Can someone explain lockpick retention?

I searched for lockpick retention but didn't see the answer I was looking for, and with this weekend coming up....

Well I'm currently at r5 lucky (Golden) and r6 Treasure hunter. I'm running NM picks in Fronis Irontoes lair, which I have been for some time. I'm just shy of 6k chests opened. My supposed retain rate is 68%. I know that this is per individual lockpick, not overall retention rate. However, my actual overall retain rate is less than 30% (I don't keep perfect logs, but I know how many picks I've bought and how many chests I've opened) This seems odd to me.

Is it possible that running the same chest over and over is causing my retention rate to tank? I've noticed that without fail, the more I run this chest the worse my rate gets until I'm breaking picks 8 out of 10 times (and I have to imaging the odds of that are astronomical, assuming my 68% retention rate) This happens every day I run this chest, it wasn't just some fluke thing. Now it seems after I've opened the chest 20 or so times, even if I take a break and do something else and come back, I'll break just about every pick. Any insight? Is there something I can do to 'reset' so I actually approach the retention rate I should have? (and I'm not looking for overall 60% retention, but over 50% would be nice)

hi, i sorry i cant help with any info in reset lock retention, but imo as u said you've been opening same chests, with same keys, they way i see that drop on lock picks, seems to be same idea with mobs, if ur questing or grinding mobs or chests for a good drop, it might be better to grind or quest chests / mobs in another area, and maybe the key % drop mighten be as bad............. as bad sorry if that doesnt help much still new myself, just going on what ive played, as to what i know = / hope u get your answer

Yeah, people don't like how the retention rate actually plays out. Lemee trey to dig a thread up for you.
_____
Ok this isn't just about the lockpick retention rate, but you may find some info from other posts since it has to do with opening chests.
http://guildwarsguru.com/forum/showt...ck+ retention

There's no evidence that anything can affect the advertised retention rate.
And since we still haven't seen proof that retention rates are wrong, all I can suggest is that you are unlucky.

When you use a lockpick, the game rolls a random number between 1 and 100. If you have a retention rate of 68%, the random roll has to be between 1 and 68 for you to keep the pick. If the random roll is 69 to 100, you lose the pick.

If you retain the pick, the next time you use the pick you get a brand new random roll. For example, if you have 10 picks and a retention rate of 68%, statistically, the chances of getting random rolls 1, 1, 1, 1, 1, 1, 1, 1, 1, 1 is exactly the same as getting the sequence 99, 99, 99, 99, 99, 99, 99, 99, 99, 99, and is exactly the same as getting the sequence 1, 88, 42, 17, 21, 51, 68, 70, 34, 91.

In the first sequence you'd retain all ten picks (seen as lucky), in the second sequence you'd lose all ten picks (seen as terribly unlucky), and in the third sequence you'd retain seven.

yeah i hear your complaint...i had the same thoughts when i was chestrunning.

ive run 8500+ chests now, and ive had several theories to why they retain or break (ive had 15+ retain streaks, and 10+ break streaks)

one theory was, when i run early in the morning, when less people are on, i retain more (yeah right =P)

another was, its like loot scaling, and it depends on how many chests you ran in that area (i proved that one wrong, well enough to convince myself)

another one was, if you run too many chests per hour, they start to break a lot (happened to me repeatedly, thought this one might prevent bots from earning from chestruns)...still unsure about this, ive run a ton all at once and it eventually evens out in the end

My current theory: it actually is a random number generator. You will have all kinds of luck, and sometimes it sucks. just keep running em and, if your luck doesnt change, yer just plain unlucky.

at 68% retain, you should be able to run 750 chests per stack of keys...plus or minus 100~

Both WiKi's are always your friend.

wiki Lockpick

wikia Lockpick

Hope it helps

DING DING DING - we have a winner. This is how it works - same thing with drops. There is no magic formula (other than some items will only drop in certain areas).

Anything other than that answer and one might as well tell me what you think it is and have me agree. Since they are all wrong any other theory is as good another (and just as provable).

Humans hate random more than nature abhors a vacuum. Nature will at least let a vacuum exist, when most humans encounter "random" they come up with tricks to fix the game. It may be a lucky rabbits foot, rubbing a statues head, chanting something, and all sorts of things. In games this tends to be geared toward what others are doing - any bad luck *must* be due to others and you just have to figure out what to do in order to "fix" it.

Unfortunately many (if not most) humans tend to be happier with unfair non-random systems than with totally fair random ones. Even if they are the ones always being screwed then they still have the power to change things - with random there is nothing to do other than persevere. They want this so bad they invent their own systems (and usually get angry when others don't adopt them - especially if others have opposite lucky rites they perform). Of course, if one ever thinks gamers are bad go look up what some gamblers honestly believe makes them win money (I'm not talking rituals people do for fun - say blowing on dice and such - but real life things they do to supposedly create a lucky streak).

Everyone knows that computers don't do random numbers properly. What a lot of people don't know is that the reason it's so easy to figure out that it doesn't generate random numbers properly is that streaks of similar numbers are very common. If you have a computer roll a d100, for example, and you get a 20, the odds are far higher than they should be that the next number will be somewhere between 10 and 30.

I, too, have had a very high number of both retained and broken streaks (though far less picks run than you, around 1500), and I do think that it's just the flaws of random number generation coming through.

Information source?

Because this sounds more like poor implementation then error in design.

USE WIKI NUB! But I kid.

http://en.wikipedia.org/wiki/Hardwar...mber_generator

Random numbers don't exist in software, only in the physical world. It would be exceedingly easy to create a piece of hardware that creates truly random numbers (Atari did it, but I do not know if modern consoles do), but until computer manufacturers hop on that boat, we're stuck with formulae that can only *appear* to be random, and can only do so for a limited amount of time.

You should try coding a combat system that uses random numbers sometime. You'll pull your hair out at how often lucky/unlucky runs affect the winner. I don't know for sure, but I believe that is why spells don't use random damage; it avoids that problem altogether.

I accept the challenge.

Given a random number p_i pulled from a flat distribution for the interval [0,1[ the expectation value for the distance d = |p_{i+1} - p_i| between it and another random number p_{i+1} from the same distribution is EX(p_i) = p_i^2 - p_i + 1/2. For a series of N numbers p_1 ... p_N the mean deviation from the expected value can be calculated as k = (1/(N-1)) Sum_{i=1}^{N-1} d(p_i, p_{i+1}) / EX(p_i)

If what you claim is true and there is such a memory effect in computer generated 'random' numbers then for a sufficiently large N the value of k should be much below unity. I gave my computer a run for its money and calculated k for a sample size of N = 1e6. The result was k = 1.00048, so consider your general claim proven false.

Some poorly implemented linear congruential generators do have several modes of failure when it comes to statistical randomness of the generated numbers. However, even if ANet programmers would have been utterly inept (which I don't believe for a second) and used such a generator, it wouldn't matter anyway since there isn't a separate seed for your pickies. Between opening two chests the dice will always have been rolled enough times for any memory effect to have disappeared.

Random enough for all practical purposes is good enough for me as well.

If by 'limited amount of time' you mean 'the estimated lifetime of our Universe and then some' then I agree with you but find the point a bit moot

8 of my last 10 Lockpicks were broken on a Chest where I should have 60% retention rate.
Also 4 of my last 5 items that I salvaged were broken and my chance to break it is 29%.....

Something is going very wrong....... TT

I think instead of listening to "everyone" you need to actually listen to someone who knows how random number generators work. Even the worst ones we know of do not remotely work like that.

There is absolutely no way Atari created a "true" random number generator in your home. So far the only known method for true random is radioactive decay, even then most think that some pattern will eventually emerge. Atari used noise from a circuit that is supposedly random, however in some cryptanalysis it is found to be less than random (but then, so has pretty much everything else except radioactive decay).

While true that PRNG's eventually repeat, decent RNG's have a period long enough that it would take over your lifetime to find reach the end of it.

That is still random - even in radioactive decay you get lucky/unlucky streaks. In fact if you work a system out where that *does't* occur you have a horribly non-random sequence. If your PRNG does it enough that it is statistically invalid then you need a different PRNG - if you use them much I would suggest Knuth's The Art of Computer Programming Vol II, Seminumerical Algorithms.

There are many fast and simple algorithms that are reasonably cryptographically secure. That is, they need modern supercomputers and weeks to figure out any patterns in it.

For all intents and purposes that *is* random, in fact it is *more* random than 99% of the physical things people want to use (coin flips, tossing dice, etc). It only matters that the sequence is statistically random - it doesn't matter how it was arrived at. What type of RNG you used is irrelevant.

And yes, I have written software using them. In fact I've even had to do formal proofs (both in a university and outside of them for journals) using them. I've even had to do some of the statistical testing on large sequences.