, which should amplify any discrepancy between recharge from cast and recharge from arcanatime).

Hopefully I'll have an answer this week sometime, although it might take to the weekend to execute all of them with high precision.

I can't explain. Most of my tests have been consistent with recharge starting after cast time expiration not arcanatime expiration, including attack looping (using the same attack over and over again and measuring total cycle time). I'll have to investigate further (although I'll continue the discussion in the other thread where it makes more sense to do so).

Actually, Regen gets damage mitigation from regeneration, heals, and +health. If you count +health and +res together, at standard levels of difficulty regeneration gets about equal amounts of protection from healing and regeneration, and slightly less from resistive/health mitigation.

As to its performance, the only argument on regen are at high levels of performance with high levels of player skill. In the hands of average players playing at average levels of difficulty, I don't think the devs see any difference in performance across the playerbase for Regen relative to other scrapper sets. If they did, they would probably act upon them. In the past, though, its always been implied that SR (prior to its buffs) and Invuln (to an extent) were the lesser performing sets. That past includes the recent enough past for regeneration debuffs to be fully in play.

At high levels of performance, there is a potential argument to be made but at those levels of performance balance arguments typically have to be pretty sophisticated to be convincing, of a kind I don't generally see being made. Or at all, really.

The repost of the original article on "ArcanaTime" is here.

The reason why Real Numbers doesn't report this time is, besides the fact that Real Numbers was coded before the invention of Arcanatime, the real cast time is also important to know. Arcanatime is only strictly speaking useful for attack chain calculations. But when it comes to things like computing recharge, its still the case (so far as we know) that powers begin recharging after their *cast time* expires, not when arcanatime expires. Especially because Arcanatime isn't a true time window that can "expire" but rather a measure of how fast the game servers can process attack commands from the players even if lag is reduced to zero (which it is if you queue attacks fast enough - the power gets queued on the server and thus no network lag is experienced between one power activation and the next).

I've tended to have to install Mids on Win7 with "Run as Administrator" to get it to work. Don't know why, but its not like Mids is the only program to fall into that trap (for example VMware Workstation still has the same nasty bug where if you try to update it from within the program it fails in the middle with the same permissions problem and you end up with it being uninstalled and unrunnable until you reinstall manually - unless you're in the habit of running VMware as Administrator in the first place).


Also, if you need the PvP DR equations, you know where to find me.

Well, lets start here. Suppose I slot 3 level 50 damage IOs into Inferno. That would be 42.4 * 3 = +127.2% damage buff before ED, and +99.08% under ED. There are invention sets that can go higher than that: if you slot everything but the proc in Armageddon you end up with +101.86% damage after ED. Lets call it +99% for now.

Typically Blasters can use both BU and Aim (if they have BU) which is +100% and +62.5% respectively. The total damage strength is now 361.5%.

Invention set bonuses can easily get you +15% to +25% damage buff, sometimes more, but lets just say +15% for now. That takes us to 376.5%.

We then have defiance. The best defiance buff you can build up is somewhere around +50%. To be conservative I'm going to say that the blaster only accumulates +30% defiance buff before using Inferno, which takes us to 406.5%.

At that point, Inferno will average, using my average base calculation, 1892.87 points of damage, compared to Blizzard at the damage cap of 2001.96 points of damage. To beat Blizzard the Blaster would have to find about +24% damage buff, either in more slotting, more damage bonuses, more defiance, or some ally buff in the team. Its not a lot and my calculations were on the conservative side. Could I specifically *build* a blaster optimized for damage strength that could close that gap solo? Yeah, I'm pretty sure I could. With purple sets, absolutely (due to the +4% damage set bonuses in those). On the other hand, short of resistance debuffs there's nothing you can do as a defender to increase Blizzard above 2001.96 because that's the damage cap. Heck with fitness becoming free Blasters are probably going to be in a better position to take Assault, which adds +10.5% damage to the blaster, cutting the gap in half.


So it kinda depends on what you consider reasonable self buffing for the Blaster. With enough slots and inventions dedicated to damage buffing, its still possible for a Blaster to spike an Inferno that beats a non-scourging Blizzard. With no effort at all a Blaster can get pretty close to what a Defender can do only by damage-capping themselves with Fulcrum shift. It depends on how hard you want to work at it as the Blaster.

Actually, *not* all the time. I like Mids - its the only planner I've ever recommended, way, way back when it was first written, and that was at a time when I rejected the idea of planners - but it does have just one flaw in this circumstance. Actually, one flaw and one small approximation. But hold that thought.

First of all, Mids agrees with my calculations. It shows Blizzard with base damage 500.5 (which agrees with my 500.49) and Inferno with 472.2 (which doesn't quite agree with my 465.65 because Mids doesn't make CancelOnMiss calculations - that small approximation I mentioned - when I rerun with the same approximation I get 472.21 which agrees with Mids). But that is under the average calculation mode. You said:

To see what happens when all ticks hit, even probabilistic ticks (like especially the last two of four Inferno waves) you have to tell Mids to do that. Options -> Configuration -> Effects & Maths -> Show Max Possible Damage.

When you do, Blizzard stays the same (because its calculation already presumes every tick triggers) but Inferno goes up to 544.3, which *also* agrees with my calculation of 544.29.

So if all ticks hit, the maximum possible base damage for Blaster Inferno is higher than the maximum possible base damage for Blizzard, both by my calculations and for Mids if you are using Mids in the correct mode to see maximum damage. The default is average damage not maximum because that's usually what people want to know: on average, if you use this power a lot, what would you expect its damage to end up being. It most closely tracks with the value of an attack like a tier 9 (DPA also becomes important to the value of an attack for an attack intended to be part of an attack chain).

Switching back to average mode, Inferno can still outpace Blizzard when buffed, but only if you recognize that Mids has one tiny error (at least in the version I'm using): it doesn't honor damage strength caps consistently correctly. Blizzard can be buffed to 500% damage in Mids, but actually has a 400% cap. If you know that and cap the buff accordingly yourself, Inferno at the (500%) damage cap outdamages Blizzard at the damage cap in Mids, in exactly the way I described.

So yeah, Mids basically agrees with me (except for the damage cap thing). It would be weird if it didn't really, since the fundamental basis for its calculations is basically the same as my own.


Incidentally, separate from everything else, Blizzard is a grey area power for another reason. We call it "an attack" for mostly historical context reasons. Its a tier 9, like other tier 9 blaster attacks, so its an attack. But its not really an attack: its a summons that generates a pet that fires 75 attacks against all targets in range. Because its lifetime is so long relatively speaking (15 seconds) its not like normal attacks or even normal tier 9s which usually take no more than about 4 seconds to deliver their damage (Inferno takes about 9 to complete its DoT). Technically speaking, if Blizzard is an attack, then Voltaic Sentinel is an attack.


In any case, as I originally stated:

Only if you compare Blizzard's average unbuffed base damage to Blaster Inferno's average unbuffed base damage does Blizzard outdamage Inferno.

But the question is somewhat more complex than that.

Those threads pop up from time to time. I usually don't bother with them at all unless someone says something interesting or different. You'll notice I haven't posted in that thread.

It wasn't a map trap. It was a weird spot LR just strangely would stand in and not leave unless we specifically led him out of it.

As to the AI, all I know about the "movement brain" is what I've tested: I've done a lot of testing of critters moving, running, being led from anchors and running from threats. Its been a hobby of mine, so I've done a lot of extensive testing there. The "shooting brain" is something I know a little more about: it used to be exploitable but I pointed out in what ways back in I14 beta and eventually those exploits were closed by the devs by simplifying the AI so the critters stop trying to figure out what the best possible attack to use is now and just shoot you in the face with whatever's handy. It prevents them from getting stuck not shooting because before you could induce a critter to "pick" an attack by doing something, and then exiting the range of that attack and forcing the critter to try to move towards you to use it, and after a while it would give up and switch to another attack and then you could do something else that would convince the critter not to use *that* attack. Its hard to describe without pictures.

Castle once hinted on the open forums that he also knew something about how this worked when he said he could do weird things in testing to make AVs less of a threat to him because he knew how the powers were designed and how the critters would use them, or not use them. It was a long time dirty secret about the critters that they were dumb as tacks when it came to using their attacks, and also a big hassle for the powers team when they would give a critter a bunch of attacks and discover the moron would often only use two of them and stand around picking its nose the rest of the time. If you recall a patch note ever saying something like this: "critter X now uses power Y. It always had it, but just never used it before" if Y was an attack that's usually the problem I'm talking about. That doesn't happen anymore, or only happens much less often.

What I know about aggro AI is fuzzier. I know critters keep an aggro list. They actively seek out the thing at the top of the list that has the most aggro (or "hate" as the case may be). But it has triggers to prevent it from doing futile things. If a critter tries to attack you for some (I think adjustable) period of time and it fails to be *able* to get a shot off, at some point a trigger in the AI goes off and flags you as essentially worthless and the critter goes off to attack something else. I think its even flagged to be "afraid" of you if you can attack it and it can't attack back after a certain period of time, because if the critter is not aggroed to anything else it literally tries to run away from you rather than chase you. But this timer can be reset by giving it the opportunity to shoot at you at least once and can sometimes be reset by giving it the opportunity to shoot at you at a time when it can *queue* an attack to be used but can't actually *trigger* that attack because its doing something else at that moment in time.

Technically, I suppose that's an exploit, but its so difficult of an exploit to leverage its usually not worth it. You're certainly not going to be able to leverage it for any substantial reward that couldn't be gotten by someone else just obliterating the thing you're playing around with.

Its a problem, but not a widespread one. And *most* of the scrappers I know that are regulars in the "high performance" discussions would never say that the value of Scrappers or any other archetype was based primarily or solely on being able to solo AVs.

If nothing else, soloing AVs is something that the scrapper forums haven't considered the benchmark for performance in years. First it was my Rularuu challenge, then the Rikti Crash Site challenge, then the Storm Palace challenge, then the RWZ challenge, and now odd corner case weirdness like my AE mission and (for offense) the pylon standard. Beyond being silly, the soloing AV thing is also so 2006.

To the extent this is a problem, its a problem that affects all archetypes to some degree and AV-soloing is not the biggest offender. There are still people who judge Defenders by their ability to heal or their ability to match the offense of offensively focused archetypes. There are people who judge controllers purely by their ability to farm. Some people judge tanks on the assumption that aggro control is redundant or valueless. Some think Stalkers should be the equal of Scrappers in scrapping and have placate and assassination strikes just for fun.

All stupid, all fortunately uncommon, but all annoying nonetheless. The question is should you honor such nonsense just because it happens occasionally. I don't think so.

The hard limit for defense is archetype based, just like the resistance cap. Except since defense is a buff, it can exceed the limits of what is *useful*. For most archetypes the defense cap is somewhere out around 200%, plus or minus. But since the base tohit of most even con critters is 50%, anything more than 45% defense doesn't help in those situations, because you cannot reduce the intermediate tohit value to less than 5%, the "floor."

So 45% is considered the "soft cap" to defense. Soft, because you can get more, and it can help sometimes - when the critter's tohit is higher (pets, turrets, and some special critters), when the critter has tohit buffs (Rularuu eye balls, bank guard LTs, some red side AVs with Build Up, etc), or when the critters have defense debuffs (which will lower your defense, meaning having more than 45% allows for a "buffer" against those debuffs).

The hard cap, on the other hand, is something you're not likely to see very often, because its very high. Although it also scales by level, so its possible to hit it at very low combat levels with a lot of buffs, but still you're not likely to see it often.


Good time to mention that the soft cap or the soft floor as its also called only allows you to reduce *intermediate* tohit to 5% and no lower. Its called the intermediate tohit because its multiplied by Accuracy afterwards, so the final value is potentially higher than that. That intermediate floor didn't always exist, but it was discovered in I3 (I4 testing) that in situations with high defense, specifically PvP, you could have enough defense to drive someone's tohit *negative*. If you did that, Accuracy wouldn't help such a person: stacking more accuracy would only make that negative number more negative. So they introduced the intermediate floor, so that Accuracy couldn't be made absolutely worthless with enough defense. You can't use defense to drive tohit negative anymore, and as a result accuracy always has a positive effect for the attacker.

So the tohit "floor" isn't really 5%, the intermediate floor is 5%. The final tohit floor is sometimes 5%, and sometimes higher for higher accuracy critters. Note: this isn't unfair to defense. The maximum mitigation from resistance for players is 90% (less for most archetypes). So you'd think that the best you could do with defense is 90% also. But for a critter that hits 50% of the time normally, 90% mitigation drops him to 5%. And for a critter that normally hits 75% of the time, its 7.5%. But in the old days, a high defense player could drop something like a high level AV that had 90% tohit all the way down to 5%, which is a net damage mitigation of 94.4%. That's what made perma-Elude much more powerful back then. Today, best you can do is 90%, which means the "floor" for a critter is usually one tenth their normal chance to hit, which is not always 5 percent.

Its been a while since I tried this, but way back when I was able to draw LR to leap up pretty high up the buildings and then confuse him into leaping back and forth between perches, all while so far away from the action (vertically) that he didn't seem to want to aggro on the action below. This required constantly shooting at him though, and not just the occasional snipe: I had boost range which meant I could hit him pretty much constantly while staying just out of his range. The trick is you have to be far enough away that he can't quite shoot, but close enough that he thinks he can. Also: if you stay out of range permanently the AI will eventually register you as an impossible to hit target and can decide to either run away from you or run after someone else. The trick there is to re-enter his range when he cannot shoot at you, and that's possible if you do that as he is playing an uninterruptible jump animation. I call it "tease-kiting" and it takes practice, and dying a lot to figure out how to make it work.

I've actually practiced that on the actual STF map on the couple of occasions I was on STF runs that failed for some reason, but you can also get a lot of practice on the AVs in RV. Hard to simulate other aggro coming from other sources, but at least you can practice trying to kite an AV in a restricted area by pulling them near to a building and setting arbitrary boundaries for yourself. Yeah, I've done that, and yeah, that's probably a bit mental.

I'm not saying its necessarily the best way to go: I'm just saying if I tried to tank LR with any of my blasters I'd be dead in fifteen seconds. But pulling is something I'm extremely good at, so I was playing to my strengths there.

If I recall correctly, the last time I ran an STF there was also this little corner down below the platform off to the right just inside his teleport tether that he could sometimes get stuck in (or rather decide not to leave under certain conditions). Sometimes the AI makes bad decisions when it comes to movement. It used to also sometimes make bad decisions when it came to shooting: you could game LR to actually not fire for extended periods of time, or fire infrequently, with the right hokey-pokey movement (I mean that literally: by moving into and out of the range of his attacks, particularly around 60-80 feet). You shouldn't be able to do that anymore, because the firing brain of most critters should now be much harder to confuse in that way (actually it should be practically impossible because they don't "think" as much about which attack to use).

Nah, we're talking about Issue 4ish or so, and I was already a loud mouth by then. Its more that I was still getting a handle on the intent and deep mechanics of the tohit system. This was approximately the same era when Stargazer posted her conclusions about accuracy mechanics which was a big hole in our understanding of the tohit system until that point.

Intent was extremely difficult to deduce until after I4 and I5 when the devs started opening up some of their balance thinking surrounding melee defenses, perma powers, and the GDN. And communication lines got much better frankly when Castle joined the crew and particularly when he became open to questions in CoV beta (although other developers such as pohsyb also became more chatty around the same time - its pohsyb that acted as my sounding board to work out the details of the tohit algorithm and the changes within I7).

I'm sure Castle still has nightmares where he hears the words "so we're thinking of making something called 'archetype representatives' you interested?"

+30% tohit. The problem is the red tower which is giving him a +105% damage buff. At that level of damage buff even the ~10% of attacks that land while floored with defense will deal more damage than even most tanks can absorb solo without healing assistance. If that tower is gone you have a fighting chance of solo tanking his damage if you are soft capped and also have significant resistance and/or healing.

Also: his Arm Lash has a -10% defense debuff and he sometimes spams it. Its stackable and lasts 10 seconds per application. Although at +4 that would be -14.4% debuff. Without defense debuff resistance the first hit from that would reduce 80% defense to 65.6% defense and Lord Recluse would start hitting at a net 30% chance (1.4 * 1.5 * (80-65.6) = 30.24). It probably gets quickly worse from there.

In an all blaster STF I wouldn't try to tank him with a blaster, I would shoot at him and get him to run around following me all over his tethered area, and load up on healing insps and break frees just in case. His maximum range is 110 feet, and its possible to snipe-aggro him and then stay out of his range almost permanently with careful pulling. And like all critters, he's pretty brain dead if you see him often enough to figure out what confuses him (deciding on leaping directions is one thing I recall, although that was a while ago).

If there are blasters actually tanking LR in the STF without massive ally buffs, more power to them but my guess is that would have to be a highly optimized build for that purpose.

I can't take credit for this one. Someone else mentioned it on the forums first. What's funny is that I knew about this problem for a long, long, long time and never bothered to mention it except in passing because I thought that was just how it was intended to be. Back in the day I used to team my MA/SR with an Ill/FF controller a lot, and she would often remind me to flip off my toggle(s) to save endurance because really, we knew that while bubbled they weren't doing much of anything.

It was when this started to be discussed and Jack started implying this *wasn't* supposed to be that I immediately jumped in and started pointing out that these stacking problems existed pretty much everywhere, including the power pool powers. Ironically and somewhat humorously in retrospect I mention this very problem in an early version of my guide to defense just in passing, and apparently no one noticed or bothered to question me on it. I removed it in the version just prior to when the discussion flared up on the forums. But I consider this one to be the biggest duh-moment of my time here.


And by the way, my understanding is that the multiple-type defense stacking behavior was a bug, and never intentional. Or possibly intentional on the part of the original programmer, but not in keeping with the design intent of the game engine. But its difficult to be certain, because there's strong evidence that at the beginning of time no two game developers had exactly the same understanding of how defense and tohit worked or was supposed to work. We were still learning things about what the game engine was programmed to do as late as Issue 8.

Say what?

I'm pretty sure I've always disputed any such claim in any discussion thread I've been a part of. In fact, I'm on record as stating that as fun as it is, and as much as its not a big priority for the devs to try to address, being able to do it at all is a sign the game's broken in ways very difficult to fix.

Also, I'm very twitchy about people making *any* claims about the "worth" of an archetype based on any narrow metric, not just soloing AVs. In fact, I tend to slam people pretty hard for making such claims.

I hate to ... no, actually I love doing this. In my opinion, if you're going to be absolutely certain, its important to also be correct.

So lets get to it. Blizzard deals 75 ticks of 0.06 scale damage lethal and 0.06 scale damage cold over its 15 second lifetime. This is 75 pulses of attacks, not a 75 tick DoT (more on that later). Total damage is scale 9 damage. Its a pseudo-pet so it deals damage on the pet damage scale. That means at level fifty it deals 55.6102 * 9 = 500.49 points of damage.

Inferno deals Scale 1.0 smashing, plus Scale 2.0 fire, plus 75% chance for Scale 1.5 fire, plus another 50% chance for another Scale 1.5 fire. If all waves hit, Scale 6 damage (basically the same as Nova). *And* it also deals nine ticks of Scale 0.3 fire damage over eight seconds with 99% chance for each tick (and its coded cancel on miss). If *everything* hits, that's a total of Scale 8.7 damage (6 + 9*0.3).

But for Blasters that's on the blaster ranged scale which is higher than the pet scale. At level 50, that would be 62.5615 * 8.7 = 544.29 damage. Oops, looks like Inferno beats Blizzard and it deals that damage almost twice as fast to boot.

Of course, that's maximum damage. Neither power will hit all of its ticks all the time. The *average* damage depends on your accuracy. Important to note that the Blizzard pseudo pet damage power has standard accuracy (1.0) while Inferno has tier9-style accuracy (1.4 - just like Nova). But lets assume both powers are stuck at the tohit ceiling of 95% vs its targets. In that case the average damage of Blizzard is fairly easy to calculate: its just 95% of the total, or 0.95 * 500.49 = 475.47.

Inferno is a bit more complex because its waves have less than 100% chance to trigger, separate from the power actually hitting the target. The average of the base damage is 1.0 + 2.0 + 0.75 * 1.5 + 0.5 * 1.5 = 4.875 scale damage, or 304.99 damage at level 50. The DoT is slightly more complex because its a cancel on miss: if any tick fails all the other ticks fail also. That means the average number of ticks is 8.56, and the average damage from DoT is 8.56 * 0.3 * 62.5615 = 160.66. Total average damage: 465.65. With the 95% tohit ceiling that gets reduced to 442.37. So actually Blizzard does more on average.

But not so fast. Blizzard is a pseudopet, and thus has a 400% damage strength cap. Inferno is a direct attack, and thus has a 500% damage strength cap (by virtue of the Blaster having one). At the damage cap, Blizzard does 1901.88 damage on average, while Inferno does 2211.85 on average. Inferno deals more.

So, actually, you're wrong. Blaster Inferno's maximum base damage is higher than Blizzard's. And Blaster Inferno at the damage cap deals more damage than Blizzard at the damage cap for either the maximal case or the average case. Only if you compare Blizzard's average unbuffed base damage to Blaster Inferno's average unbuffed base damage does Blizzard outdamage Inferno. By 33.1 points of damage at level 50. Interesting final question: at what level of damage buff does Inferno overtake Blizzard on average damage? At 430% damage buff. What's the damage buff of a blaster slotted to the ED soft cap and with BU and Aim stacked onto Inferno? 357.5%. So at the point where the team has about 72.5% ally damage buffs floating around, Blaster Inferno will be able to exceed the damage Blizzard puts out assuming the targets remain under its area of effect for the full 15 seconds. Less if we assume Defiance is running some level of buff on the Blaster at that moment.


Not so simple, is it?