Pickpocket is Absurdly Broken and Needs Complete Overhaul

[DagothRocketeer]

I've been messing around with the GECK, and testing the variables that affect pickpocket success. What I've found shocks me.

 

WARNING: ARITHMETIC INCOMING

 

From what I can tell, there are four pairs of global settings that affect Pickpocket success:

 

First:

fPickPocketActorSkillBase: This is the raw score for pickpocket success before factoring in the player's Sneak. By default, it is set to 40.

fPickPocketActorSkillMult: This is the factor by which each point of Sneak skill increases the raw success score. By default, this is set to 0.6.

 

For example, a player with 20 points in Sneak will have a raw score of pickpocket success of 40 + (20 * 6 / 10) = 52. Meanwhile, a player with 50 Sneak will have 70, and a player with the maximum 100 Sneak will likewise have a raw score of 100. Don't worry if this seems high; this isn't the actual success rate, but a value that will be opposed and reduced by other factors.

 

Second:

fPickPocketAmountBase: This would be a raw baseline score for opposing the check calculated above. However, this is set to 0 by default.

fPickPocketAmountMult: This is the factor by which each cap-worth of an item's value this raw opposing score increases. By default, it is -0.5.

 

For example, an item worth 10 caps would have a rating of 0 - (10 * 5 / 10 ) = -5. An item worth 50 caps would be rated -25, and an item worth 100 caps would be rated -50. With me so far?

 

Third:

fPickPocketMaxChance: A bit easier to grock; this is the maximum percent chance of success a player has to Pickpocket. It's set to 85.

fPickPocketMinChance: By contrast, this is the minimum percent success chance. It is set to 5 by default.

 

This means that no matter how 'easy' an item is to steal or how high your sneak is, you will fail 15 times out of 100. However, even a character with the lowest possible Sneak attempting to pickpocket the 'hardest' possible item will still succeed 5 times out of 100. Easy.

 

Fourth:

fPickPocketTargetSkillBase: This would serve as another baseline opposed score, but it is set to 0 by default.

fPickPocketTargetSkillBase: Contrary to my 1st assertion, this is a factor that increases with the target NPC's own Sneak skill. It is set to -0.6.

 

Therefore, an NPC with an Agility of 5 and a Luck of 5 for whom Sneak is not a tagged skill (which is a very common state of affairs among NPC's) would possess a Sneak of 15 and therefore generate a score of 0 + (15 * 6 / 10) = -9. An NPC with those stats who had tagged Sneak will possess 30 Sneak and generate -18, and a ninja-wannabe with 100 Sneak will generate a score of a whopping -60.

 

EDIT: I initially conjectured that this was based on Perception and had a much smaller effect. Quick testing after the fact proved that I was wrong. The revised data does drastically little to improve my opinion of Pickpocketing's usefulness.

 

So let's add it all up for a practical example. My character, Doctor Norman, fancies the Egg Timer in Cliff Briscoe's pocket. Doc has 40 Sneak, Cliff has 17 Sneak, and the Egg Timer is worth only 1 cap. So when Norman reaches into Cliff's pocket to take it, the game does the following calculation:

 

(40 + (40 * 6 / 10)) + (0 - (1 * 5 / 10)) + (0 - (17 * 6 / 10)) = 64 - .5 - 10.2 = 53.3

 

Therefore, Doc has a 53.3% chance to successfully steal Cliff's Egg Timer.

 

Let's say, for the sake of this hypothetical, that Doc fails that Pickpocket attempt, ruining their beautiful friendship forever. Harsh words are spoken in anger that can never be withdrawn. Doc is now in need of a weapon to kill that f***ing rat for good, so he can take the Egg Timer from his cold, dead hands. Luckily, Doc knows that Jeannie May Crawford just across the way has just gotten a brand-new, perfect-condition 9mm Pistol, the perfect tool to knock off cliff and pin the crime on her! Not wanting to piss off an armed woman, Norman wisely uses a Stealth Boy before the attempt, which temporarily increases his Sneak to a perfect 100.

 

While thusly enhanced, he attempts to steal the pistol, which has a value of 100 caps, from Jeannie May, who has a Sneak of 18. The good Doctor reaches out to take the gun, and the game does the following calculation:

 

(40 + (100 * 6 / 10)) + (0 - (100 * 5 / 10)) + (0 - (18 * 6 / 10)) = 100 - 50 - 10.8 = 39.2

 

A 39.2% chance? What's this? Due to the item's value, our chances are even worse than stealing the Egg Timer, even with 100 Sneak.

 

So let's say, for the sake of argument, that this attempt also fails. Now the town of Novac is up in arms, chasing our hapless Doctor Norman down the road. Vowing to show them all one day, show them once and for all, Norman spies a prospector carrying a broken suit of T-45d Power Armor on his back. Thinking it will be better protection from the angry mob than nothing (it wont), Norman makes one last stab at larceny and attempts to pluck it away before the angry citizenry catches up. Having read an issue of iLa Fantoma! to amuse himself as the mob formed outside his motel room, his sneak is temporarily bolstered to 60. The power armor, completely broken, is worth absolutely nothing, and the prospector's Sneak is 1 thanks to a monastic vow of perpetual loudness he took in New Mexico (don't ask). Norm reaches out his hands, and the following check is performed:

 

(40 + (60 * 6 /10) + (0 - (0 * 5 / 10)) + (0 - (1 * 6 / 10)) = 76 - 0 - .6 = 75.4.

 

The check succeeds for once, and good ole Doc, having failed to lift both a 1-pound Egg Timer and a 1.5 pound pistol from senior citizens, easily relieves the cacaphonous looter of forty-five pounds of riveted steel. Unfortunately, broken armor can't be equipped, and Norman is beaten to death on the pavement. Time your eggs in Hell, Cliff.

 

So, do you understand how Pickpocketing works in New Vegas? Does anyone else see the problems here? Item value is so heavily weighted that even a master Sneak-er can be completely stymied by even common items. In fact, any item with a value of 190 caps or greater will reduce the chance of Pickpocket success to the minimum 5%. Meanwhile, a master thief stealing a valueless item from a blind, sleeping idiot will still fail and get caught %15 percent of the time. The utility of Pickpocket is so harshly limited by the strictness of the opposing value-based score and the rather heavy arbitrary cap on success chance that it may as well not be affected by Sneak whatsoever; anything a player would conceivably want to steal from an NPC, such as a rare weapon, puts a master on the same playing field as a total novice. Meanwhile, the chance of getting caught while stealing petty, low-value items like a handful of caps or some ammo is still so steep regardless of Sneak that it's hardly worth risking getting caught around anyone you want to maintain good relationships. And if you don't care about your reputation, or you have a good Sneak, you can silently kill whomever you wish to steal from so much more easily than you can pickpocket from them.

 

On the Vault wiki, they seem to think weight has something to do with Pickpocket success. They are mistaken. Weight has no effect whatsoever. I know this for a fact, because I set up a test NPC that carried various weights and values of items, up to 10,000 pounds and no value, or no weight and a value of 10,000 caps. For these tests, I changed the min and max pickpocket chance to 0 and 100 and decreased fPickPocketTargetSkillMult to 0, meaning that only the player's Sneak and the item being stolen factored into the check. Under these conditions, you can successfully relieve someone of 10,000 pounds every single time, provided that it has no value. Meanwhile, an item worth 100 measly caps can be stolen half of the time and an item valued at 200 caps can never be stolen, ever. Even if weight were taken as a factor to be considered with value, an item with a weight of 10,000 lbs. and a value of 1 cap should definitely be un-stealable. But I was able to steal such an item every single time for several dozens of attempts, which seems to confirm its theoretical success probability of 99.5%.

 

Furthermore, I have no reason to believe weight has any factor on Pickpocketing even without having to test it, simply because there is no global setting regarding weight as part of a pickpocket attempt. By altering fPickPocketAmountMult to very high or low factors, I was able to easily observe the effects this bore on success; set to very high amounts along with the above conditions, even low-value items became difficult or impossible to steal. Set to very low amounts, the opposite was true, and set to 0, even an item worth 10,000 caps was stolen every time.

 

So that's the long and short of Pickpocket. It's worth pointing out that the workings of the system are completely hidden from the player, other than knowing it is (marginally) affected by Sneak. This makes it all the worse that the system works completely contrary to very reasonable expectations; what I would consider to be the most natural and intuitive targets for would be felons- small, light items of high value, just as they are in real life- are all but futile targets, or at least completely invalidate even the maximum possible skill at picking pockets. Meanwhile, exactly what no one should expect to work- stealing bulky, heavy, highly visible goods- is treated with utter indifference by a system presumably themed around perception, detection, and sleight-of-hand. Just as demonstrated above, a master of Sneak and a complete novice have exactly the same odds of stealing a 1.5-pound .45 Auto pistol due to its high value of 1500 caps. The master is distinguished from the novice only in their ability to steal comparatively worthless things... like a broken suit of power armor, which is as large as a man, weighs 45 pounds, and has no intrinsic worth.

 

The system is clearly broken. Yet, I can't conceive of a way to fix it. If there were an obvious global setting related to weight's effect on pickpocketing success, the obvious thing to do would be to increase that factor to a workable amount and either drastically reduce or eliminate outright the effect of value on the equation; the rationalization, I suppose, is that a would-be mark would be more aware of his more valuable belongings in the event of an attempted theft, but this only makes to a point, not to the drastic degree emphasized in the mechanics. But there seems to be no such weight-related setting. A Skyrim modder confirmed that in addition to the pickpocket-related settings found in the GECK, there is a global setting in that Skyrim's Creation Kit called "fPickPocketWeightMult" that serves this function, but this setting is nowhere to be found in the GECK. I'm tempted to say something could be worked out with scripting, but I admit I believe that only due to my total ignorance of scripting, and by extension, the fact that such a task must fall to someone else if it were possible.

 

I initially (and foolishly) assumed that Perception was the NPC stat opposed to the player character's Sneak check, but this isn't so. Frankly, this seems wrong to me. It should be no mean feat to take even an item of small notice from the pocket of an NPC with 10 Perception; a score of 10 in any SPECIAL stat marks someone as truly among the greatest. Yet as it stands, Perception has no bearing on success or failure. That's bizarre. Furthermore, it doesn't really make sense for the NPC to oppose the PC's Sneak check with one of their own; on one hand, it makes some sense that an experienced sneak-thief would be more aware of others' attempts to beat them at their own game. But the way it works in the game is preposterous; NPC's with a Sneak higher than 20 or 30 are very rare indeed, yet Pickpocketing is already so absurdly imbalanced that even that little bit of proficiency is enough to drive chances of successfully taking anything unacceptably down. Given that it seems so much more intuitive for Pickpocketing to be opposed by the target's Perception (which the folks at the Vault wiki incorrectly seem to believe), I'd really wonder if it could somehow be changed to reflect that instead, especially given that the SPECIAL attributes of a given NPC are so much more more variable than their skills, which tend to be either tagged and very high or (most often) untagged and very, very low. In an otherwised-balanced system, using PE to oppose the PC's Sneak check would be ideal, in my opinion. It could easily output the same range of values as Sneak, though not with the same granularity; simply decrease fPickPocketTargetSkillMult to -6 rather than -0.6, and you have a range from -6 to -60, as opposed to Sneak's potential -.6 to -60.

 

Yet as it stands, it might only be possible to attempt to rebalance the effect of value and forget about weight or skill/attribute. But laying aside the fact that it makes no damn sense, the value of items varies so much more wildly than item weight that the sweet spot of probability- values of fPickPocketAmountMult, fPickPocketActorSkillBase, and fPickPocketActorSkillMult for which a novice and a master observe a demonstrable and reasonable disparity of success and failure- is entirely too small. If the values are geared so that the most common sorts of items like food, ammunition, stims, and caps are stolen with a reasonable rate of sucess for a reasonably skilled thief (say, a PC with a Sneak of 60), then things the player would actually want to steal, such as even the most common of weapons and armor, remain balanced so that their value totally invalidates the skill of even a master. But balancing them around those things seems impossible; the range of potential item values is too vast to center upon without making weapons worth a few hundred caps as easy to steal as an Egg Timer, while the items at the top of the range- such as weapons or armor worn by NPC's the player doesn't want to kill or piss off, which are some of the most commonly-targeted items for pickpockets in the first place- still totally 'immune' to the Sneak skill, so to speak. And of course, balancing the values around those rarest of items makes everything else trivial for even a beginner.

 

Of course, there may well be a sweet spot for those values that I'm just not able to compute. The spirit is willing, but the calculus is weak.

 

Even if all of that could be sorted out, though, I can't help but take offense at the arbitrarily steep minimum and maximum pickpocket success chances. On one hand, the system is so broken that the minimum success chance is an absolute requirement to steal anything of any appeal whatsoever, along with savescumming like a bastard. But that makes it burn all the more that even a character with no skill in Sneak whatsoever ends up with the same odds as a master for stealing anything of interest. In a balanced, reasonable system, I'd eliminate the minimum success rate altogether, or make it a function of Sneak itself, if that were somehow possible; an inept butterfingers has no business getting lucky in all but the rarest circumstances, but a practiced thief should be able to maintain some tolerable baseline of success in even the toughest cases. The results of hard work often looks like luck to saps, as a wise game developer once said.

 

Similarly, I'd do away with the cap on pickpocket success chance whether the system were balanced reasonably or not. A much smaller lien on the maximum chance might be reasonable, but a flat 15% screw-you failure rate? The idea that the finest criminal in the wastleand or someone who is literally invisible could successfully make off with the gum in a sleeping boozehound's back pocket less often than a given brand of toothpaste can garner the recommendation of a given dentist is a contemptible notion. At some point, you just have to wonder if they were just trying to marginalize the pickpocket mechanic with the gravest prejudice possible. A chance of 5% I could understand, IF the rest of the system were balanced. It's good enough for tabletop gaming. I wouldn't think to question it, at least. But on top of this systems multitudinous failings, it just burns my ass. It should be pointed out that even for an item with no value, a success rate of 100% would still be impossible since it's impossible for a target of pickpocketing to have 0 Sneak, giving absolute maximums of 97% through 60% for targets with 5 through 100 Sneak (5 Sneak being the theoretical minimum possible skill, from an NPC with 1 Agility, 1 Luck, and Sneak not Tagged).

 

Comments? Observations? Ideas? Egg Timers?

Edited November 27, 2014 by DagothRocketeer

[JohnPascal]

While I'm even less capable of fixing this problem than you, I have to commend how thorough your testing was. I haven't tried to reproduce your results but someone should definitely get on that so we can confirm your observations.

 

As a side note that hypothetical about the egg timer was pretty funny.

Edited November 27, 2014 by JohnPascal

[sullyvanj93]

Your post was informative, entertaining, and spot on with my aggravation of this. While I don't possess the skill to help resolve the issue, I join in your frustration and hope that something can be done about it.

 

On another unrealistic note, there are many items worth 0 caps that seem to be remotely impossible to not successfully steal from NPCs. There has been a huge spree of larceny in my mojave, mostly concerning the following items: bullet casings, drained energy cells, and (my personal favorite) the poor NCR soldiers dogtags, stolen from RIGHT AROUND THEIR NECKS!

[DagothRocketeer]

I've had it suggested that this could be fixed using a perk that altered these values, but I'm not sure how to go about doing it.

 

Theoretically, if the fPickPocketAmountMult and fPickPocketTargetSkillMult settings were set to 0, item value and opposing Sneak would becomes irrelevant. From there, the perk (which would just be added automatically by a mod, or through the console) would check the weight of an item and alter fPickPocketTargetSkillBase to lower and lower amounts (since this is an opposing value that lowers the chance of success) with higher and higher item weights. Meanwhile, a similar component of the perk (or even a different perk, for easier setup/debugging purposes; these aren't to be taken on level-up, but added manually, so it doesn't matter) could check the Perception of the target and lower fPickPocketActorSkillBase (since the only other values are apparently hard-coded to values we want to ignore), which would work fine; if the target's Perception directly opposes Sneak, we could leave the theoretical ideal fPickPocketActorSkillBase at 40, while it would always be reduced by a factor of four times the target Perception.

 

Actually, I think per-pound item weight could easily be considered directly proportional to percentage-chance of success, ie, one pound of the stolen item reduces the chance of success by one percent, and so on.

 

Take an example. To make it easy, let's just assume I have 100 Sneak, because Doctor Norman is now a larcenous ghost. If each point of Sneak is still multiplied by .6 and our base chance is ( 40 - ( Target's PE * 4 ), and if item weight is -1 percent per pound (which would not be a calculation by the engine, but manually set by tables programmed into the theoretical perk), then if Doc rises from the grave to steal a 1 pound Egg Timer from that rat bastard Cliff Briscoe with his so-so 5 Perception, it would be:

 

( 40 - ( 5 * 4 )) + ( 100 ( 10 / 6 )) - 1 = 20 + 60 - 1 = 79

 

A 79 percent chance of success which... actually still seems really low. Honestly, stealing anything a pound or under from someone with average PE should be all but granted for a character with 100 Sneak. Actually, we could set fPickPocketActorSkillMult higher, like to 1, while altering the effect of Perception to something like -8, which would make that equation

 

( 40 - ( 5 * 8 )) + 100 - 1 = 0 + 100 - 1 = 99

 

99%! Now, that's more like it. That might even seem unreasonably high, but keep in mind that's for 100 Sneak. Look how that would work with a more perceptive target and a heavier item. And a more reasonable Sneak skill.

 

Corporal Sterling. Undoubtedly one of the most popular targets for pickpocketing in the Mojave Wasteland. He's badass, he's NCR, and he's surrounded by the might of Camp McCarran. And he carries La Longue Carabine, a unique cowboy repeater. He's not a character that most players will just want to walk up and murder, but at the same time, the attraction of unique, shiny items can't be understated.

 

Sterling has 8 Perception, a score befitting a First Recon sharpshooter, and there aren't many NPC's with a score like that. La Longue Carabine weighs a modest 5 pounds. Norman, meanwhile, has been returned to life by mysterious means, has learned nothing from his former life (and death), and has a Sneak of 60: not too high, but good enough to try something so dastardly. Mix in a bundt pan and you get:

 

( 40 - ( 8 * 8 )) + 60 - 5 = -24 + 60 - 5 = 31

 

A 31% chance of scoring a repeater like no other and escaping scot-free. Daunting, to be sure, but an issue of iLa Fantoma! bumps that to 41% or 51% depending on whether or not the character has the Comprehension perk, and a Stealth Boy jacks it up to 71%. That's no free lunch, but it's better than the measly five percent bare-minimum that a player of ANY skill would get under the game's base rules, and I think that's perfectly appropriate for trying to steal a First Recon sniper's cherished rifle right off of his back.

 

The allure of using weight as a determinant of pickpocket success is two-fold; it makes logical sense, and it doesn't create an extreme delta of values that invalidate the player's skills at all but one narrow point in that range. Very low chances of success feel fair if the numbers are right. For instance, stealing a 20-pound missile launcher from a 6-PE no-name Boomer Guard would have a maximum success chance of 72%, which seems reasonable to me, while stealing Cliff's Egg Timer (a highly sought-after item, apparently) would be all but guaranteed at max Sneak and vary almost directly-proportionally with skill.

 

A character with 100 Sneak trying to steal the heaviest item in the game (albeit tied), a suit of T-45d Power Armor from a theoretical character with 10 Perception would have a 15% chance of success. Which... seems totally reasonable, actually. That should be damn near impossible, and it is. And having each point of Sneak equal an extra point of Pickpocket success percentage makes each point in that skill feel meaningful. (Which is not to say that Sneak isn't already awesome. It is, trust me, I know.)

 

So I really, really like the math on this. If I could implement it, I would do it in a heartbeat. The only problem is... I have no idea if you can alter those values via the perk interface. Or if you can alter those value dynamically at all. Only two settings need actually be altered dynamically, while fPickPocketActorSkillMult is set to 1 and fPickPocketAmountMult and fPickPocketTargetSkillMult are set to 0 and left there. The idea to do it with a perk was given to me by a Skyrim modder who did a major overhaul of that game's stealth systems, but they seemed totally ignorant of Fallout and the GECK, and I'm no veteran modder.

 

Has anyone messed with custom perks enough to know if this is possible, or am I barking up the wrong tree? Is there any other resource for changing these numbers on the fly?

 

Ideas? Insights? Egg Timers?

Edited December 7, 2014 by DagothRocketeer

[JohnPascal]

This seems like a great solution to sneak in New Vegas. I used sneak quite a bit in Skyrim (great for getting rich, leveling up quickly, and giving my archer easier kills) but I never really used sneak in Fallout because it seemed much harder, probably because of the explanation you gave. If someone managed to implement a sneak system similar to this I'd love to give it a spin.

 

On another note, you should keep writing about these scenarios where characters do normal things (or, in the case of Doctor Norman, try to do things and die horribly) in the game. They are quite entertaining and really help support the points you're making.

[CardTaken]

I was thinking about Pickpocket and wondering if anyone had made improvements to it (so far, my searches have been fruitless, but I found this topic, which was interesting).

 

My thought is, rather than trying to fix the broken values, if there is some way to trigger a mini-game, in much the same way that lockpicking or hacking does, I suspect that would make it far more engaging.

 

I've no idea if such a thing is plausible given the game setup, but here are some design thoughts on the matter:

 

What I'm imaging first of all is, if possible, is keep Sneak affecting Pickpocket success chance in some capacity, but get rid of the whole "you need to be Hidden to have a chance at successful Pickpocket." It's so stupidly arbitary to begin with, trying to find an angle where nobody sees you so the monitor will change to "Hidden" and you can attempt to Pickpocket without automatically being detected.

 

It's not remotely realistic anyway, unless you're Pickpocketing someone who is sleeping. Like, "Yeah, I'm Sneaking around your office in plain sight. Oh hey, I got the right angle so you can't see me for a moment. Now I have a chance to take your stuff" ???

 

Second: Mini-Game - I'm trying to think of something that would be both in the spirit of Pickpocketing and relatively easy to program with pre-existing assets. Much like Lockpicking and Science, item value might lock you out of trying and tell you that you need higher Sneak. So for example, if you're trying to steal a 2k cap item, it might require something like 75 Sneak to make the attempt and it'll tell you that when you click on the item in the person's inventory.

 

One way might be a mini-game that functions like the Shell Game (see: https://en.wikipedia.org/wiki/Shell_game). You have three objects (something NV-themed that are like shells). Under one of them is an icon, which represents the item you're trying to Pickpocket. The game is not rigged at all, unlike some forms of the Shell Game. Instead, it's simply a reflex game, keeping your eyes on shells as they move and then picking the correct one (if you've ever been to a baseball game, it'd be like the animated shell games that they do there). You'd get something like 3 attempts to do this before you get caught and much like hacking, you can back out before you use the last attempt, with no penalty.

 

Item Sneak level requirement (25, 50, 75, etc.) would determine the speed at which the shells move.

 

That's just one idea. Again, I don't know if it's plausible. And mind you, this is nothing along the lines of a mod request. I simply figured I'd share my design thoughts on the matter, since I'm in the dark at the moment on NV's modding limitations. And cause honestly, the concept is burning a hole in my brain in much the way that money burn's a hole in one's pocket.