As interesting as this is. I find it difficult to believe that anything will come of it. I would imagine that the way WoW does loot is probably both time and space sensitive.

By that I mean that they probably have a list of loot with values assigned that all add up to essentially 100%. Then the values are rolled by a random seed. There can only be so many different combinations of loot. If Server side they saved a value for your boss. That value may equal out to 2 specific tier drops, 2 specific random drops, and no glaive. But next week if the same random number was put in by chance, why couldn't they have rearranged the values of the %s based on the server start time that week?

I hope that makes sense but I think I rambled a bit too much.

Just another thought about the PRNG itself. What if the Algorithm(sp?) that is used to create the numbers is based on the server start time planted within it somewhere. That means the exact pattern of numbers would vary from week to week. Making it almost impossible to predict the results of a given spawn time for Illidan, and even more difficult because even if you cracked the algorithm they used you would still have a variable that you can neither control nor pinpoint, the precise server start time.

Celestara, you are correct. Nothing will come of it. And even if the algorithm for the seed is cracked it's all but useless. It's not as if one piece of loot yields another piece of loot yields another piece of loot. As often as we've all seen Loot Table A, Item 1 paired with Loot Table B, Item 3; all of us that are rational have also recalled the countless times where item A1 is paired with B4 or B5,etc. etc.

I mean knock youself out doing this. And I can almost predict someone "proving it" by "creating" an instance where they willed a specific piece of loot. But when they go to repeat said creation, it won't work.

Celestra: It sounds like you're saying that WoW probably makes only a single roll for the table. I think it's more likely that they roll separately for each loot slot rather than trying to make a single, hashed roll. The computation requirements of rolling are not large, and until the mob is dead all that's needed is either the raw random seeds or itemID numbers. While it wouldn't be hard to script a combined table based on separate loot slots I also don't see any nontrivial benefit to doing so.

*shrug* depends how many bits of information they need in order to determine the loot. If a given loot slot has 8 items of equal drop chance, they only need 3 bits. If there are 3 items of approximately equal chance (i.e. a token drop), they need a lot more bits in order to approximate 1/3 in binary. Now it could be that they request (say) a 16-bit random number per drop, or they may request some other amount of random bits and then parse it as required thereafter.

[Plans: Black Grasp of the Destroyer], [Codex of Prayer of Healing V], [Grimoire of Immolate VIII] are not currently linkable in the Ruins US battlegroup. Nor did zoning in reveal any of them out. I'm not sure if the instance can be reloaded (soft reset) and whether or not that will trigger the bosses to change their guid.

"Hey, I'm not sure this number is random enough."
"Grab the next number from the RNG."

You can do it ad nauseum, but it's not going to add to the actual randomness or strength of the method - mostly, it'll just waste server clock cycles that could go to better use.

Well, firstly, the server start time is based (usually) on milliseconds since some date - not by the time of day. Secondly, since it's based on milliseconds, it will vary no matter how precisely they attempt to time the server start, simply due to miniscule variations in conditions causing performance fluctuation. Thirdly, even if it were possible to somehow "crack the seed," doing so would be essentially useless on a cryptographic strength RNG, because there is no reasonable way for you to calculate the next result - the fact that the loot generation table is used for every single loot event in the game makes it much, much too complicated.

I have no idea of why people seem to have the funny idea that they could force the RNG itself. The notion of being able to tease the system into telling us what loot is in the instance is plausible, although still unlikely. The notion of being able to predict loot based on something (exact time that you logged into the server) is laughable. Even if they literally used the time the first person zoned into the instance and spawned all of the mobs as the seed for the instance's loot - which would be pointless, since they could just request another RNG from the already seeded server-wide loot RNG - it would not be possible to accurately predict the time that you zoned into the instance. We're not talking about being accurate to the second here. The most conservative estimate would put this at 1/100th of a second: No one has a ping that stable. I'm willing to bet that it would be difficult to manage that kind of precision from a computer on a different switch on Blizzard's internal network, let alone from an external connection.

You cannot "seed" an instance in any way. The only reasonable way you may be able to influence loot mentioned in this topic is through linking itemIDs that should not be available yet to see if they're made linkable by the mob spawn, then resetting the instance if you don't want those items. There is a reason they call them cryptographic-strength RNGs and use them for security.

Of course, now I'm probably going to get yelled at by the people who know considerably more about RNG theory than I do, but at least I hopefully haven't made quite so much of an ass of myself as some others.

True, on both counts =P.
A cryptographically secure PRNG means that you can't infer the seed from any string of outputs in a reasonably small amount of time computationally (and the length of string necessary is also unreasonably large), and that lookup tables are generally on a size scale comparable to the universe itself. Given the seed, finding the next value is easy--that is, in fact, what the PRNG itself does. That shift in division of labor doesn't change the basic point though: you still can't predict the next roll based on the previous ones.


This PRNG stuffness is all completely a fantastical aside anyways. The issue at hand is being able to see the value before losing the chance to re-roll it. Prediction and method of generation don't matter; you keep trying until you find the instance you want, regardless of how they are generated.

You can get a new raidID fairly quickly.

All you have to do is promote someone else to leader, leave the raid, and form another raid with you as the leader. Worked wonders for farming ZG rep, and should still work today.

Even if you could game the system, and force the loot you wanted, would you? Personally I think it'd be at least as bad (game) ethically as using a bug to exploit a boss.

GM's were running around with glaives and other stuff on the TTR.

This thread is a testament to how annoying and broken the RnG system is for raiding guilds at large for loot. Why would we ever want to remove the status? Because the status has more to do with getting lucky then it has to do with actually clearing Black Temple and being an elite guild. Because it puts guilds in Sunwell at an advantage who have been lucky over guilds that have been screwed on this drop for no real good reason. Other then " Random Loot is Random "