90% chance my butt

SolidusPrime

The frequency of pickpocket failures at a 90% success chance in Vvardenfell seemed like it was way off, so I started collecting a little data. Any time an NPC displayed a 90% chance to succeed at pickpocketing, I would record rather the attempt was a success or failure on a spread sheet. This character has maxed out Legerdemain. All attempts were made while the NPC was standing still, from a close enough distance that my character did not move when initiating the attempt.

300 attempts
103 failures
197 successes

I'm going to collect another 300 results while my friend does Hew's Bane. I never seem to have this many failures outside of Vvardenfell though, and I've been stealing since the Thieve's Guild was released. Think I'm just unlucky or what?

disintegr8

I have only recently gotten into TG and DB but I was initially getting peeved off with pickpocketing early on due to 90% and 95% chance attempts failing.

Now I don't care if I get caught or not, I just move to the next victim and run away from guards if I get caught. I am sure if I kept stats on every % chance and counted my success or failures, they would not tally up but it's only a game.

Chronicburn

Something feels fishy about the 90%-95% picks to me also ... even in other zones. Thanks for compiling the stats on this. I'd be interested to see your next 300.

For me, I just assume I'm gonna get caught, so I have a dedicated thief character. I just fill my pockets, run up a bounty, and then log until the bounty is gone. The fence limit is daily anyway so it's no big loss...

DeathHouseInc

...90% of what?

Yakidafi

DeathHouseInc wrote: »

...90% of what?

Of failure..

Ulfgarde

that's what you get for running 0 in luck smh

NetViperX

There is definitely an issue with the percentages. I regularly fail multiple times in a row at 90%, and almost always at 75% or lower, and that is with maxed out Leger and Thieves Guild skills. Lately if I am trying to keep bounty to a minimum I only steal at 100%, and just use the shank on anything lower.

Runschei

I rarely get caught while doing my daily route in Wayrest. But interesting results you have there

Minyassa

So it's NOT just me!!! I've gotten so I don't dare even try it if it's less than 100% because 90% has always meant, for me, a 100% chance of getting caught. I know that RNGesus hates me but I thought that was getting kind of ridiculous. Maybe something's broken.

Rohamad_Ali

That's closer to 75% on average . @ZOS_GinaBruno Would you be so kind to have someone look into this ?

Zbigb4life

I haven't noticed it before, but now you mention it, you're quite right.

Asardes

I've noticed it matters a lot how the eye reticle looks. If it's almost closed then the displayed pickpocketting chance is accurate. If it's more open it decreases and you can get caught even with 95%. So try to wear 7 medium and maybe one of the sets that decreases detect radius like Night Mother's Embrace. Also even if the chance is 100% wait until the chime when the number turns green to press the button. I do get caught sometimes and stack bounty but do that part at the end of the day so my char can log off and clear it until the next day.

ThePrinceOfBargains

This is definitely the case with forcing chest locks too. Far too frequently I’m failing 3-4 times in a row to force a simple or intermediate chest at 80-85% chance of success. That’s absurd. It felt like I had better success before I got the last two upgrades.

Mystrius_Archaion

That 90% calculation is server side. Yes, it is a roll of the RNG just for your attempt, but all other samples are still part of the whole effect and contribute to the overall odds. It's like picking your numbers for the lottery and using the same numbers every drawing. Your chances are correct for the one drawing, but if you choose random numbers every drawing or consider multiple drawings as part of your odds calculations then you introduce a whole multiplier of the variables and thus minimize your odds.
This is why RNG tends to just suck. You never really get even the advertised odds because both sides of the equation are technically not fixed as one side should be.

I did a whole calculation on this years ago when playing World of Warcraft on the chances of getting a 1 in 100 chance of a specific drop from a 25 person raid where everyone had to roll for the item.
The chances of any one player essentially guaranteed getting the item was 1 in 100 runs times 25 players equal to 2500 runs. But, if you change out any players or add more groups running it then the odds get much worse.

That's not even considering personal loot chances. Sure, personal loot chances may seem better, but the odds of a 1/100 runs drop are now 1/100 for each player ending up at actually 1/(100*99*98*97*....76) for 25 people which is a very bad chance. This is why RNG is very unforgiving even if a system seems better at first glance and why drop chance needs to be very carefully balanced.

Mystrius_Archaion

TheMaster wrote: »

This is definitely the case with forcing chest locks too. Far too frequently I’m failing 3-4 times in a row to force a simple or intermediate chest at 80-85% chance of success. That’s absurd. It felt like I had better success before I got the last two upgrades.

I wish it would at least push intermediate chests to 100%. The only ones with any failure chance should be advanced and master.

Mystrius_Archaion

Asardes wrote: »

I've noticed it matters a lot how the eye reticle looks. If it's almost closed then the displayed pickpocketting chance is accurate. If it's more open it decreases and you can get caught even with 95%. So try to wear 7 medium and maybe one of the sets that decreases detect radius like Night Mother's Embrace. Also even if the chance is 100% wait until the chime when the number turns green to press the button. I do get caught sometimes and stack bounty but do that part at the end of the day so my char can log off and clear it until the next day.

That eye mechanic is entirely separate. You are revealed every time you pickpocket, so when they eye is partially open you will likely be spotted by somebody because it is warning you that you are close to being detected because you're likely on the edge of detection range or within it if not for being in sneak.

Kaliki

Thanks for the effort and collecting data on this.
I have long had the feeling the chances were worse than displayed.

But I think the chances for forcing locks are even worse than pickpocketing. Its a waste of time to try and force at 75% because youll very likely end up needing 3 trys.

FoolishHuman

Sometimes the NPC turns around just at the moment you want to pickpocket, and since the calculation is server side it might still look like they are looking away on your screen. Also it's possible some other NPC saw you. They also might have come into vision just at the moment you pressed the button.

SpiderCultist

Same goes when upgrading a weapon... just don't do it if you don't have a 100% success.

Teridaxus

I once failed a 100% easy target...

One time i also got reported by the same person which i pickpocket while no other npc around. I got both the item and a small bounty from it.

eso_nya

Yeah last night i got caught at 100% aswell (the one and only time so far).
Seeing the same results for 90% chance.

Bug or feature: Whenever i break into a home, the npcs inside dont unstealth me, but start pointing at the point where i entered and yell nice stuff, that i should go away and not defile their homes, i also dont get a bounty. For a while they keep "acting weird" like turning in random directions and shaking their fists and stuff. (is kinda funny when i instantly go for a BoW-kill and they keep talking while being dead)

My char for crime actions is a bosmer nightblade and im using 5 nightssilence and 3 nightterror. For heists and sacraments i could pee on the npcs legs and they'd say "my boots getting wet, but i cant see anything"...

Another thing i noticed, alot of times after BoWing, the target faces a completely different location, than the one i saw it in when activating. esp noticeable when i get the throwing knife animation. that one shouldnt hit into the chest, right?
That being an indicator for npcs being turned into a different direction than what it looks like clientside, i still get the same bad successrate for 90% pickpocketing vs targets sitting on a chair.

Mojmir

Simple chest, 80% chance to force lock
Failed 3 out of 4 times
Lol

phileunderx2

Yes popping locks on chests doesn't work like the passive says .Most chests regardless of difficulty take multiple attempts to open. That is on a maxed character. Pickpocting is the same.

I_killed_Vivec

Silly, it's 90% for 90% of the people who aren't you!

Though to be serious, we all have a skewed view of RNG because we only see our own results. In the time that it took you to make 300 attempts there could have been over 300,000 attempts game-wide, which makes your personal experience insignificant.

Think of it as an allegory for life. You might be an individual, unique, but you are only one of 7 billion on this planet, circling a star that is one of 200 billion in a galaxy, which is one of at least 100 billion galaxies in the visible universe.

Now that's true insignificance. Yet we are all bothered about perceived unfairness in the RNG in a video game...

eso_nya

I_killed_Vivec wrote: »

Silly, it's 90% for 90% of the people who aren't you!

Though to be serious, we all have a skewed view of RNG because we only see our own results. In the time that it took you to make 300 attempts there could have been over 300,000 attempts game-wide, which makes your personal experience insignificant.

Think of it as an allegory for life. You might be an individual, unique, but you are only one of 7 billion on this planet, circling a star that is one of 200 billion in a galaxy, which is one of at least 100 billion galaxies in the visible universe.

Now that's true insignificance. Yet we are all bothered about perceived unfairness in the RNG in a video game...

While i totally agree on the human insignificance, what u write above is not how rng works.

A sample of 3 x 100 attempts is a very good mesurement for a 90% chance. "Very unlucky" would be ~50 or 60 failures. The more likely assumption would be, that the light fingers passive isnt working as intended.

Mystrius_Archaion

eso_nya wrote: »

show previous quotes

I_killed_Vivec wrote: »

Silly, it's 90% for 90% of the people who aren't you!

Though to be serious, we all have a skewed view of RNG because we only see our own results. In the time that it took you to make 300 attempts there could have been over 300,000 attempts game-wide, which makes your personal experience insignificant.

Think of it as an allegory for life. You might be an individual, unique, but you are only one of 7 billion on this planet, circling a star that is one of 200 billion in a galaxy, which is one of at least 100 billion galaxies in the visible universe.

Now that's true insignificance. Yet we are all bothered about perceived unfairness in the RNG in a video game...

While i totally agree on the human insignificance, what u write above is not how rng works.

A sample of 3 x 100 attempts is a very good mesurement for a 90% chance. "Very unlucky" would be ~50 or 60 failures. The more likely assumption would be, that the light fingers passive isnt working as intended.

No, it is working perfectly as intended for a server side RNG calculation.
Your specific pickpocket attempt is only one of thousands of attempts on the servers. So you may have a 90% chance, but the odds of the server side RNG rolling a number in that 10% range of failure is much higher due to your roll actually being 1/1000 or worse. That makes your 10% range actually 100 out of 1000 RNG rolls. If that goes up to 10000 attempts then 10% is 1000 failure rolls. It's not exactly that way, but it essentially is pretty horrible.
You're not the only one picking random numbers from the same RNG. The good numbers may have already been picked by other players.

Elsonso

FoolishHuman wrote: »

Sometimes the NPC turns around just at the moment you want to pickpocket, and since the calculation is server side it might still look like they are looking away on your screen. Also it's possible some other NPC saw you. They also might have come into vision just at the moment you pressed the button.

I have noticed that the percentage can be very dynamic, often changing unpredictably and quickly. If there is any sort of latency in the notification of that change, the pick pocket attempt could be performed while "90%" is still on the screen, but the actual percentage is in the process of changing to something considerably lower.

For this reason, I usually attempt immediately upon seeing it increase to a high percentage. If there is a latency, the window of opportunity is smallest immediately after it has been raised.

If there is too much variation in the number, and it is changing too frequently, I don't attempt it. Unless I have to for a quest, of course. This is to prevent the number from changing on me while attempting it.

For this reason, I am almost never observed pick pocketing, or assassnating, the various inhabitants of Tamriel. I have even been known to successfully (get item, not observed) pickpocket the justice system guards.

Mystrius_Archaion wrote: »

t's like picking your numbers for the lottery and using the same numbers every drawing. Your chances are correct for the one drawing, but if you choose random numbers every drawing or consider multiple drawings as part of your odds calculations then you introduce a whole multiplier of the variables and thus minimize your odds.

Huh?

eso_nya

Mystrius_Archaion wrote: »

show previous quotes

eso_nya wrote: »

I_killed_Vivec wrote: »

Silly, it's 90% for 90% of the people who aren't you!

Though to be serious, we all have a skewed view of RNG because we only see our own results. In the time that it took you to make 300 attempts there could have been over 300,000 attempts game-wide, which makes your personal experience insignificant.

Think of it as an allegory for life. You might be an individual, unique, but you are only one of 7 billion on this planet, circling a star that is one of 200 billion in a galaxy, which is one of at least 100 billion galaxies in the visible universe.

Now that's true insignificance. Yet we are all bothered about perceived unfairness in the RNG in a video game...

While i totally agree on the human insignificance, what u write above is not how rng works.

A sample of 3 x 100 attempts is a very good mesurement for a 90% chance. "Very unlucky" would be ~50 or 60 failures. The more likely assumption would be, that the light fingers passive isnt working as intended.

No, it is working perfectly as intended for a server side RNG calculation.
Your specific pickpocket attempt is only one of thousands of attempts on the servers. So you may have a 90% chance, but the odds of the server side RNG rolling a number in that 10% range of failure is much higher due to your roll actually being 1/1000 or worse. That makes your 10% range actually 100 out of 1000 RNG rolls. If that goes up to 10000 attempts then 10% is 1000 failure rolls. It's not exactly that way, but it essentially is pretty horrible.
You're not the only one picking random numbers from the same RNG. The good numbers may have already been picked by other players.

u r kidding me, right?
Edit for clarification:
U trigger an event: <generate random number> this events sends a number back, usually between 0 and 99. if your number is between 0 and 9, your attempt fails, if its between 10 and 99 u r successful. There is not a finite pool of lucky numbers, which is empty when u start to late.
If the numberrange would increase by participating players, it would be either impoosible to fail or impossible to win after a very short time.

Integral1900

I would love to know why the force master lock works nine times out of ten, yet if the same charecter tries it on a simple lock it fails more than half the time... it just doesn’t make sense...

UnseenCat

The percentage can definitely be misleading -- it's meaningful, but only in the instant it first appears. Real conditions can and do change faster than the display. Probably because of round-trip times between server and client, plus what's going on around you and the mark.

As I built up my character on the Thieves' Guild line, I felt that the percentage became more reliable as I upped all my skills relating to stealth and legerdemain. It's not just a simple progression focused on pickpocketing.

Improving stealth radius is a big part of making the chances more reliable. A lot of pickpocketing attempts fail when an NPC turns around or walks up from behind an obstacle. Situational awareness is key, especially at a lower detection radius when you're starting out.

Racial passive bonuses can help; building up my Kajiit racial line certainly hasn't hurt.

The class assigned to the NPC mark you're targeting matters, too. Nobles pay attention to their purses! They'll react faster and may catch you regardless of the percentage, particularly if you haven't leveled up your stealth much yet. Beggars tend to be the least aware and are easier to start with. (And the RNG doesn't seem to discriminate as to the quality of loot you can lift based on the NPC's class, so they still can yield good hauls.) Merchants, craftsman classes and bards are somewhere in the middle. Guards are hard, and are of course the riskiest if you do get caught.

Distraction seems to play a part. If the NPC is paying attention to a bard, a performance, or a speech, they seem to be a little easier. I've had good luck with Bards busy performing -- they seem to have a slightly better than 50/50 chance of not noticing you even at lower percentages of success. At least that's the "feel" I've gotten from the game. (On the other hand, there's usually an audience of two or more NPCs watching the bard, and they'll notice you and kill your chances unless you can stay out of their sight lines.)

Another tip -- 90% (or any percent) in white isn't the same as the percentage in green. It pays to wait for the chime, particularly while you're still at lower levels of Thieves' Guild skill and stealth. When the indicator goes green, not only is the mark not likely to notice, but NPCs are probably not aware at the instant the chime sounds. Both factors decay very rapidly, though. Percentage in white is purely the base awareness level of the mark, NPCs may or may not be factored in.

Watch your stealth indicator "eye" icon, watch the class of the NPC you're targeting, be aware of NPC movement patterns as well as sight lines to the mark and surrounding NPCs.

Lag, of course, will ruin your day.

After building up my character's stealth and Thieves' Guild passives, pickpocketing guards at least once became possible. Haven't emptied one yet -- I still have more skill development to go, though. But their success chances drop far faster with each successful lift than any other NPC type. Patience...

As to those frustrating NPCs that still manage to barge in and call you out on a perfect 95%-100%? Khajiit has claws -- and Blade of Woe -- to deal with them.

devilfunk

100% I don't care.