Game Mechanics: Feign Death Success Rate

[Raev]

I want to preface this post by saying that I don't know what the true FD success rate should be. The best I could find was these patch notes (implemented here in 2011:

- The Monk's feign death success rate has been improved. It was previously capped at 95% at 100 skill and now it ramps up the success rate for higher skill levels.

Anyway, I feel anecdotally like I have been failing FD quite a bit recently. Today I tried to CR a group from Disco 2, and I failed first directly under gruplinort (somehow I survived) and then again under the commander (had to cap). Now when I first made Raev I made an FD hotkey like so:

/autoattack off
/doability 1
/t raev blah

which produces the message 'Talking to yourself again?' in the log file. I added this because I wanted to see how badly I was failing to get FD off vs casters and never took it out. Since I don't talk to myself normally, pretty much every such message is an FD attempt.

zonko@vanderboom:~/.wine/drive_c/Program Files/Sony/EverQuest/Logs$ cat eqlog_Raev_project1999.txt | grep "Raev has fallen to the ground" | wc -l
145
zonko@vanderboom:~/.wine/drive_c/Program Files/Sony/EverQuest/Logs$ cat eqlog_Raev_project1999.txt | grep "Talking to yourself again" | wc -l
1889

145 / 1889 = 7.7% chance to fail. However, sometimes I hit my hotkey multiple times like a scrub when the timer hasn't expired yet. So my real FD failure rate is probably mildly above 8%. Wolfram Alpha is quite sure this is not chance, giving it a probability of less than 1 in a million if the real probability is 5%.

Doing the same thing for my beta logs:

zonko@vanderboom: Logs$ cat eqlog_Raev_P1999Beta.txt | grep "Raev has fallen to the ground" | wc -l
30
zonko@vanderboom: Logs$ cat eqlog_Raev_P1999Beta.txt | grep "Talking to yourself again" | wc -l
348

That gives 30/348 = 8.5% which is even higher, although Wolfram Alpha suggests there is a .2% chance that it could be the result of random chance.

The real question, of course, is what the true chance should be, which is why I didn't title the post 'FD IS BROKEN'. The patch notes seem pretty reasonable to me, but IIRC Cobblestone said people did real tests but he lost the files. Perhaps one of our classic afficionandos

[Pringles]

cat to grep? [You must be logged in to view images. Log in or Register.]

[Crom]

think server should be using this version from 1999

> November 16th, 4:30pm
> ---------------------
>
> Feign Death was recently changed by the Tuning Team to reflect evolving
> gameplay issues. Since two of the three classes that get Feign Death get
it
> through spells, I thought this might be a good place to explain the
changes.
>
> Previously, when you used Feign Death (and assuming you cast the spell
> successfully or, in the case of the Monk, made your skill roll) hostile
> monsters would decided that you were dead and wander off. When you later
> got up, they would still remember that they were angry with you. If you
> were within sight, they would immediately come after you. If not, they
> would come after you at some later point. This might happen within a few
> minutes, or could happen even a few hours later. The only way to be
> absolutely sure that a monster had forgotten about you was to zone.
>
> With our new changes, about a third of the time that you successfully
Feign
> Death, the monster will immediately forget about you, similar to an
> Enchanter casting Memory Blur. If they do NOT forget about you, they will
> immediately come after you when you stand up. No more monsters coming
back
> at irregular amounts of time - it should be immediate or not at all.
>
> This is a change - the old way, if you Feigned Death a SINGLE TIME, the
> monster would ALWAYS remember you. This way, they remember you about
> two-thirds of the time.
>
> Using the old Feign Death, you could Feign Death multiple times to try to
> confuse the monster. After about seven or eight times doing this, most
> monsters would eventually forget they were mad at you.
>
> With the new Feign Death, if you Feign Death once the monster has about a
65
> percent chance of remembering you. Twice and the monster has about a 50
> percent chance of remembering you. Four times means about 20 percent, and
> six times means about 8 percent. This is very similar to the old Feign
> Death.
>
> Another addition to the new Feign Death involves a monster "going home."
> When you Feign Death, most monsters will return to where they were before
> the fight. If a monster returns to its initial location, it will
> immediately forget about you greater then 90 percent of the time.
> Therefore, if you Feign Death and stay down long enough for a monster to
"go
> home," you should be safe when you stand up a majority of the time. If
the
> monster IS still mad at you, it will return immediately, rather then
> wandering back later at some irregular interval.
>
>
> *FEIGN DEATH UPDATE*
>
> After further testing and watching these changes on the Live verses the
Test
> Server, we will bechanging Feign Death again on the next patch.
>
> New changes will make it so that if the creature is less than level 35,
the
> monster will always forget about you. If the creature greater then level
35,
> then the monster will use the changes detailed above.
>
> - The EverQuest Team


With this update:

FEBRUARY 17, 2000
Announcements:

The City of Paineel, home to the infamous Heretics of Erudin, is now available via the Optional Patch. In addition, Erudites may now also be Shadowknights in addition to the other classes available to them. All Erudite Necromancers, Shadowknights, and Clerics who worship the God of Fear, Cazic-Thule, now start in the City of Paineel. As such, if you wish to create a character of this type from this point forward, you *must* download the Paineel zone-files via the Optional Patch.

Please note that if you downloaded the Paineel zone files recently, you must run the Optional Patch again to insure you have the correct version of the files.

In related news, all of the entrances to the Qeynos Catacombs have been fixed.
Spell Changes:

The recast delay for many of the Shaman buff-spells have been reduced.
"Reoccurring Amnesia" has had its casting time reduced.
"Yaulp" and "Augmentation" should now stack.
"Solidarity of Vision" has had its duration increased.
The Druid spell, "Skin Like Nature", has been returned to its former status due to popular opinion. As such, the damage-shield component has been removed and the regeneration component has been re-applied.
"Sense Magic" has been replaced in all cases by the "True North" spell. If you have already inscribed the "Sense Magic" spell, the spell icon will now automatically reflect "True North".
A bug that existed with several damage-shield spells, causing their damage to be subtracted rather than added when stacked, has been corrected. The spells will now stack properly.
The "Feign Death" spell, when cast by a Necromancer, will now have a fixed duration. The Necromancer will receive several warnings prior to the spell wearing off. This change was necessary because some Necromancers found a way to gain experience through their pet while being AFK for an extended duration, such as being in bed for the night. Note: This does not affect Shadowknights or Monks.
New Necromancer spells are now in game. Check your spell vendors.
Alchemy has been revamped. Check with your local Shaman.
DOT (Damage Over Time) spells will no longer hinder camping.

[Daldaen]

That is some solid information, however that speaks to the rate of memblur on FDs.

Not the rate of FD succeeding (IE no fallen to the ground).

Raev, your logs mention you made that Hotkey early on. Think many or some of your FDs are from before you have 100 skill? Those may throw it off. If you perma-log just search for when you got 100 FD skill and remove all "you can't talk to yourself" logs from before that point.

I doubt it would majorly affect your rate of success anyways, but it's worth looking at. Also the way I read that patch note is weird. It says 95% chance at 100 skill and even better chance as it increases. Does that mean skill 100-225 are irrelevant, and the 95% success chance is capped at skill 100? Should FD even go beyond skill 100 in classic?

[Haynar]

The increasing chance by reflopping is coded. Not patched yet.

Applying statistics to the RNG here will drive you batty. Applying to eqlive will make you insane. As random as it is, its not perfect. But here its not near as streaky as eqlive.

Effective skill is capped at 100, and was like that until 2004.

Its actually coded to 95.2%, but streaky RNG will make it seem off.

H

[Raev]

Originally Posted by Haynar [You must be logged in to view images. Log in or Register.]

The increasing chance by reflopping is coded. Not patched yet.

Nice!

Originally Posted by Haynar

Its actually coded to 95.2%, but streaky RNG will make it seem off.

I checked older log files and its definitely 95% there. I also checked before and after the 6/24 patch and the rates seemed unchanged. So I suppose its not impossible that I have been really truly insanely unlucky. I am not sure how a streaky RNG would affect this; even if the RNG was implemented by a moron as "return counter++ % 100" there would be so many RNG calls between my FD attempts that I would expect them to be nearly random anyway. So color me confused.

I will back up this log file and start a new one to see if I revert to 5% fails over the next week or two.

Thanks for the quick response Haynar.

[Kender]

Originally Posted by Haynar [You must be logged in to view images. Log in or Register.]

The increasing chance by reflopping is coded. Not patched yet.

Applying statistics to the RNG here will drive you batty. Applying to eqlive will make you insane. As random as it is, its not perfect. But here its not near as streaky as eqlive.

Effective skill is capped at 100, and was like that until 2004.

Its actually coded to 95.2%, but streaky RNG will make it seem off.

H

the chance shouldnt change. the chance is always 35%.
it's the accumatated chances that mean the likelyhood of a memblur event will occur that works out at 50% 80% etc

Their numbers are a little off though

The chance of at least 1 mem blur calc...

failed blur chance .65 so the formula is 1 - .65(to the power of n) where n is the number of flops

results (rounded to 2 decimal places)

1 = 0.35 (1-.65*.65)
2 = 0.58 (1-.65*.65*.65)
3 = 0.73 (1-.65*.65*.65*.65)
4 = 0.82 (1-.65*.65*.65*.65*.65)
5 = 0.88 (1-.65*.65*.65*.65*.65*.65)
6 = 0.92 (1-.65*.65*.65*.65*.65*.65*.65)
7 = 0.95 (1-.65*.65*.65*.65*.65*.65*.65*.65)
8 = 0.97 (1-.65*.65*.65*.65*.65*.65*.65*.65*.65)

[Haynar]

Originally Posted by Kender [You must be logged in to view images. Log in or Register.]

the chance shouldnt change. the chance is always 35%.
it's the accumatated chances that mean the likelyhood of a memblur event will occur that works out at 50% 80% etc

Their numbers are a little off though

The chance of at least 1 mem blur calc...

failed blur chance .65 so the formula is 1 - .65(to the power of n) where n is the number of flops

results (rounded to 2 decimal places)

1 = 0.35 (1-.65*.65)
2 = 0.58 (1-.65*.65*.65)
3 = 0.73 (1-.65*.65*.65*.65)
4 = 0.82 (1-.65*.65*.65*.65*.65)
5 = 0.88 (1-.65*.65*.65*.65*.65*.65)
6 = 0.92 (1-.65*.65*.65*.65*.65*.65*.65)
7 = 0.95 (1-.65*.65*.65*.65*.65*.65*.65*.65)
8 = 0.97 (1-.65*.65*.65*.65*.65*.65*.65*.65*.65)

Thanks for you input. And this describes what they refer to as the old FD. Where each chance is the same. Their new description is not the same, but at higher attempts the results do converge.

I do not believe they did some hand waving to give the illusion that the mechanics changed.

H

[kaev]

Originally Posted by Haynar [You must be logged in to view images. Log in or Register.]

Thanks for you input. And this describes what they refer to as the old FD. Where each chance is the same. Their new description is not the same, but at higher attempts the results do converge.

I do not believe they did some hand waving to give the illusion that the mechanics changed.

H

Haynar I love your work, dont' be mad at me for disagreeing with you, but if my reading comprehension did not fail me on that patchnote, what Kender posted is what they describe as the new mechanic: 35% chance of the monk being forgotten each and every time he feigns(*). The old mechanic is not described in explicit detail, except for the bit about a single feign death never being enough to get the mob to forget you.

(*) Also the 90% chance of forgetting the monk when the mob resets is part of the new mechanic and not part of the old.

Old mechanic:
- First feign NEVER makes the mob forget.
- Additional feigns may make the mob forget, % chance not specified.
- Mob will delay random amount of time before coming after the monk if he remembers him

New mechanic:
- First feign has 35% chance of making the mob forget the monk
- Additional feigns also have 35% chance of making the mob forget the monk
- Mob reset while feigned has 90% chance of making the mob forget the monk
- Mob will come after monk immediately if he remembers him

[Haynar]

Originally Posted by kaev [You must be logged in to view images. Log in or Register.]

(*) Also the 90% chance of forgetting the monk when the mob resets is part of the new mechanic and not part of the old.

I never saw this. Ever. When a mob was reset totally, including rotating to the position they are guarding, all aggro was lost.

Being an avid SEQ user who played a class that could FD, you could use it to watch them rotating. And when they stopped you always lost aggro.

There are also other mechanics which they describe as monster not seeing you. This was not described in the new system. This was one of the ways that the instacast CoS was overpowered. You could FD, and if a monster did not see thru invis, you could stand and cast invis and almost always lose aggro. Darn near 100% of the time if you were fast. You typically had about 1/2 a sec, before the reaggro happened.

So there is more to what they describe.

More monks that I have talked to, think it is the progressive chance that was what they meant, where the chances increase. Much of this based on doing mob splits in velious.

So in summary. We can agree to disagree on some aspects. Even though the math can be made to reproduce the numbers. From my testing working on splitting mobs, it seemed like a closer feel to what I remembered. But mileage may vary.

If I get a lot of negative feedback, then I will change it. But right now, the feedback I have gotten is the current way does not feel right. And it is set at about 34% chance per FD, with no increasing chance. We never had it be like the original FD was, as they described.

There was also not a description of the chance mobs had to forget you, if they were roaming.

Did you ever play in the Umbral Plains in luclin? OMG, totally different mechanics there of how aggro was passed and FD worked.

The funnest aggro zone to figure out was Grieg's End. Did you know traveling in the group, the first person they would go after, was always the last person to zone in? Was awesome, when it was the new guy, and they could never figure out what was going on.

H