|
Is taking up gathering jobs very helpful if you're doing crafting ones? I'm thinking of maybe trying out some crafting once I finish the Heavensward content, but I'm worried I might not have enough gil to make much headway (I don't really mind blowing almost everything I have, since there aren't many other uses for gil, but I only have like 850k right now).
|
#
?
Aug 10, 2018 17:58
|
|
|
|
| # ? May 24, 2024 16:25 |
|
|
Rainuwastaken posted:Pfft, Warframe's RNG is fine. See, this whip has a 500% chance to crit every time I swing it. Consistency! There were similar exploits to Warframe's RNG in WoW that abuse their entropy compensation at least twice that I count. One related with bad luck protection during MoP/WoD and maybe as early as Cat. I believe this is common knowledge. You could "save" your luck by doing lovely runs, and waiting for a long streak of no loot, and then go to a good raid. The other was during Vanilla through the end of WOTLK. Very, very few people know about that. But basically - a very few amount of people managed to control crit chances through extensive statistical analysis. The signature was not easily viewable by humans - you had to parse scripts from the public logs and check for the entropy in the distribution of the values. At peak ICC there were five feral druids who could achieve consistent >100% simulated damage, all around 105%, effectively compensating the "5% hybrid tax" that existed at the time. I don't know what methods the others used but mine involved custom snapshot rotations. I strongly suspect they did the same. Through perfect manipulation of crits there was however a Korean Druid who clipsed us all and made that #1 DPS spot impossible I think he did 107%-110% simulated damage and his rotation didn't look like anything we knew. I'm pretty sure that guy could figure out how to win the jumbo cactpot reliably if he wanted.
|
|
#
?
Aug 10, 2018 17:58
|
|
|
Elentor posted:I'm pretty sure that guy could figure out how to win the jumbo cactpot reliably if he wanted. Only method I need is buying 6969, 0420, and 8008 every week. Any other choices would be a violation of my moral code.
|
|
#
?
Aug 10, 2018 18:00
|
|
|
Ytlaya posted:Is taking up gathering jobs very helpful if you're doing crafting ones? I'm thinking of maybe trying out some crafting once I finish the Heavensward content, but I'm worried I might not have enough gil to make much headway (I don't really mind blowing almost everything I have, since there aren't many other uses for gil, but I only have like 850k right now). Gathering is also a good source of money since there are plenty of crafters that either don't want to do it themselves, or don't want to wait on time-gated materials.
|
|
#
?
Aug 10, 2018 18:15
|
|
|
Elentor posted:There were similar exploits to Warframe's RNG in WoW that abuse their entropy compensation at least twice that I count. One related with bad luck protection during MoP/WoD and maybe as early as Cat. I believe this is common knowledge. You could "save" your luck by doing lovely runs, and waiting for a long streak of no loot, and then go to a good raid. Huh, I'm a founder and had absolutely no idea. That's super interesting though. I'm always of two minds when people crack open a game and expose its guts for the world to see. Part of me hates how it kinda strips the magic out, but on the other hand it's utterly fascinating to see exactly how something works (and how easily broken it is sometimes).
|
|
#
?
Aug 10, 2018 18:15
|
|
|
Ytlaya posted:Is taking up gathering jobs very helpful if you're doing crafting ones? I'm thinking of maybe trying out some crafting once I finish the Heavensward content, but I'm worried I might not have enough gil to make much headway (I don't really mind blowing almost everything I have, since there aren't many other uses for gil, but I only have like 850k right now). It's a time investment, but gathering your own mats saves you a lot of money. I leveled all DoH and DoL classes pretty evenly and the only thing I might regret about it is prioritizing that above MSQ on several occasions.
|
|
#
?
Aug 10, 2018 18:17
|
|
|
Gathering isn't too painful.. there's a barrier when you get into high end with regards to scrips, but it's not too bad. You can passively make a few 100k a week filling in gaps on the MB with mats. It's not too bad, but right now is the between patch lull and everything is super down on the MB. When patches drop, there's a huge opportunity to offload stuff you've stockpiled and really go nuts. When 4.2 hit, I made ~8 million that first week just selling retainer fetched mats. The only reason why I'm kind of broke now is because I did the endgame crafting gear meld lottery and watched 80k disappear into dust over 200 times.
|
|
#
?
Aug 10, 2018 18:24
|
|
|
Ytlaya posted:Looks like Michael Cera.
|
|
#
?
Aug 10, 2018 18:37
|
|
|
Solitair posted:It's a time investment, but gathering your own mats saves you a lot of money. I leveled all DoH and DoL classes pretty evenly and the only thing I might regret about it is prioritizing that above MSQ on several occasions. It's probably simpler to gather 1 thing that nets a tidy profit, and then buy the things you need tbh. Traveling around to get a dozen different mats that also have to be crafted into another mat by different DoHs before being used in the final recipe can get ridiculous.
|
|
#
?
Aug 10, 2018 18:38
|
|
|
Head Hit Keyboard posted:Wait wha? What does that even mean? This is what Warframe's RNG looked like:  This image isn't made-up by someone else, it's straight out of a Warframe's Dev. Notice how it is "perfectly balanced" if all you want is the average. There are 50% 0s, 50% 1s. Every RNG has something called its period. It's how long it takes to revolve around itself, so to say, and finish, well, a period. Then it restarts. Good, modern RNG generators have periods that are extremely long, we're talking 10^70+ long. Numbers bigger than the amount of everything that has been processed. However, for the most part, a lot of PRNGs have had (and still do, to this very day, in many implemenetations) extremely small periods. The period in Warframe was 16. It basically alternated between 00, and 11, with one exception. I'm gonna give you an example of how to exploit it, using made-up skills. Imagine you have a period of 2: 1 and 0, which is not that far away from Warframe's RNG. Suppose you have an auto-attack that hits every 0.5s, and a big strike that deals 500% damage that you can use once every second. Assuming you have 50% crit, you AA as soon as you touch the enemy, and you crit for 100% more damage, the optimal rotation is to AA+Big Strike, then immediately cancel your attack until Big Strike is back, for a guaranteed 1100% damage every second. If you had a 50% crit chance perfectly split between them, then your DPS would be (2 * 1.5 + 5 * 1.5) = 1050% instead of (1 * 1 + 5 * 2) = 1100%. You now have a 5% DPS advantage by cancelling an attack. There are other things that are not related to the periodicity. For example, it's very unlikely that each character has its own PRNG seed, but it's possible that instances have their own RNG seed in order to decouple them from server state (or even if they have their own servers). Depending on the PRNG, even algorithms with very long periodicity, and I mean, even the >best ones< have very similar starts and take a while to diverge. The .NET implementation of random, which many people are gonna use because it's the default, is notoriously bad. Here's an image of my own. Each row is the sequence of a different seed of your build-in System.Random that anyone on Windows can use:  I'm not gonna go in-depth about all the properties and exploits you can extract from this but this isn't unique to System.Random is all I can say.
|
|
#
?
Aug 10, 2018 18:38
|
|
|
Warframe is a poo poo pile though
|
|
#
?
Aug 10, 2018 18:40
|
|
|
Elentor posted:The other was during Vanilla through the end of WOTLK. Very, very few people know about that. But basically - a very few amount of people managed to control crit chances through extensive statistical analysis. The signature was not easily viewable by humans - you had to parse scripts from the public logs and check for the entropy in the distribution of the values. At peak ICC there were five feral druids who could achieve consistent >100% simulated damage, all around 105%, effectively compensating the "5% hybrid tax" that existed at the time. I don't know what methods the others used but mine involved custom snapshot rotations. I strongly suspect they did the same.
|
|
#
?
Aug 10, 2018 18:58
|
|
|
FractalSandwich posted:Can you elaborate? I'm not surprised that you might start to see evidence of patterns in combat logs with a large enough sample size, but it's not clear to me how that would necessarily be exploitable in any way, especially in a server-based game with as many moving parts as WoW. Uh, I mean, I just gave an example a few posts above, but if you need even more specifics: * There are literally web apps that reverse-engineer sequences for you based on known databases since the majority of algorithms are predictable, and as I've shown, there's been plenty of history of real games loving RNG up. * There are other PRNG traits other than warmup time and periodicity that are exploitable. * There are situations where you can extract pure samples. If, for example, you enter an instance alone and it is known that each instance has its own RNG stream, you can make an educated guess of which algorithm the company is using and write a tool to reverse engineer possible sequences from the stream from the combat lot. If you know the stream, you can play God with the RNG by writing a custom, CPU-intensive tool that will reliably predict future outcomes in order to maximize your RNG and reduce its entropy to a minimum. If you don't, then you can apply statistical analysis to the observed RNG properties. For example, many high-performance RNGs have upper boundaries to the extent with which a sequence can be generated, so a 1 in a trillion chance of 0s and 1s in a row is sometimes literally impossible to happen. Others have boundaries constrained by the deviation of the sample size: That 1 in a trillion chance cannot happen as the first inputs of a sequence, but can if the sequence starts going on long enough. With a large enough sample size you can determine the expected type of noise (white noise, blue noise, etc). Whether you have experience with fourier transformations, bayesian statistics, anything goes here. Even if you know the frequency with which a fairly good RNG will start to fail in a crowded environment, you can turn that into an advantage. Here's a fun property of a traditional 2^16 period RNG: Every number alternates between odd and even. I can't list every predictable RNG property here, I didn't do it even in my own thread about game dev., but I hope these examples are more practical for you.
|
|
#
?
Aug 10, 2018 19:20
|
|
|
FactsAreUseless posted:You have weird ideas about what Michael Cera looks like.  Looks like brown Michael Cera with a wide nose 
|
|
#
?
Aug 10, 2018 19:23
|
|
|
Does battle retainer item level matter past the point where the chat log starts saying that your gear will improve the chances of it finding rare items, or is it just a cutoff?Solitair posted:It's a time investment, but gathering your own mats saves you a lot of money. I leveled all DoH and DoL classes pretty evenly and the only thing I might regret about it is prioritizing that above MSQ on several occasions. OneEightHundred fucked around with this message at 19:46 on Aug 10, 2018 |
|
#
?
Aug 10, 2018 19:42
|
|
|
Gruckles posted:It's probably simpler to gather 1 thing that nets a tidy profit, and then buy the things you need tbh. Traveling around to get a dozen different mats that also have to be crafted into another mat by different DoHs before being used in the final recipe can get ridiculous. I never had that good a handle on what makes the most money, especially in relation to the effort it takes to gather it. The most profitable things I usually sell on the MB are the HQ equipment I get from quest rewards.
|
|
#
?
Aug 10, 2018 19:44
|
|
|
Elentor posted:This is what Warframe's RNG looked like: Do you know much about XIVs RNG? This is fascinating.
|
|
#
?
Aug 10, 2018 19:47
|
|
|
Solitair posted:I never had that good a handle on what makes the most money, especially in relation to the effort it takes to gather it. The most profitable things I usually sell on the MB are the HQ equipment I get from quest rewards. e: Most of money lately has been selling Moggle weapons at 300% markup. OneEightHundred fucked around with this message at 19:50 on Aug 10, 2018 |
|
#
?
Aug 10, 2018 19:48
|
|
|
Poops Mcgoots posted:Only method I need is buying 6969, 0420, and 8008 every week. Any other choices would be a violation of my moral code. You forgot 0710
|
|
#
?
Aug 10, 2018 19:57
|
|
|
OneEightHundred posted:Does battle retainer item level matter past the point where the chat log starts saying that your gear will improve the chances of it finding rare items, or is it just a cutoff? Just a cutoff, I haven't updated retainer gear for a couple of patches now because it hasn't been necessary. Eventually the ilvl target will go up again, probably in 4.4.
|
|
#
?
Aug 10, 2018 19:58
|
|
|
oh no blimp issue posted:Do you know much about XIVs RNG? This is fascinating. Not as much as I'd like. All I know is that it's one of the worst I've seen in an MMO. It's the streakiest RNG I've taken note with several deviations from the norm all across the board over dozens of thousands of samples. It also tries really hard to prevent you from figuring it out. For example, long crafting repetitions will scramble the values and start forcing misses on you. To turn it into something practical: Understanding that there was a reduction in the Good % Proc from ARR to HW from 25% to 20% had to involve crafting many different items because the Proc value goes down the longer you stay crafting the same item. So whenever you see someone somewhere saying something inconspicuous like "the odds of ___ are x%", keep in mind someone had to do this:  And that every single time the values change, before you can get access to the sweet theorycraft that handed it to you, the people who noticed the change are met with hundreds of "it's just RNG being RNG, you're biased" and so on. FFXI apparently had some crazy shenanigans like facing a constellation increased the odds of your crafting success so on top of all the noise we still have to be paranoid and superstitious about external factors influencing the RNG which doesn't help. Elentor fucked around with this message at 20:10 on Aug 10, 2018 |
|
#
?
Aug 10, 2018 20:03
|
|
|
i hear if you craft on top of the giant Rhalgr statue you get to catch Mew
|
|
#
?
Aug 10, 2018 20:06
|
|
|
Huh, didn't know it forced misses to obfuscate itself. That would explain some things.
|
|
#
?
Aug 10, 2018 20:22
|
|
|
Elentor posted:Uh, I mean, I just gave an example a few posts above, but if you need even more specifics: I understand wanting to be cagey about it, but given that it was apparently fixed years and years ago, I don't see it as a terribly irresponsible disclosure at this point. Unless Blizzard specifically asked you not to talk about it, in which case I've already made you say too much.
|
|
#
?
Aug 10, 2018 20:22
|
|
|
Ouhei posted:There's some really great ones in there and I hope they do what they did with the Tank stuff and use it as Tome gear so we get versions for other roles. They've actually done it with all of them so far. The tank winner became the Lost Allagan tome set, as you mentioned. The Healer winner became the Skalla set and the Caster winner is the Bonewicce stuff from Swallow's compass. So, presumably, they'll make sets for everyone based on these winners too.
|
|
#
?
Aug 10, 2018 20:26
|
|
|
Elentor posted:Not as much as I'd like. All I know is that it's one of the worst I've seen in an MMO. It's the streakiest RNG I've taken note with several deviations from the norm all across the board over dozens of thousands of samples. It also tries really hard to prevent you from figuring it out. That's really interesting, how significant is the reduction in proc values and how likely are you to see it in the wild as a normal player? Is there a difference between making say 10, or is it on the scale of thousands?
|
|
#
?
Aug 10, 2018 20:29
|
|
|
FractalSandwich posted:Okay, cool. Thanks. So there are all these kinds of attacks that might theoretically be possible in some cases, but I'm especially interested in this concrete example of one you actually carried out on a live game in the wild. I'm interested in what, specifically, was the nature of the exploit you used in WoW? How did you go about setting up for it, and how did the execution differ from playing the game as intended? math.random() is deterministic and basically any prng that doesn't have some sort of way to use quantum mechanics to be involved in generating the seed, you're going to be able to reverse engineer and figure out the seed. if you're using a large enough data set, like everyone posting their fight logs for example, you can give it to any college sophomore with a copy of R and 2 semesters worth of statistics to fart out where the patterns lie.
|
|
#
?
Aug 10, 2018 20:29
|
|
|
I'm paying attention to the ambient dialogue, and there is a LOT of smut in this script. Good lord, how did some of it get past the censors, haha.
|
|
#
?
Aug 10, 2018 20:47
|
|
|
pray return, m'lord
|
|
#
?
Aug 10, 2018 20:48
|
|
|
oh no blimp issue posted:That's really interesting, how significant is the reduction in proc values and how likely are you to see it in the wild as a normal player? Is there a difference between making say 10, or is it on the scale of thousands? Keep in mind I no longer have any idea what is and what isn't in the game, I haven't actually crafted since late HW., but you're not gonna see it as a normal player unless you're going for server first crafted items and knew the super secret rotations to keep items going forever. EponymousMrYar posted:Huh, didn't know it forced misses to obfuscate itself. That would explain some things. This guy gets it. FractalSandwich posted:Okay, cool. Thanks. So there are all these kinds of attacks that might theoretically be possible in some cases, but I'm especially interested in this concrete example of one you actually carried out on a live game in the wild. I'm interested in what, specifically, was the nature of the exploit you used in WoW? How did you go about setting up for it, and how did the execution differ from playing the game as intended? I mean, it's pretty much what Phone said. Outside of literally giving you the code I've given you a lot of information that's not typically talked about. The extent of what I did in WoW was pretty tame, I had a custom addon that scanned for probabilistic patterns and did a heavy simulation on the fly telling me when to expect the maximum EV off of a snapshot. It was fed 5 months of raw data. That made me fluctuate between fifth and second place in global DPS rankings and a stable second place in simulated DPS % at a very consistent 105% against ICC Heroic Bosses. As for the actual literal manipulation, only one Korean dude as far as Ferals go knew how to do it, and I can't tell you the exact mechanisms of what he did because I don't know. He stopped playing during Cataclysm and then that was it. Analysis of his logs showed he was manipulating misses and crits. I don't know how, my bet is he simply was more efficient at reverse-engineering the data and had found a way to solve it rather than approximate a solution like I and others did. The people who go the extra mile don't usually spill the beans. If you want more stories of crazy nerdy poo poo my team and I have done I really don't mind about spilling the beans, I don't particularly care about talking of shadier poo poo because I make a point of never doing anything in my account that is ban-worthy and none of us do to be honest. But as far as RNG manipulation we operated pretty much raw statistics.
|
|
#
?
Aug 10, 2018 20:50
|
|
|
Elentor posted:Keep in mind I no longer have any idea what is and what isn't in the game, I haven't actually crafted since late HW., but you're not gonna see it as a normal player unless you're going for server first crafted items and knew the super secret rotations to keep items going forever. Well thank you, that's pretty interesting. (And the rest of the post I didn't quote) I presume you use an automated solution for your data collection? I can't image you hand crafting like 5000 of the same item repeatedly. Or are you implying that as you continue working on a single item the proc chances will go down?
|
|
#
?
Aug 10, 2018 21:11
|
|
|
Ytlaya posted:
|
|
#
?
Aug 10, 2018 21:24
|
|
|
Elentor posted:Keep in mind I no longer have any idea what is and what isn't in the game, I haven't actually crafted since late HW., but you're not gonna see it as a normal player unless you're going for server first crafted items and knew the super secret rotations to keep items going forever. See, this is the part that gets me curious. Given that with sufficient gear levels, pre-SB, it was trivial (if a bit tedious) to 100% items every time regardless of NQ materials, the apparent existence of an infinite loop boggles my mind. The usefulness of said loop seems dubious after you've geared yourself up, and given the fixed number of possible inputs available to the player, I really don't see how this could happen - I'm not saying that it's impossible, but that I don't understand it. It's beyond me. It must be have been super secret, because none of the crafters I know would have even considered such a thing as possible
|
|
#
?
Aug 10, 2018 21:32
|
|
|
Sounds like a specialist rotation, which almost nobody bothered to learn how to use because the skills were annoying and required paying attention. I never learned myself, because specialist rotations also weren't necessary. But I'm pretty sure you could loop things. Stormblood took a much better approach to specialty skills - just let me hit one button and get my 3 Inner Quiet stacks.
|
|
#
?
Aug 10, 2018 21:44
|
|
|
Michael Cera would be a bard not a gunologist
|
|
#
?
Aug 10, 2018 21:45
|
|
|
Edit Okay that went through twice even though the forums doesn't allow that?
|
|
#
?
Aug 10, 2018 21:45
|
|
|
Safeword posted:I'm paying attention to the ambient dialogue, and there is a LOT of smut in this script. Good lord, how did some of it get past the censors, haha. It doesn't, right there on the ESRB: rated Teen, content descriptors: Suggestive Themes, Sexual Themes, Language. FFXIV is both incredibly subtle and incredibly thorough with it though. I mean, hang out around the Missing Member in Limsa, the area where the female pirate crew hangs out. And then realize how much of a double entendre the bar's name is and how well it fits in with everything.
|
|
#
?
Aug 10, 2018 21:45
|
|
|
oh no blimp issue posted:I presume you use an automated solution for your data collection? I can't image you hand crafting like 5000 of the same item repeatedly. I usually automate data collection as long as I'm not botting, with the exception of FF Opera Omnia which I botted on an emulator because there was literally no way in hell I was gonna play through all the combats needed for the data gathering. For FFXIV, my automation was a TPT that scanned the results and added to a DB coupled with a parser to scan patch notes which drove me crazy because the fuckers changed the formatting in the patch notes every time. I never botted the actual crafting though, I used macros and alt-tabbed to do something else and my scanner saved the results. At its peak while it still worked, it'd save values from the MB automatically:  I could have automated it more but all in all I had had enough fun coding the stuff and any more would be overkill. Having an updated database made was useful for the end-of-expansion when we were crafting gear to help newbies:  Olesh posted:See, this is the part that gets me curious. Given that with sufficient gear levels, pre-SB, it was trivial (if a bit tedious) to 100% items every time regardless of NQ materials, the apparent existence of an infinite loop boggles my mind. The usefulness of said loop seems dubious after you've geared yourself up, and given the fixed number of possible inputs available to the player, I really don't see how this could happen - I'm not saying that it's impossible, but that I don't understand it. It's beyond me. It must be have been super secret, because none of the crafters I know would have even considered such a thing as possible There's no real usefulness for it. People had managed to reach infinite steps pre-specializations. After specializations it became kinda... trivial to be honest, but then crafting became so easy that even specializations became superfluous. But it's basically how Godbert and I did the first 100% HQ Chance crafter attire of Excalibur in HW. The two existing HQ ones before ours were 70 and 80% respectively (according to the crafters):  I don't have any fancy 99-step screenshot proof but those were out in the wild like I said pre-HW. I also didn't make my rotation a secret during HW, I was pretty open about it to the linkshell.
|
|
#
?
Aug 10, 2018 21:45
|
|
|
Meanwhile I'm sitting here at 70 on all my crafters, barely able to use the macros I stole from a friend. You guys are seriously impressive.
|
|
#
?
Aug 10, 2018 21:49
|
|
|
|
| # ? May 24, 2024 16:25 |
|
|
Steal macros from the site that makes them for you based on your stats and cross-class availability
|
|
#
?
Aug 10, 2018 21:51
|
|