Arcanaville

I believe Blaster Inferno will beat non-scourging versions of Blizzard. I think the numbers were discussed up-thread.

That matches what I saw. So its probably a bug. I'll report it to the devs: its not the first time a power wasn't reporting tohit in chat, and it makes herostats statistics invalid for that power until its fixed.

Yep, there's gotta be something wrong with that. And I just logged in one of my Fire tankers in on live, and I notice Blazing Aura isn't generating tohit rolls in my combat chat on hits. I see the damage, and I do see misses, but no actual tohit roll messages on hits, only misses. Maybe that's just me: I'll do a bit more testing tomorrow, but without those messages herostats won't be able to generate proper statistics. Your numbers might be faulty for that reason.

When you actually look in your combat chat, do you actually *see* tohit rolls and hits or misses from Blazing Aura? Something like "Blazing Aura had a 95% chance to hit, and rolled a 46.23." or similar message? Or are you just getting the "Blazing Aura deals 37.07 points of fire damage to X" and "Blazing Aura missed!" messages and the miss rolls ("Blazing Aura had a 95% chance to hit, but rolled a 96.23.")

I notice your last session showed 134 misses and 9 hits, which implies your entire recorded session generated 143 ticks of Blazing Aura, with a tick being one pulse of BA hitting one target. That's like the number of BA ticks you'd generate in about a minute or two of game play. That doesn't sound right.

Also, my Blazing Aura does seem to be hitting targets at a much higher than 5% hit rate when I watch the actual damage.

How is it that the critters don't kill you when they shoot you in the head and you aren't superhumanly resistant to brain damage? A combination of comic book logic and game simplifications.

Hidden within the game simplifications is a lot of real world possibilities. Perhaps its not that you literally missed a motionless target. Perhaps the target moved suddenly at the last instant, or you stepped on a twig while executing the strike and the target got lucky and turned towards you in just the right way for the blade to miss. There are all sorts of ways for such an attack to fail to hit the target other than you literally can't hit the broad side of a barn.

They are not the only ones. Pets (including enemy pets) and turrets (like the ones Malta engineers spawn) still have 75% chance tohit. I think its partially that they always had higher tohit (even moreso than other things prior to I7: at one time turrets had 105% base chance to hit), and partially that the devs wanted to keep some things around that still made life a little harder for defense sets, just like certain damage types tend to do for resistance sets. Its reasonable for such threats to defense to exist if it doesn't get out of hand.

Turn on chat logging (there's a way to do this in the options menu), make sure tohit rolls are being logged in at least one visible tab, and when it happens again PM me an excerpt of the entire sequence surrounding the problem. Trust me when I say the odds are very good there will be a logical explanation for the apparent problem that is not due to a broken random number generator or a malfunctioning tohit system. This is something I know a little something about.

You can grant a whole power to a target. You can't edit a power on the fly, either your own or a target's. You can, of course, do things like apply buffs which affect the power, which is actually in a weird sense how Dual Pistols ammo works - by buffing and debuffing the chance for certain effects to occur. But you cannot create an effect that doesn't already exist and add it to a power that already exists.

Nothing is superior to a sword composed primarily of awesome.

The rand the devs use for the tohit system is, I'm told, basically the standard (modern) rand() from the standard C libraries. Its known to be a strong generator, and it doesn't show streakiness of a kind a human being could possibly detect. My tests of it show that the tohit rolls generated by the game are consistent with using that generator. Its not cryptographically secure unless well-seeded with an entropy source and some implementations I've heard have LSB weaknesses, but its close enough for game purposes.

The streakbreaker code is not there to correct a problem with the rand. Its there to break unlucky streaks because players complained about them way back in the old days enough to make a dev decide to implement a guaranteed miss streak limiter that frankly I'm guessing was as much to allow them to *know* when someone was lying or mistaken about miss-streaks as to actually provide a benefit to the players when attacking. I know its the first thing that tips me off when someone claims behavior the streakbreaker makes impossible. Its *always* either an error of perception, or frankly not the truth, every case investigated.

Not before I get Super Reflexes renamed to Super Xtreme Cosmic Evasion Hyper-Dodging™

Man, that joke's almost as old as the debate over whether Magical Trevor was a radiation defender.


Its going to be very confusing to players playing the end game when all the Fire tankers earn so much -Regen buffs by Issue 23 that they are killing the end game villains in Issue 19 before the players get to see them in Issue 20. But then again, people keep complaining this game is too linear anyway. And kill-stealing becomes a thing of the past by being a thing of the future. Sort of.

Wouldn't really work. Pets don't "immediately" attack something that attacks the mastermind, they will aggro on things that attack the player because that attack is signaled to the pet AI as if it was attacked. But if the pet is taunted, say, attacking the mastermind doesn't automatically generate a retributive strike. While taunting is not an issue for a pseudopet or untargetable pet, aggro also doesn't translate immediately to a counter attack because pets don't attack instantly. And if you're attacked by a number of foes in rapid succession, there's no way to get the pet to counterattack all of them. And giving pets instant attacks creates other potential problems, like convincing them not to just spam them all the time. This would likely be vulnerable to all sorts of AI-related race conditions.

To be honest, the easiest way to implement deflection/counterattack into the game engine would almost certainly be something most players never suggest but makes more sense design-wise. Make the game cause *every* attack to generate a retributive counterattack, and add a way to turn it *off* by flag. Then set the flag "off" by default, and make the devs have to turn it on in special circumstances.

This way, there's no need for massive computations to figure out how and where the counterattack would go. Every attack basically simultaneously self-damages the attacker unless a particular flag is off, and the flag starts off that way. The flag would then be set not on the attacker but as a target check: if the target of the attack had the flag set, the counterattack would happen. So defensive powers and abilities would in effect "turn off the counterattack suppression" and the game engine would allow the attack to self damage the attacker. No auras, no reverse calculations, no CPU expensive conditionals. But you would need to make a fundamental change to the way damage attribmods are processed, which won't be trivial. That's a bit of an oversimplification of how I would do it, but it illustrates the basic idea. Sometimes its a lot easier to compute when something shouldn't happen then when it should, and that's often the path of least resistance.

Trying to do this with what we have now is likely to be impractical. People have thought about it for years, and rethink it every time a new game engine feature arrives. I personally don't think its possible with the current toolbox.

Correct. There's no way for a power to change the definition of another power on the fly, so there's no way for a player power to "add" a self-damage component to a critter's power. There *is* a way for one power to flip some of the effects of another power off and on, so if all the critter attacks were changed to have a self-damage component that was normally off, a player power could flip that on temporarily. But of course, that requires literally changing all of the critter attacks to support this feature. There's no way to do it dynamically.

In all seriousness, no. If Positive regeneration increased the *size* of the regen ticks, then negative regen would reduce the ticks to negative values, which would effectively be self-imposed DoT rather than regen. But positive regeneration doesn't increase the size of the ticks, it speeds them up. In effect, your regeneration rate tells the game server when your next tick of regeneration should occur. Zero regen would make the tick interval infinitely large and cause the next tick to never happen. Negative regen would, if it were possible, dictate to the game engine to apply the regen ticks in the past. Which of course it can't do.

The server quantum is 1/30th of a second. The animation clock is also 1/30th of a second. The combat clock is 1/8th of a second (0.125s). There are some other clocks I've inferred in there, but they may just be derivatives of those two. The "ArcanaTime" factor of 0.132s is a "beat" that occurs due to an interaction between the animation clock and the combat clock.

But I believe the game server world itself updates in terms of discrete 1/30th of a second intervals. A lot has to happen in that 1/30th of a second to prevent server-lag from occuring. Blowing the server quantum was the primary source of lag during old-school Hamidon raids: doing so actually slows down the zone's** effective chronometer, which causes time to slow down (which slows down among other things recharge, mez duration effects, and power activations, all significant to Hamidon raids)


** I say "zone" but its not 100% conclusive that a slowdown in one part of a zone slows down time in all parts of the zone all the time. I think it sometimes does, but its also possible zones are subdivided and lag doesn't propagate to all elements of the zone equally. But no one thought to do conclusive experiments along those lines back then. Still, when there is a zombie or rikti zone event, packed hot spots don't automatically lag out the entire zone equally, do they?

Depends on the threading and the core architecture. Nehelem dual cores can consistently get above that, and some other architectures that implement even more independent cores like Bobcat can probably start to approach 100% on threaded applications that don't stall the floating point unit.

Sorry, no. One, the floor for regeneration is zero. But also two, even if the floor wasn't zero, I think the game engine has no concept of negative regeneration. Regeneration affects the regen tick rate. I don't think the game servers are capable of making regeneration ticks happen backwards in time. Might have to wait for CoH 2.0 for that feature.

There's a couple new critter types with base 64% tohit. Its not something specific to the DE. However, I don't have a list of what uses those critter types off hand.

Just a minor correction, I believe those have 64% base tohit, not a 64% tohit bonus. That's just unfortunate wording in the Real Numbers displays.

Interesting that that Greater Devoured is classified in ... that way. Going to have to check other GDs elsewhere.

CoH abstracts that away, and most MMOs that do so do so deliberately to emphasize other elements of gameplay. The tohit system in CoH is no less "genuine" in that regard.

And until you see MMOs integrated with Wii Fit boards or Microsoft Kinects, you won't see a "genuine" dodging mechanic either.

Abstractions are often important elements to successful games. In an age with 3D graphics and complex combat simulations, people still play chess, the ultimate in abstracted combat.

MMOs all have - even WoW - niche appeal. No MMO can or ever will satisfy the majority of all game players. I would bet that the vast majority of CoX players would quit out of frustration if we had the cut-throat and nearly unregulated economy of Eve Online. On the other hand, Eve Online has more players than we do. Is that because Eve Online has the better model than we do? Not really: WoW players would also revolt, and they outnumber Eve Online players over thirty to one. This is what makes it so dangerous to extrapolate either player preferences or subscriber numbers into "what all/most gamers want." Most don't want City of Heroes no matter what our combat mechanics are. Of those that do, some like hitting and missing, and some don't. Thinking you know for certain what would happen if you changed that dynamic radically is never a safe assumption.

When CO was in beta I predicted that the endurance building mechanic, which allowed for a much higher activity level than CoX, would not be universally seen as a good thing. Some people would love it, and others would hate the gameplay it promoted, and neither would be the vanishingly small minority to the other. And I think its fair to say that prediction was confirmed: some people definitely liked it, and like it to this day. For others it became a strong contributing factor to becoming quickly bored with the kind of combat it encouraged.

There was no "right" choice there. You were going to lose some very large percentage of players either way. The question was not how to keep them all, but which ones you wanted to keep. The same thing is true with hitting and missing. You lose missing, and you keep the people that want to hit all the time. You lose the people that actually want to see evasion in the game, which explicitly requires missing. You pick one, and you let the chips fall where they may.

Problem #1: "Beginner's luck" is a tohit buff, not an accuracy buff. There's no universal conversion between the two. Beginners luck is stronger than comparable accuracy when tohit gets worse and weaker when tohit gets better, and that is almost certainly intentional.

Problem #2: The intent was to provide the most benefit to the lowest level players when not only enhancements but slots themselves were in short supply. Increasing the strength of accuracy enhancements while leaving the rest alone simply creates the situation where accuracy slotting is so much more beneficial to use than any other kind of enhancement that its essentially a design-forced behavior. At the moment accuracy and damage have comparable benefit at lower levels which makes it an entirely personal preference-driven choice to slot to hit harder or slot to hit more often. In terms of overall average DPS and average DPE, the net result is the same.

Problem #3: Exemplar. To make this work you'd have to make special weird scaling tables for accuracy under exemplar with special code needed to implement it. A lot of extra work for questionable benefit.

Not universally true. People play all sorts of games where their success percentage is lower than that. Not everyone who plays basketball, for example, has a 70% shooting percentage. Even moreso baseball. And when people play games of chance, like slot machines or casino table games, virtually all of them have lower than 70% success rates. I'm honestly not sure what your reference is for the 70% number, because its not a number I've ever seen come up in any psychological or games study.

Edit: responded before reading the rest of the thread. Normally I don't do that but I got distracted. In any case, studies suggest the tolerable success rate seems to be dependent on circumstance, but in many cases can be very low. Also, what defines "success" is also significant: hitting with a swing is a form of success, but for many people its less important than defeating the target before dying, which is another form of success.

How high is high enough? Player base chance to hit an even conning critter with no defense is three out of four. That's already hitting "most of the time." Everything you hit less often than that was *designed* to be hit less often than that. Increasing player accuracy to compensate is counterproductive: the devs would just make the critters even harder to hit, since that's what they did in the first place.

If the devs had it to do over again, they probably wouldn't even make a system where you could miss a meaningful amount of time. Some players can handle missing, and some just can't. It has nothing to do with fun, unfun, too much, too little, too random, not random enough. Some people simply cannot handle missing in a game. We're really just dancing around that fact with mechanical compromises.

*Why* this is true is a separate discussion issue. The original devs came from a very PnP-influenced mindset where missing happens all the time. They probably didn't even think it would be an issue. And honestly while people can throw theories around (if you can think it, I've probably already posted it) I really don't know for certain the fundamental reason why some players cannot handle missing. But I suspect its due to the fact that the devs made so much of the rest of the game's mechanics too predictable. Predictability breeds the expectation of predictability everywhere. Because there is no randomness or variation at all in power effects, we can actually calculate almost precisely what our powers will do. And they always work in exactly precisely the same way under the same conditions. *Except* for whether the powers hit the target or not. If damage was variable, if effects were variable, if there were chances to avoid or shake off secondary effects rather than perfectly predictable mechanics for this, I think players would focus less on missing.

Its neither here nor there, though. If the devs get a do-over they will probably deemphasize missing. If I had a do-over I would deemphasize predictability. I doubt that philosophical difference will ever be put to a definitive test anytime soon.


By the way, I slot accuracy TOs and DOs. Some players think its pointless, but to be frank most players think all sorts of odd things, not just restricted to accuracy. You can try to help players out, but at some point you can't target incorrect player perceptions and still have a remotely sane game.

Just to give my take on this, first of all the fact that the streakbreaker follows lowest tohit chance in the miss streak *doesn't* benefit the player. Its not supposed to. The streakbreaker *itself* benefits the player, at least when on offense (it hurts a bit when on defense). The streakbreaker adds hits when the random number generator otherwise says you miss. It always helps the attacker. The fact that it doesn't help as much as you want it to is a separate issue.

And why it doesn't help you as much as you seem to want to is that its only intended to "nudge" the chance tohit without skewing it all over the place. To amplify what Father Xmas posted, I dug up some quickie calculations I did on the streakbreaker from way back when it was first discussed:

Sorry, Firefox is currently converting TABs to spaces and messing with the formatting. This is basically an estimate of the streakbreaker performance. SB1 and SB2 are the streakbreaker break points. M is the miss streak tolerated (the Streakbreaker will break streaks *higher* than that number). MS1 and MS2 are the probability of exceeding the streakbreaker threshold for the top and bottom of each chance band. H1 and H2 are the number of hits random chance would produce per 10,000 rolls. HS1 and HS2 are the estimated number of hits injected into the streak of swings due to the streakbreaker, which is the MS1 or MS2 probability divided by the streak length requirement (M + 1). SP1 and SP2 are the resultant actual chance to hit averaged over long runs of swings. In all cases, the SB doesn't shift the chance to hit by an overly strong amount, usually only a percentage point or so, while its breaking miss streaks. Which is its intent: to break the worst streaks of misses, without radically altering the actual number of hits and misses to the point of breaking the whole idea of chance to hit altogether.

These are just estimates. The "productivity" depends on how many actual swings you generate and the changing circumstances of those swings, and there is an independence error implicit in these calculations which I think changes things a little bit. But they are useful rules of thumb. The actual calculations for determining precisely what the streakbreaker does is slightly more complex. So much so, you're probably better off monte carlo-ing the whole thing.


As to the issue of complaining its too complicated, to be honest I find that to be a bit of a quixotic complaint. The *simple* thing to do is to not have a streakbreaker at all and let the dice fall where they may. But once players start claiming that "random" is "unfairly streaky" they are really asking for a complicated thing, whether they know it or not, whether they admit it or not, whether they understand it or not. The devs tried to put in a system where your overall chance to hit was still very close to the calculated values, but where long streaks of misses magically vanished. Satisfying both requirements simultaneously cannot be done with a simple system.