Arcanaville

Just to be clear, when you reach the streakbreaker limit for any tohit range, the next attack autohits without consulting the roll. It doesn't "force the roll" to be 100%, and there's no way for the random number generator to somehow "override" through malfunction this behavior. Because the roll is simply discarded. You can tell by looking at combat logs: when a streakbreaker hit is induced, there's no roll: the logs say attack was "forced" to hit by the streakbreaker.


Also, just for giggles, I've decided to take another look at the rand used by the tohit system. In the past I've data-harvested combat rolls over long periods of time, and theoretically speaking that could create a skew all its own if the rand was somehow affected by game patches: I could be datamining across radically different implementations. However, its now possible to harvest millions of random rolls in a very short period of time, without any real effort from me. I've been looking at those. I want to analyze about a hundred million combat rolls just to see if there is *any* evidence of a problem with the rand. Frankly, I doubt there is one, but just from the data I have so far I can say there's still no evidence of statistical anomalies with one exception which would probably be impossible to detect in actual play (a hint of a spectral anomaly affecting the least significant bits of the rolls, which some non-cryptographically hard random number generators are subject to, and which most people don't care about - this would only affect how often the decimal part of the roll was a particular value across all random rolls, which is unlikely to affect anyone materially).

Normal MMOs are not designed to have that level of mindless simplicity, even if some people would prefer that they do.

It sounds like you would be more interested in a slightly less complex game.

Theoretically true, but I've never actually had that happen. I doubt it happens often, because such situations would annoy experienced players and consistently kill novice players. Missions which kill novice players up to thirty times before completion with regularity would be something I would think the devs would notice.

As a general principle, I agree.

Because he decided those effects made more sense with accuracy, not because he has a vendetta against tohit. +Acc shows up in focused accuracy, in the invention bonuses, and in the VEATs. Its not a wide-spread goal to eliminate tohit altogether: lots of things still have plenty of tohit and would have been easy to change if Castle wanted to change them.

Actually, that sounds more like me than Castle. I don't think Castle has a strong aversion to tohit buffs in general, although he realizes the problematic nature of them in certain cases. I've never disliked tohit buffs either as a mechanic, but they have serious balance implications in PvE and different ones in PvP. So I've always believed they should be used only when their full effects are intentional. For example, if you're going to make a DE eminator that provides a 100% tohit buff, you'll be basically neutralizing defense and ramming everyone basically to the tohit ceiling. If its explicitly intentional that this hits non-defense sets a little and defense sets a ton, because that eminator is explicitly intended to be a thorn in defense sets' sides, then its appropriate. If its just supposed to allow the DE to hit more often in general, its probably not. The huge preponderance of -DEF debuffs in the game and their common proliferation is a general pattern that this principle is not followed, and one of the reasons why defense debuff resistance was essential to add to defense-oriented sets.

But conversely, I've always opposed players that have suggested changes to the game mechanics that *removed* tohit buffs as a mechanic from the game as a whole. There are circumstances where its outsized and disproportional effect appears directly intentional. For example, the tohit buffs in Aim seemed to be geared that way: that power is intended to ensure to a high degree that you hit what you shoot at most of the time regardless of most levels of defensive protection. This is something only tohit buffs can easily offer, and its not improper to allow when the full effects are intentional and properly balanced for.

But I don't think Castle has ever stated an aversion to tohit buffs in the general case.

To be honest, while I think the two caps are logically consistent, there's no overt reason behind them other than the fact that the numbers were at least partially modeled after twenty diced dice by Geko (or whoever implemented the tohit algorithm). I once proposed changing the floors without changing the ceiling, and Castle has also stated that if he knew now what he knew then, he might have proposed something similar when he first joined.

I've oddly solved more people's accuracy problems that way than any other.

Also, its very unlikely at this point that there exists a flaw in the random number generator for the tohit system, as its been checked and double checked many times over the years. But its not impossible, and there have been random number generator errors in other places in the game. So an actual record of tohit oddities would be useful if an actual oddity occurs, because at this point the devs are unlikely to investigate a problem without evidence. They will, if there is evidence to support the problem, and the chat logs are the easiest way to have those always available if a problem occurs.

I'm guessing WoW and most other MMOs don't also have trivial ways to radically improve your chances to hit either. An even acc SO increases net chance to hit by 33.3%. How common is that in WoW? Build Up has +15% increase to tohit and is amplified by Accuracy. Inspirations fall frequently and the smallest of those boosts tohit by 7.5%.

The fact is that CoH has much stronger buffs and debuffs than any other game I've seen, and allow them to stack much more strongly than in any other game I've seen, and that works both for and against the players when it appears.

Way back when, I used to hunt the Rikti Crash Site (long before it became the RWZ) with my MA/SR scrapper, and one of the things I used to hear is that a lot of people didn't like to do that because of all the Paragon Protectors, many of which go MoG. It took me a bit of time to realize that what I was seeing was not what many others were seeing because I was two-slotted for accuracy (this was pre-ED, mind you) and ran focused accuracy. MoG was flooring other people, but not me, because I had "grown up" under perma-Elude and the old school I2 days of fighting +5s and even higher on teams, so I built for a lot of accuracy so I could hit things. MoGed PPs were hittable for me.

That's part of the options this game gives you. You can choose to expend effort dealing more damage per hit, or you can choose to expend effort becoming more accurate with each attack. In a game where everyone hits everything all the time, that choice is gone. Its a choice many players don't want, but others do. For the choice to become a very accurate hero (or villain) to be valid, there have to be things that are hard to hit.

Also, although its debatable whether the devs went too far in many cases, the very strong buffs and debuffs in this game are also there to provide players with stronger (as in more distinguished) choices in powers to use. If CoH was a game where the "base" performance was pretty high, and all modifiers to that base performance were fairly low, buffs and debuffs would be a lot less interesting. Tell someone in WoW you have armor that provides 6% better tohit, and that might mean something. Tell that to a CoH player, and they'll look at you funny. Because 6% benefit is considered small in CoH. It means you're going to hit about 6% more often (depending on the kind of buff it is). In CoH, I can buff myself from 75% chance tohit to 95% chance to hit with slotting, which is an increase in effective damage of 27% (95/75). Its a dramatic difference if I choose to slot my powers that way, so most people do. Giving up that flavor just so people can hit (almost) all the time is not something to do lightly.

This was discussed in the past, actually within the context of drop mechanics back in I9, but its also applicable to tohit. Most of the proposals for doing this *don't* also ask for *hits* to be similarly "normalized" only misses, and the net result is that your actual rate of hits goes way up relative to what they are supposed to.


Do you actually have chat logs that show this behavior, over and over again? Seeing this often, as opposed to the one in a thousand chance its supposed to be, would be evidence of a broken random number generator. So far, every time I've asked for such logs, the problem disappears as soon as the person actually starts logging.

(You can enable chat logging in your options per character and have *everything* basically logged to local log files without having to do anything after that point, and the logs don't really take up significant space relative to the size of the game. This is an effortless way to see if the combat logs back up your perceptions of how often you really are missing.)

Chain induction jumps a maximum of four times. Theoretically, there's no limit if the devs wanted to make it jump more times than that. In fact, it would be easier to make a power that just kept jumping indefinitely until it either missed everything or ran out of targets. Making it jump exactly (or up to) four times is slightly more work. I think it may have actually originally done so in an early beta test of the power.

Very build dependent. Overall I'd say Willpower tends to be the strongest mitigation set across the widest range of possible builds from SOs to IOs. Shields isn't actually that strong, except in builds where you can build to soft cap where it then becomes a serious player because like Willpower it has other mitigation it can layer under the softcap (which is a problem for SR when talking about the strongest possible builds: its easy to softcap, but commensurately harder to get much else under it once you do).

I'm unlikely to update those comparisons any time soon, as they were done as much to explain the methodology of the analysis as anything else, which hasn't changed. In terms of comparisons, I shifted to the mitigation spreadsheet, which used to be downloadable but the link broke to it. I haven't had a chance to update it and upload it somewhere else yet. I was going to earlier but the Brute changes made me back off and decide to update it for those and upload a copy that was up to date, then the Brute changes were backed out and made that moot. Still, I should sanity check it before uploading the next version, as the last time I updated it was the I13 Invuln changes.

Nope. You can't chain jump to a target that was just chain jumped, whether it was one of your chains or someone else's. As long as there are other targets around, though, your chain induction will simply avoid that target and select another one to hit.

You'd be adding extra conditionals within one of the very critical loops in the combat engine. And you'd be retrieving power data from the powers database which is at the moment not processed in this way.


As far as I know, while there are major differences in the details, all modern MMOs are programmed this way.

Yep.


When a target is hit by the chain, it momentarily (for three seconds anyway) sets a flag on the target. The chain is configured with a special requires clause that basically makes the jump entities unable to target a foe that has been recently hit by a chain.

Oh no, not another one of these discussions.

MMOs are not programs where everything is written in code. There's a game engine, and then there's data which represents all of the entities in the game, like powers. The tohit system resolves hits. Its not "passed" information about the power, its the other way around: when a power is activated the tohit system determines if that power hits its target, and then the rest of the information about the power is read to determine that power's effects. This allows the game engine to be generic, while the data describing the content of the game can change.


No, its that you don't understand how systems like this are written. In *every* game I've ever seen, the engine and the data are separated. Its how mods are made: they aren't made by actually hacking the game code. And when you make mods, you can only do what the engine already allows you to do, based on what the engine is written to handle.

No one is going to go randomly fiddling with the tohit code of the game just to start adding in special cases to it, like "if this power is green, go do something else entirely." And without that sort of special case code, there's no way to reference whether a power is a "tier 1 power" or not.

That doesn't mean it can't be changed, only that certain changes are much easier than others. See: Elusivity.


Powers are either flagged "autohit" or not (technically, they are flagged as either autohitting certain kinds of targets or not). If they are autohit, they do not roll a tohit roll before affecting their targets. If the are not autohit, they roll a tohit roll which they must pass for their effects to affect its targets (this is also why the suggestion above is nonsensical: its more that the powers call the tohit system rather than the other way around).

Chain induction requires a hit. However its an AoE with a 1 target cap (or rather, the jumps are) and the jump has enhanced accuracy. If there are no targets in range or the jump actually misses all of the targets in range, the jump fails.


Because the game engine can only reference what its programmed to reference. The tohit algorithm is not designed to reference the specific details of the power invoking it. Doing so would make the tohit code - which is used all over the place for a lot of things - larger, slower, and more unwieldy. Its not sloppy. Frankly, you're not the first person without experience coding games to make that claim. I'll give you this much: at least you haven't mentioned inheritance yet although you came close by mentioning objects. Powers are actually defined in an Excel spreadsheet that is then imported into custom database tables that are loaded into the game engine. This is common knowledge among players that keep track of such things.


1. I've been discussing the game implementation with the devs longer than most people have been playing the game now.

2. In particular, I discussed the tohit system with pohsyb until I was blue in the face back in I6 running up to the I7 accuracy changes. I actually posted a fairly detailed simplified algorithm for it back then, and I know quite a bit more since then.

3. All of the information I've provided here is also generally known and has been discussed on the forums in the past. I'm not the only player who could have answered your questions. I can think of a dozen more that could off the top of my head.

4. One third of what I know comes directly from the devs. One third comes from extremely systematic testing of the game itself. One third comes from basically remembering everything every other reputable player has ever discovered and described. That makes my batting average when discussing game implementation very, very high.


I'm suggesting not portraying certainty when you're actually wrong. It hurts credibility. That's why you should never guess, and really never guess and think your guesses couldn't possibly be wrong. Arcana's rule number one: all uninformed** guesses about City of Heroes game implementation are wrong. I've been saying that since 2006 or so, and so far no one has ever managed to break that rule.


** The only guesses that have ever been even partially right have been either informed by the devs, or informed through direct and meticulous testing of the game engine. Guesses based on textbooks are batting zero for the century.

Oh my god.


Word of advice: if you don't know, its probably best not to guess. Neither of these approaches would work for reasons having to do with a couple of misunderstandings you have about the way the game is coded and implemented in data.

There's a completely different way to implement such an idea in theory, which I'm not going to describe in detail because I think its an especially bad idea. But the first attempt cannot succeed because power data is not hard coded into the game engine, thus the tohit system has no way to know when a "tier 1" attack is being used: the code cannot reference that fact, ever. The second attempt fails because there's no way to signal the miss from one entity to the other, and no way to spawn the entity only on a miss.

I definitely think we need to take your catch-phrase to the next level.

This seems to be a compromise of sanity more than anything else.

More the former than the latter. But technically, it will be the enemies that were near him relative to where the server thought he was at the moment of power activation, not where he appeared to be in your client.

I'll check it when I get home. Maybe something invalidated it recently that I didn't notice.

Edit: Ok, so the AE did one of its "I miss you, please come back and edit something" things. The Scrapper Challenge mission was flagged invalid, but for no actual reason: I hit edit, there was no problems listed, I republished, and it was back up. I should probably figure out why it occasionally does that, but eh, lots of other things to think about at the moment.

I haven't played it recently myself, so if an AE change turned all the fire blasters into Smurfs or the Elite Boss is now a cross-dressing robotic plant controller, let me know. So I can download it and save it.

FYI I removed the listing from my sig to make room for my shameless advertisement of my favorite web comic, but for those wanting to play it the arc is ArcID 9713: Scrapper Or Anything Else Challenge Mission. I keep meaning to update it, but I keep getting distracted from doing so.


I haven't heard of Aeon's Army: going to have to play that on the weekend. But as to how brutal the Scrapper Challenge mission is, nothing beats what it looked like in I14 beta, before they put in controls to make it impossible to make missions like I was making (I swear, half the AE limits were designed explicitly to reign me in). I have only one visual record of what one early version of it looked like, I recorded pohsyb attempting it with his mega-mega test server controller (which has basically every controller primary and secondary power).

Funny but true: pohsyb thought I was going to help him. I was actually standing back and watching to see how far he would get. Like heck I was going to draw aggro in this monstrosity, although I eventually did try at the end (and lasted all of about thirty seconds while eluded). I was actually surprised he lasted that long.

Why I'm getting shoved around during that video? The original versions of the "Scrapper" Challenge specifically had confuse in them to deal with people who attempted to bring controllers into the thing.

rroll = rand(iseed) * 100.0
if (rroll < 0.95) {
grue.eaten(self)
}

Oh sure, he tells you, you tell someone else, and the next thing you know:




You do realize this is a forum for a rated T for teen game, right?


I don't know about multiple inheritance, but I do know the knockback code has been multiply disowned. And the animation code has multiple personality disorder.