STEALING EXPERIMENT: suggestions?

[The Outcast Sarchus]

Dear friends, enemies and strangers,

While it seems that we all understand the variables that effect stealing, such as armor worn, etc, there is little in the way of concrete evidence to detail how, exactly, they effect it. I myself have a number of questions. Namely:

1) When stealing from a pile, does the armor I'm wearing merely effect my rate of total failure, or does it also effect the degree of my success? This leads to another question:
2) Is my success measured in percent of the total pile? For example, is a full success say 50% of the pile, so the same success would steal 500 gold from a 1k pile, but a mere 50 from 100gp? Since we know it is possible to steal %100 of something if the amount is small, could this percentage operate on a curve, or could there be a max-stolen-per-steal cap?

In order to begin to dissect some of these equations I'm going to run a test in-game, the results of which I'll post here. I am at work now, so seeing that I have a few hours before I get to work on this, I figured I might present the study to see if anyone had any suggestions. I, being scientifically minded, but not scientifically skilled, will probably make some sloppy mistakes without a more studious eye to guide me.

The test will run like this:

Without any armor on, I will steal from an alternate character who is holding 1,000 gp. I will repeat this 50 times, each time replenishing the pile of gold to 1,000 pieces before attempting to steal. I will record my results in terms of minimum gold grabbed, maximum, and average.

I will repeat the above experiment wearing a dexer suit, and then again in full plate, recording min, max and average accordingly.

Then, I will repeat the entire process (50 steals armorless, 50 in a dexer suit, 50 in plate) against a pile of 500 gp, to see if the success is at all related to the size of the target pile.

Besides increasing the sample size, is there any obvious way to improve this model that I'm forgetting? Additionally, are any of the answers to these questions already known?

[MatronDeWinter]

That's not exactly how stealing from a stack/pile works.

The server determines the max amount that you could possible steal (based on total weight). It may be all of the pile, or a small proportion of it. It then picks a random amount in that range and attempts the steal. The weight of that amount is calculated v. your stealing skill, armor, and nearby mobiles to determine your success chance, and degree of success.

The data from this "random" selection (that determines which amount of the total to steal) is in a bell-curve, so you are rarely stealing 1gp from a pile of 10k, or stealing 10k from a pile of 10k.

How armor plays in this would be interesting to read about. Particularly if you were to run an experiment (with a decent sample size, say 1000 steals per) on items of different weight at different armor levels. I would not even bother worrying about how stacks/piles, or individual items of weight are dealt with, it's very simple to calculate.

[Mikel123]

The data from this "random" selection (that determines which amount of the total to steal) is in a bell-curve, so you are rarely stealing 1gp from a pile of 10k, or stealing 10k from a pile of 10k.

Really? Have you tested that? That seems really bizarre. I don't even know how you'd program that. I'd have bet money it was a simple random number between 1 and 10k.

[nightshark]

The data from this "random" selection (that determines which amount of the total to steal) is in a bell-curve, so you are rarely stealing 1gp from a pile of 10k, or stealing 10k from a pile of 10k.

Really? Have you tested that? That seems really bizarre. I don't even know how you'd program that. I'd have bet money it was a simple random number between 1 and 10k.

Weird, the one time I did play my thief I stole 1k bandages from someone macroing. I never stole more than about 50 bandages. I assumed you just stole somewhere between 1-10 stones worth of whatever you were stealing.

[Populus]

The data from this "random" selection (that determines which amount of the total to steal) is in a bell-curve, so you are rarely stealing 1gp from a pile of 10k, or stealing 10k from a pile of 10k.

Really? Have you tested that? That seems really bizarre. I don't even know how you'd program that. I'd have bet money it was a simple random number between 1 and 10k.

Weird, the one time I did play my thief I stole 1k bandages from someone macroing. I never stole more than about 50 bandages. I assumed you just stole somewhere between 1-10 stones worth of whatever you were stealing.

I'm pretty sure it's based on stones. But I think it's different on UOR/AOS tho.

[MatronDeWinter]

From 1000 bandages, the max value possible is 100 (10 stones). Insert the bell curve formula (that I suspect) and you end up always stealing around 50 bandages.



That's the problem with people stealing stacks of items. If you are around 50 skill it will gain just fine statistically. Which is why most people are like "yea that works to gain" when they start their newbie (50 skill) thief. The efficiency of gains are disproportional to your skill at this rate. When you are like 80 skill for example, there is a statistically lower chance that you will steal something that is 7-9 (~8) stones required to gain. So while you may very well be able to gm doing this, it will take you much longer than if you just set the proper weight non-stack item and macro on that.

Adding or decreasing armor will "shift" the curve down, and not change any real data points.

Since we know it is possible to steal %100 of something if the amount is small, could this percentage operate on a curve, or could there be a max-stolen-per-steal cap?

Well, if the stack is under 10 stones, there is always a chance you can steal the entire thing. The max-stolen-per-steal is the total of items in that stack equating 10 stones.

[Mikel123]

Yeah, I'm telling you - I would bet a lot of gp that it's not a bell curve, but rather a straight line, such that each number of bandages, from 1 to 100 (or perhaps, 10 to 100, since maybe it will always steal at least 1 full stone worth?) has an equal probability. I don't know much about programming, but my guess is that it's much simpler to say, "Steal between 1 and X, where X is the greater of 10 stones worth of bandages or someone's entire stack" than it is to force it into a bell curve. Weapon damage follows a bell curve because it's divided into dice already. You can't easily divide a randomly-sized stack into dice, and have one formula that applies to all scenarios, since 5 stones of one item is a different number of them than 5 stones of another item.

[MatronDeWinter]

Yeah, I'm telling you - I would bet a lot of gp that it's not a bell curve, but rather a straight line, such that each number of bandages, from 1 to 100 (or perhaps, 10 to 100, since maybe it will always steal at least 1 full stone worth?) has an equal probability. I don't know much about programming, but my guess is that it's much simpler to say, "Steal between 1 and X, where X is the greater of 10 stones worth of bandages or someone's entire stack" than it is to force it into a bell curve. Weapon damage follows a bell curve because it's divided into dice already. You can't easily divide a randomly-sized stack into dice, and have one formula that applies to all scenarios, since 5 stones of one item is a different number of them than 5 stones of another item.

I am very tempted to agree with you, it would be very simple to just pick a random number between. But if you steal from a large stack over and over, you get the median value the majority of times. Unless my sample size is way off-base, I think it is curved or at least in an interval set like..

If Randomnumber() > 1 && Randomnumber() < 3 { Steal 30/2 bandages}
If Randomnumber() > 3 && Randomnumber() < 5 { Steal 50/1.5 bandages}
If Randomnumber() > 5 && Randomnumber() < 6 { Steal 50 bandages}
and so on forming a step-curve.