Get rid of the damn tohit cap

[Akisan]

Still not sold on it being a good idea to bump to-hit to 100.  But that is a good mechanic for glancing blows, @Purrfekshawn. Something along the lines of 135% to-hit means there's a 35% chance for a glancing blow (partial damage, may or may not proc debuffs/mezzes) if the attack *does* miss. 

 

Such a system could also be an interesting counterpoint to enemies with overwhelming defense.  As an example, say, if you're fighting the Nemesis army, and someone popped all the lieutenants early, Vengeance means everybody has a chance to hit in the teens.  But if your to-hit bonus (not including their defense) is still high enough, you would still land glancing blows (probably a lot of them, since you have an 85+% miss chance with Vengeance stacked so high).  It'd give a reason to keep attacking at such low odds, instead of disengaging and waiting out Vengeance.

[Vanden]

1 hour ago, aethereal said:

It is impossible to miss 8 times in a row if your hit chance is 95%.  The streakbreaker kicks in after one miss if your chance to hit is 90%+.  If you saw a multiple miss streak, that necessarily means that you did not have capped to hit chance.

True, but you don't necessarily care about every attack you use. It's possible to miss with your Big Attack, like Assassin's Strike for instance, use other attacks while the Big Attack is recharging, and miss again once it's up. That still feels like two misses in a row, and there's nothing the streakbreaker can do about that. And since the streakbreaker exists entirely to placate bad player feelings about missing, it's a notable deficiency.

Edited November 26, 2020 by Vanden

[Ghost]

5 hours ago, thunderforce said:

I'm not sure anyone has, even the OP (who may or may not be a "he").

 

Individual players can only really talk about what they would like to change. Then there's a discussion to find out if other players like that, or if some similar idea is more widely acceptable. The OP has done nothing wrong by talking about something they'd like to change.

You may be right.

Maybe I misunderstood the intent of his/her post

 

The title of the thread didnt make it seem as if the op was searching for opinion - it seemed like a stated fact/demand

Followed by the his/her third post, which included calling someone a name.

Just seemed to me like the person was angry, and demanding change.

 

We all know its hard to read intent on the internet - maybe I was wrong.

 

Edited November 26, 2020 by Ghost

[Akisan]

Related to the glancing blows above, potential chance to crit was mentioned earlier in the thread. 

 

Perhaps if they were tied to -def, that is, if your raw chance to hit is above 100% without any tohit or acc bonuses (so, at least -25% def on most enemies), then you get a crit chance based on how far down their defenses are.  For example (and a 1:1 def:crit scale), a Rad/Devices blaster has landed enough hits to get her enemy's def to -40%.  She (or anyone else, for that matter) would have a 15% chance (75+40=115) to land a critical hit with any attack.

 

Some offset for mezzes could be taken into account for this (like being held could count as -15% def, but only for crit chance calculation), though it would likely require a re-balancing of certain sets, like Earth Control.  Such a system could make for some interesting slotting decisions - for the blaster above, would it be worth more for her to slot for raw damage, or splash -def for extra crit chance?

[thunderforce]

13 hours ago, srmalloy said:

Or, to slant your argument differently, increasing your chance to hit from 5% to 6%, you reduce your chance of missing by about 1%; increase your chance to hit from 95% to 96%, and you reduce your chance of missing by 20%. Statistics can be so endlessly flexible, able to be twisted in ways that can create almost any impression you want.

This is probability, not statistics; and it would be more accurate to say you can pull the wool over the eyes of people who don't understand probability. For example, here, you are producing a figure which is numerically correct but not meaningful. Vanden's figures give a meaningful view of the change in effectiveness.

[LQT]

On 11/26/2020 at 2:05 PM, Ghost said:

The OP wants to change something because it frustrates HIM

 

The OP isn't the only one who has addressed this. I have. Others have. For you to state otherwise means to ignore those posts which makes thread participation questionable outside of trolling and/or ignorance.

 

On 11/26/2020 at 2:05 PM, Ghost said:

I want my tanks to be impervious to damage - after all, that's what they are supposed to be

I want my blasters to kill everything with one shot 

My stalker to never ever be seen.  If I want to stealth past level 50's at leave 2, I should be able to.  Its what I want. 

Swords.  If I wield a sword, it shoulda automatically kill anything it hits - its a sword after all.  

These are hyperbolic strawman arguments that are neither on topic for the thread nor backed up in any way whatsoever as to why the need looked at.

The "functional d20 system" the game uses for managing attacks does have flaws which the OP, myself, and others have described in this thread a few times now as well as providing examples where extremely bad luck can occur. However posts like yours aren't constructive in the slightest and aren't adding anything to the discussion of the game mechanic in question in any way.
 

19 hours ago, VV said:

That's a lot of hate for something that only happens once out of 20 times. Seriously, I hardly notice. If your build requires the need to hit every single time, it's a bad build, that's all. You know you're gonna miss every now and then, take that into account when you build.

The checks are independent of each other, so that can and does happen far more often than 1 out of 20 times when at 95% chance to hit.

Check previous posts myself and others have made in this thread regarding attack chains that end up as 50% streakbreaker because each non-streakbreaker attack misses.

I'm actually back to this thread for such a reason as I'm about to post a combat log of such an attack chain.
 

You, or others, not noticing is a topic already discussed in this thread some pages back regarding observation bias towards certain attacks due to their significance for certain ATs.

As for builds requiring the need to hit every single time, that's literally how certain powersets work as build/spend mechanics. It's also how certain ATs work.

Are you going to call Stalkers a bad AT because Stalker's Focus needs hits to land for it's chance to build a charge? Or any of the sets that need attacks to land such as Beam Rifle [Disintegrate], Staff Fighting, Dual Blades, or Street Justice?

 

If nothing else your post going over these already discussed topics means you've done the same thing as the other poster I've replied to and not actually read the discussion in the thread, instead erroneously commenting on Vanden's OP and ignoring what followed from others.
 

Back on topic:
 

Be me: Decide to work on Defeat badges. Go to Grandville. Search for Toxic Tarantulas. Find 2 that are level 42, I'm a 50+1 fully t4'd incarnate Blaster. Attack. Get this.
 

Quote

[15:50:19] MISSED Toxic Tarantula!! Your Blaze power had a 95.00% chance to hit, you rolled a 98.15.
[15:50:20] HIT Toxic Tarantula! Your Taser power was forced to hit by streakbreaker.
[15:50:21] MISSED Toxic Tarantula!! Your Blazing Bolt power had a 95.00% chance to hit, you rolled a 97.68.
[15:50:21] Toxic Tarantula MISSES! Claw Pierce power had a 6.50% chance to hit, but rolled a 98.94.
[15:50:23] HIT Toxic Tarantula! Your Blaze power was forced to hit by streakbreaker.

Pic attached for proof.

As this post was taking so long to type up I continued doing stuff in game. Had this happen just now again.
 

Quote

[17:03:27] MISSED Miss Thystle?!! Your Blazing Bolt power had a 95.00% chance to hit, you rolled a 97.46.
[17:03:30] HIT Miss Thystle?! Your Blaze power was forced to hit by streakbreaker.
[17:03:33] Soldier Automaton MISSES! Nemesis Rifle power had a 5.75% chance to hit, but rolled a 12.75.
[17:03:33] Miss Thystle? MISSES! Hurl power had a 5.75% chance to hit, but rolled a 47.52.
[17:03:33] MISSED Miss Thystle?!! Your Blazing Bolt power had a 95.00% chance to hit, you rolled a 98.28.
[17:03:35] HIT Miss Thystle?! Your Blaze power was forced to hit by streakbreaker.
[17:03:38] Soldier Automaton MISSES! Nemesis Rifle power had a 5.75% chance to hit, but rolled a 83.31.
[17:03:38] Support Automaton HITS you! Storm Rifle power had a 5.97% chance to hit and rolled a 2.87.
[17:03:39] HIT Soldier Automaton! Your Blazing Bolt power had a 95.00% chance to hit, you rolled a 46.41.

Reason for that combat: Was at extreme range picking off an approaching ambush hence only using snipe and 1 other power.

Pic attached.
 

As mentioned, this happens because each roll after streakbreaker is made independently of each other.

So if you end up with: Miss, streakbreaker, Miss, streakbreaker, there is nothing stopping you from missing a third time, as happened in the previous post I made here, because each miss doesn't account for any previous ones, regardless of how immediate they may have been.

Some "fun" times this has caused less than enjoyable turnouts: Successive misses/SB'ers vs the bunker doors during a Keyes iTrial badge run.
 

19 hours ago, Ghost said:

You may be right.

Maybe I misunderstood the intent of his/her post

maybe I was wrong.

 

That's three different ways to avoid admitting you were wrong given that language choice of "may/maybe" in each of those sentences. You clearly didn't read the thread going by your earlier post, and now you're using that very ignorance to avoid admitting you snapped at the OP needlessly.

[Jacke]

4 hours ago, LQT said:

So if you end up with: Miss, streakbreaker, Miss, streakbreaker, there is nothing stopping you from missing a third time, as happened in the previous post I made here, because each miss doesn't account for any previous ones, regardless of how immediate they may have been.

Some "fun" times this has caused less than enjoyable turnouts: Successive misses/SB'ers vs the bunker doors during a Keyes iTrial badge run.

 

That sequence of misses at a supposed "95%" chance to hit where the only hits are forced by the Streak Breaker makes me think there are things wrong with the combat mechanics at some times.

[aethereal]

18 minutes ago, Jacke said:

 

That sequence of misses at a supposed "95%" chance to hit where the only hits are forced by the Streak Breaker makes me think there are things wrong with the combat mechanics at some times.

The odds of three 5% events in a row, if they are independent of each other, are only 1 in 8,000.  Certainly the audience of this thread alone have made several orders of magnitude more attacks than that on HC, so it doesn't seem particularly unlikely that someone has noticed one of the doubtless many times that a three-miss-(ignoring-streakbreaker-attacks)-streak has happened to them.

 

Now, an 8 miss streak like someone suggested above, that's a different story, and feels unlikely to have happened if misses are uncorrelated.  (It's 1 in 25,600,000,000).

[Ghost]

7 hours ago, LQT said:

 

The OP isn't the only one who has addressed this. I have. Others have. For you to state otherwise means to ignore those posts which makes thread participation questionable outside of trolling and/or ignorance.

 

These are hyperbolic strawman arguments that are neither on topic for the thread nor backed up in any way whatsoever as to why the need looked at.

The "functional d20 system" the game uses for managing attacks does have flaws which the OP, myself, and others have described in this thread a few times now as well as providing examples where extremely bad luck can occur. However posts like yours aren't constructive in the slightest and aren't adding anything to the discussion of the game mechanic in question in any way.
 

 

 

 

 

 

You sure are proud of yourself.

It was the way the OP post was stated, that I took issue with - just as you are not happy with my post, I was not happy with his.

 

The point of the "hyperbolic strawman comments" were once again in response to what I perceived as a spoiled brat wanting his way.

They were meant to be  sarcastic.  Sorry you missed it. 

 

I have already acknowledged that I may have misread into meaning.  Not sure what more you want - but that's probably because of my ignorance.  

Maybe send me your address and Ill have a quart of my blood shipped to you ASAP.  

 

 

 

  

Edited November 28, 2020 by Ghost

[Grouchybeast]

I would miss the chance to miss.

 

Without that 5% possibility of a NOOOOOOOOOO! I would find landing the big hits to be that much less satisfying.

[Ghost]

3 minutes ago, Grouchybeast said:

I would miss the chance to miss.

 

Without that 5% possibility of a NOOOOOOOOOO! I would find landing the big hits to be that much less satisfying.

I dont understand what the fun would be, knowing that you would hit every time.

 

Edited November 28, 2020 by Ghost

[KingofMonkeys]

I'd be okay if they brought it up from 95% to 99%

 

Your superpowered entity will still miss the occasional punch/shot, but it would also be less of "Oh, hi RNG"

Edited November 28, 2020 by KingofMonkeys

[Doomguide2005]

12 hours ago, KingofMonkeys said:

I'd be okay if they brought it up from 95% to 99%

 

Your superpowered entity will still miss the occasional punch/shot, but it would also be less of "Oh, hi RNG"

Oh if that were only true.  I think that on players it bothers as much as the OP it might in fact be worse.  I can only imagine the consternation of missing that crucial attack when I only have a mere 1% miss chance ... nevermind missing 2 in a row or 3 both of which are going to happen eventually with 100's of players making 1000's of rolls.  The first (missing twice) is only a slightly less common occurrence as the 3 in a row sequence noted above.  

 

 

[srmalloy]

16 hours ago, aethereal said:

The odds of three 5% events in a row, if they are independent of each other, are only 1 in 8,000.  Certainly the audience of this thread alone have made several orders of magnitude more attacks than that on HC, so it doesn't seem particularly unlikely that someone has noticed one of the doubtless many times that a three-miss-(ignoring-streakbreaker-attacks)-streak has happened to them.

Observer bias, yes. You don't notice all the times when things happen the way you expect them to, but the weird outliers jump out in your face. Something that might be worthwhile to do is to go into the source code, extract the function for the RNG, and set it up as a standalone program and run a few million numbers to look at the distribution to see if the RNG itself has an issue.

 

Some RNG functions have odd biases built into them because the developers don't think about the data they're working with. RNGs produce a binary number (32- or 64-bit, typically), and one of common way of getting a random number in a range is to mod the random number by the range (i.e., X mod 10 + 1 to get 1-10). But this introduces bias; mod 10 means the bottom four bits of the random number are all that determine the result, so you have 0-15, producing 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 0, 1, 2, 3, 4, and 5 when you apply the mod function -- 0-5 appear twice as often as 6-9. This effect gets smaller as the range goes up -- a 1-1000 range (10 bits, 0-1023) only doubles up on 2.4% of the range, compared to 60% for 1-10. I would hope, though, that the original CoH developers accounted for things like this; RNGs in games have been around for a long time, and solutions to this problem should be routine.

[Eclipse.]

On 11/28/2020 at 7:55 AM, srmalloy said:

 Something that might be worthwhile to do is to go into the source code, extract the function for the RNG, and set it up as a standalone program and run a few million numbers to look at the distribution to see if the RNG itself has an issue.

 

While this sounds all fine and good, it might not matter if what ever The Random function uses a seed  on the server that has more/less entropy than the seed does on your local hosted version.

IIRC, though, the source code uses a fairly basic pseudo-random generator that relies on os.urandom (it uses the default rand() in python). All fine and good (if "predictable") as long as the seed is set after the servers have been up awhile,  less brilliant if the seed is set shortly after the servers were spooled up. I don't remember ever seeing it being seeded differently than default. Working, no access to source right now.

Edited December 4, 2020 by Eclipse.

[srmalloy]

Seed initialization is the downfall of a lot of RNG algorithms, true enough, and they can be much harder to fix -- the game servers are likely going to be brought up as quickly as possible after a server restart, so that's going to be resistant to fixes.

[aethereal]

I think you're misunderstanding the role of seed generation in problems of randomness.  Bad seed generation creates security flaws through predictability, which shouldn't be relevant here.  It should not result in lopsided distributions of numbers.

 

There was a thread a ways back in which someone parsed a long combat log for to hit values.  Distribution of values looked pretty even, which is what I'd expect.  Now, the more subtle problem could be streakiness.  That is, you might have a situation where the probability of rolling >95 is very close to 5%, but where the probability of rolling >95 conditional on the last roll being >95 was >5%, and the probability of rolling >95 conditional on the last roll being <95 was <5%.

 

But the parsimonious explanation here is human memory bias.

[MsSmart]

I hate to tell you this...

 

But your real enemy is the defective number generator or so called range excuse used, so no matter how good your accuracy is, you going to miss far more than the so called 5% chance of missing from time to time, it seems to appear the number generator gets stuck rolling high at times, and at others it is just fine. I truly wish, there was a way for the player to nudge the random generator when it is stuck rolling high. In Dark Age of Camelot, they listened to the players, and set a command where they could change the "seed" that fed the random number generator, maybe that is what this game needs as well.

 

Sue

[tidge]

I completely accept that there is an effect of observer bias, BUT.... blaming the seed is only a small part of the issue. The following discussion is shortcutting to the corner case of always having a 95% chance to hit, with no auras or the like.

 

A slightly larger issue to me is that the RNG is trying to populate the probability space (0 < p < 1) in a FLAT manner, so the chance of a following a 1-in-20 roll with a second 1-in-20 roll is almost certainly NOT 5%. My (slightly informed) WAG is that it is more like a 4.5% chance.  However, the streakbreaker algorithm ignores the followup roll to a 1-in-20 event. In other words, it is HIGHLY UNLIKELY that there actually was a 'streak' that needed to be 'broken'.

 

I'd be much happier with streakbreaker if it actually fired when it needed to  (to 'break a streak') and not preemptively fire. This would be a PITA to track/code, so I am not surprised it doesn't work this way.

 

I'm actually curious what the effect would be if we got rid of streakbreaker altogether. From my PoV its primary purpose is to assuage 'hard feelings' about the RNG, but since it obscures RNG results (the rolls from a streakbreaker 'hit' are not displayed) I feel like it is doing more harm than good....especially for characters that regularly run with 95% final toHit chances.

 

Some tangential comments about RNG visibility in the game:

Spoiler

One area where it is possible to observe the RNG behavior that is/should be unaffected by streakbreaker is with %procs in attacks. Consider:

1. Multiple %procs can be added to a power
2. %procs will not fire on a target if you miss that target
3. %procs can be clamped at a 90% (ceiling) chance of firing
4. %procs can be put in AoE to target many enemies
5. %proc results are displayed in the combat logs
6. %proc chances are independent of each other

For all my griefing about the ToHit algorithm, RNG and streakbreaker... I absolutely do not believe that the RNG is working against the %proc chances. For all the potential for observer bias, I have plenty of characters with clamped 90% proc chances and for them, and I have never felt that RNG was treating the proc chances unevenly.

 

 

[LQT]

"This is fine."
 

Quote

 

[16:00:36] MISSED Guardian of the Amulet!! Your Gloom power had a 95.00% chance to hit, you rolled a 95.81.
[16:00:37] HIT Guardian of the Amulet! Your Storm Kick power was forced to hit by streakbreaker.
[16:00:48] MISSED Guardian of the Amulet!! Your Eagles Claw power had a 95.00% chance to hit, you rolled a 98.32.
[16:00:51] HIT Guardian of the Amulet! Your Storm Kick power was forced to hit by streakbreaker.
 

[16:01:01] MISSED Behemoth!! Your Storm Kick power had a 95.00% chance to hit, you rolled a 97.60.
[16:01:02] HIT Behemoth! Your Crane Kick power was forced to hit by streakbreaker.
[16:01:11] MISSED Guide!! Your Storm Kick power had a 95.00% chance to hit, you rolled a 98.81.
[16:01:12] HIT Guardian of the Amulet! Your Crane Kick power was forced to hit by streakbreaker.

 

2 sets of 2 misses/2 streakbreakers within 40 seconds, one being back to back usage of Storm Kick, taken specifically due to the +Def function, so that's fun to have missing for reasons outwith my control.
 

Something else, albeit negligible; technically the hit cap seems to be 94.99% due to how the game calculates a miss. 95.00% can and will miss. Given the roll range goes up to 100.00 that meanst 501 data points between 95.00 and 100.00, not 500 as would be equal to 5%.

Or the two sigfigs is just how far it will display things on the UI and so 3 sigfigs or more will count as a miss but never be displayed.

Pre-post edit:

Amusingly enough before posting this I decided to play the toon the above data is from some more. Then this happened.
 

Quote

[16:22:42] MISSED Headman Gunman!! Your Crane Kick power had a 95.00% chance to hit, you rolled a 95.00.

¯\_(ツ)_/¯