_J_ I can't believe you can argue with a straight face that a meaningful simulation of the auction house is possible in a dev environment. And the fact that you can argue that blizzard should have been able to figure out what that treasure goblin killing was the most optimal grinding option is also ridiculous.
You've basically said that you program bots to kill a boss or something. You program one million bots to kill a boss over and over again and program them to trade with each other? Do you have any idea how ridiculous that sounds?
I know you think you are really smart but the only reason these things are obvious is because there are literally millions of people playing diablo 3. If you were dropped in an empty room with 10, 20 or even 100 other people and told you to figure out diablo 3 you would have absolutely no chance. Neither would anyone else in this thread. Because understanding complex systems is really, really hard.
I don't think they could've simulated the AH but I don't think they really needed to. This discussion began because the developers and testers couldn't get through A1 Inferno, and the justification for that has been that "they didn't have an AH," not that they suck. That's a bad justification IMO.
You know what gear is possible with ilvl 61 items and you know what stats are good. You don't need to give out 100% max rolls for ilvl 61, but you can create gear that would be in the 60th or 70th percentile (and you can find out what those values specifically are) and try it out to see if your testers/developers have a gear issue or if they're just bad and don't completely understand their own game.
Simulating the AH would have other benefits but for the purpose of where this discussion started, it's not necessary. I mean, maybe the Barbs and Monks had a reasonable justification for not getting very far with the release and pre-beta versions, but the DHs and Wizards? Like Delph has shown, you can roll through A1 in sub-55 gear easily and a weird build. :P
5)figure out a value of Z that accounts for the amount of gold these players will have
This wouldn't be difficult to model. You already have playtesters going through the game solo. Just record how much gold they amass, and average it out.
Again, this isn't relevant. We're just trying to discern how much gear players will generate. It doesn't matter if they post it to the gold AH or the RMAH. For our purposes, we assume that everyone trades and buys gear in the quest to get the most gear they can as fast as they can.
8) Somehow, figure out a value proposition for items that will no doubt be very wrong because things like the importance of one with everything, increased crit and so on are currently unknown to you
Why is this unknowable? You just sit down and figure out how to get the highest DPS for a particular class. Stacking attack speed generates more DPS than stacking, say, int, so your demonhunters stack attack speed.
I'm not sure why something they playerbase figured out is impossible for game designers to know.
_J_ I can't believe you can argue with a straight face that a meaningful simulation of the auction house is possible in a dev environment. And the fact that you can argue that blizzard should have been able to figure out what that treasure goblin killing was the most optimal grinding option is also ridiculous.
You've basically said that you program bots to kill a boss or something. You program one million bots to kill a boss over and over again and program them to trade with each other? Do you have any idea how ridiculous that sounds?
I know you think you are really smart but the only reason these things are obvious is because there are literally millions of people playing diablo 3. If you were dropped in an empty room with 10, 20 or even 100 other people and told you to figure out diablo 3 you would have absolutely no chance. Neither would anyone else in this thread. Because understanding complex systems is really, really hard.
I don't think they could've simulated the AH but I don't think they really needed to. This discussion began because the developers and testers couldn't get through A1 Inferno, and the justification for that has been that "they didn't have an AH," not that they suck. That's a bad justification IMO.
You know what gear is possible with ilvl 61 items and you know what stats are good. You don't need to give out 100% max rolls for ilvl 61, but you can create gear that would be in the 60th or 70th percentile (and you can find out what those values specifically are) and try it out to see if your testers/developers have a gear issue or if they're just bad and don't completely understand their own game.
Simulating the AH would have other benefits but for the purpose of where this discussion started, it's not necessary. I mean, maybe the Barbs and Monks had a reasonable justification for not getting very far with the release and pre-beta versions, but the DHs and Wizards? Like Delph has shown, you can roll through A1 in sub-55 gear easily and a weird build. :P
Yeah.
The post I recall said something to the effect of, "We didn't expect players to have this level of gear this quickly."
And my question is, "Well why the fuck not?"
You can discern how much loot can be generated by X players in Y hours.
You can discern the means by which they would trade it.
Why was it so mysterious?
I mean other than, "We didn't expect people to use the AH to trade items, because obviously there's no insentive to do so."
So even if you did this it would still be meaningless because there was literally no way for blizzard to do number 8 before release.
Why is that impossible? It's known now. So why, in principle, was it unknowable then?
Koreans are not simulating the economy. They are giving bots very simple commands so they can operate within an existing economy. If you can't understand the distinction between using bots to operate within a human system and using bots to simulate a human system then you don't understand the problem. The only reason the bots are capable of doing what they can do is because a human can log in and say 'hmmm the market seems to be willing to pay x gold for an amulet with +y dex and +z vit'. Without the human getting a read on the economy the bot is useless.
And regarding blizzard's developers being able to come up with a value proposition that matches their players. You're right. It's theoretically possible. But doesn't the fact that it has never, ever happened in the history of game development tell you something about understanding complex systems?
Koreans are not simulating the economy. They are giving bots very simple commands so they can operate within an existing economy. If you can't understand the distinction between using bots to operate within a human system and using bots to simulate a human system then you don't understand the problem. The only reason the bots are capable of doing what they can do is because a human can log in and say 'hmmm the market seems to be willing to pay x gold for an amulet with +y dex and +z vit'. Without the human getting a read on the economy the bot is useless.
And regarding blizzard's developers being able to come up with a value proposition that matches their players. You're right. It's theoretically possible. But doesn't the fact that it has never, ever happened in the history of game development tell you something about understanding complex systems?
I think we may be arguing past each other.
My position is that Blizzard could simulate the amount of gear that X characters with Y MF% would amass in Z hours of farming.
Your position, as I now understand it, is that they could not simulate how much a particular item would sell for.
My reply to that position is that it's irrelevant: Persons will sell items at a cost that their customers can afford. It's not as if persons would go through the trouble to farm gear and then sell it at a rate no one can afford.
You seem to be primarily concerned with the economic factors. I'm saying we don't need to consider those. We just need to know how much loot is in circulation when X characters with Y MF% farm for Z hours.
what is the purpose of your proposed simulation then?
To discern the amount of gear in circulation within various scenarios.
When 10,000 people with 200 MF farm for 100 hours a week, X gear is produced.
When 100,000 people with 150 MF farm for 70 hours a week, Z gear is produced.
It's my reply to their, "We didn't expect people to have this high of gear this quickly."
I'm not saying that Blizzard could have predicted this exact future. I'm saying that they could have simulated various futures, and designed for one of the more middle options. Somewhere between "everyone plays solo and never uses the AH" and "everyone always farms and always uses the AH."
The degree to which they were baffled by the AH a week after release seems to indicate that they didn't do any theorycrafting that hypothesized an active farming community. And my point is that you can theorycraft that shit. Could they predict the exact future? No.
But they could have laid the foundations for a world in which, a week after release, they look at the AH and go, "Oh, look. It's scenario 17."
My position is that Blizzard could simulate the amount of gear that X characters with Y MF% would amass in Z hours of farming.
The entire premise of your AI bot army suggestion is so absurd you'd be laughed out of any game development studio, but you don't even have to go further than this line right here.
How do they do this?
You make it sound like it's a matter of whipping up a formula and looking at the answer on the other side of an equation. Maybe because you play this game in such a robotic and predictable way you might think that this is a reasonable analysis tool for a larger audience, but it's completely inappropriate.
You have to know so much information about effectively unpredictable things that the only way it would work is if your entire player base is comprised of wind-up toys. Where do people end up farming? What are the popular farming strategies? How much MF are they able to get away with successfully? How fast can they clear that farming run? How long do they farm in a given session? Account for the distribution of players between 30 minutes a day casuals and 10 hour a day no-lifers? How do you account for exploits and balance oversights and bugs and the hundreds of other completely unanticipated variables?
You can come up with variables all day long that are effectively impossible to predict with any kind of confidence that would wildly affect your results. The variance would be so wild that your resultant "data" would be completely meaningless. How do you pick a "mid-range" solution when your parameters are completely made-up and you have no idea where the middle will actually be or what kind of pattern you're looking at?
Can you find a single example of a well respected development studio using a bot army to simulate intelligent players in any task more complex than highly directed server stress tests?
To discern the amount of gear in circulation within various scenarios.
When 10,000 people with 200 MF farm for 100 hours a week, X gear is produced.
When 100,000 people with 150 MF farm for 70 hours a week, Z gear is produced.
It's my reply to their, "We didn't expect people to have this high of gear this quickly."
I'm not saying that Blizzard could have predicted this exact future. I'm saying that they could have simulated various futures, and designed for one of the more middle options. Somewhere between "everyone plays solo and never uses the AH" and "everyone always farms and always uses the AH."
The degree to which they were baffled by the AH a week after release seems to indicate that they didn't do any theorycrafting that hypothesized an active farming community. And my point is that you can theorycraft that shit. Could they predict the exact future? No.
But they could have laid the foundations for a world in which, a week after release, they look at the AH and go, "Oh, look. It's scenario 17."
okay so you think blizzard is the joker from The Dark Knight, rather than a game developer in the real world. Carry on then.
To discern the amount of gear in circulation within various scenarios.
When 10,000 people with 200 MF farm for 100 hours a week, X gear is produced.
When 100,000 people with 150 MF farm for 70 hours a week, Z gear is produced.
It's my reply to their, "We didn't expect people to have this high of gear this quickly."
I'm not saying that Blizzard could have predicted this exact future. I'm saying that they could have simulated various futures, and designed for one of the more middle options. Somewhere between "everyone plays solo and never uses the AH" and "everyone always farms and always uses the AH."
The degree to which they were baffled by the AH a week after release seems to indicate that they didn't do any theorycrafting that hypothesized an active farming community. And my point is that you can theorycraft that shit. Could they predict the exact future? No.
But they could have laid the foundations for a world in which, a week after release, they look at the AH and go, "Oh, look. It's scenario 17."
okay so you think blizzard is the joker from The Dark Knight, rather than a game developer in the real world. Carry on then.
It seems rather trivial to take a certain number of internal testers and extrapolate the gear they get outward in time and assume that all testers have access to all other testers' drops
I mean you won't get it perfect but you'll at least be in the ball park
I don't actually consider the gear quality that the top players have to be a problem though. No offense J, but you're statistically insignificant. I find it to be a much greater problem that the AH is required at all, when it should be a feature rather than a core pillar of design
I think most of you guys are severely underestimating what the developers of Blizzard are capable of. They are not stupid guys.
Could they have playtested Inferno with gear considered poor, mediocre and excellent? Of course they can. For whatever reason, they didn't. It's not necessary to emulate the entire playerbase's effect on the AH to do that.
You have to know so much information about effectively unpredictable things that the only way it would work is if your entire player base is comprised of wind-up toys. Where do people end up farming? What are the popular farming strategies? How much MF are they able to get away with successfully? How fast can they clear that farming run? How long do they farm in a given session? Account for the distribution of players between 30 minutes a day casuals and 10 hour a day no-lifers? How do you account for exploits and balance oversights and bugs and the hundreds of other completely unanticipated variables?
You can come up with variables all day long that are effectively impossible to predict with any kind of confidence that would wildly affect your results. The variance would be so wild that your resultant "data" would be completely meaningless. How do you pick a "mid-range" solution when your parameters are completely made-up and you have no idea where the middle will actually be or what kind of pattern you're looking at?
It seems like they could look to games such as D2 and WoW to discern a statistical range for what percent of players will be the 10 hour no-lifers, against the 30 minute casuals. It's not as if that data is completely unpredictable. I mean, unless you think that D3 is so qualitatively different from every other game that its gameplay is completely untranslateable to other gaming contexts.
As to the particular farming strategies...since gear drops are defined by act, you wouldn't need to lay out a definate path...but you could. Look at where elite packs spawn in each zone, and discern the optimal farming strat given that elite configuration.
The only weird variable in all of this would be discerning the MF% and kill time for your farm bots, since their gear and MF would be obtained via the AH that this is all meant to simulate. So your system would have some self-reference in it. That's not necessarily problematic, though. All you have to do is discern the rate at which MF% gear is obtained, and then build that into the formula. You start out farming with 0 MF, then you get upgrades slowly and that increases MF, which increases drops, which Etc.
That doesn't seem crazy.
Again, you won't be able to predict what every particular player would do. But does it sound absurd to assume that, say, 1% of 4 million people farm for 50 hours a week?
what is the purpose of your proposed simulation then?
To discern the amount of gear in circulation within various scenarios.
When 10,000 people with 200 MF farm for 100 hours a week, X gear is produced.
When 100,000 people with 150 MF farm for 70 hours a week, Z gear is produced.
It's my reply to their, "We didn't expect people to have this high of gear this quickly."
I'm not saying that Blizzard could have predicted this exact future. I'm saying that they could have simulated various futures, and designed for one of the more middle options. Somewhere between "everyone plays solo and never uses the AH" and "everyone always farms and always uses the AH."
The degree to which they were baffled by the AH a week after release seems to indicate that they didn't do any theorycrafting that hypothesized an active farming community. And my point is that you can theorycraft that shit. Could they predict the exact future? No.
But they could have laid the foundations for a world in which, a week after release, they look at the AH and go, "Oh, look. It's scenario 17."
How do you know that they didn't do this, and their assumptions turned out to be wrong, because (e.g.) they didn't anticipate so much high end gear getting out there so quickly due to people running goblins over and over?
To my knowledge, they've not said "Never in our wildest dreams did we think that people could ever have this much stuff at this time." They've said things like "We didn't think the community would have this much gear at this time." This is consistent with them having multiple scenarios, but having assessed a set of scenarios as being the most likely, and the actual scenario did not fall in that range.
EDIT: The point is that any assessment they do on these lines is going to have huge error bars on it, and we don't know that they didn't do the best they could and the 'best they could' just turned out to be inaccurate for any number of factors that they couldn't have predicted because if they had predicted them, they'd have fixed them pre-launch.
Invictus on
Generalísimo de Fuerzas Armadas de la República Argentina
goddamn, how much do you charge for rolls like that? I'm really, really tempted to throw a figure at you for that second one, but I'm relatively low on cash.
I'm not sure. I may just wait until tomorrow because there's nothing else in that range up atm, so I don't know how to price it. Last one that was similar sold for 15m, but I set the base bid too low.
0
Options
Big Red Tiebeautiful clydesdale style feettoo hot to trotRegistered Userregular
Koreans are not simulating the economy. They are giving bots very simple commands so they can operate within an existing economy. If you can't understand the distinction between using bots to operate within a human system and using bots to simulate a human system then you don't understand the problem. The only reason the bots are capable of doing what they can do is because a human can log in and say 'hmmm the market seems to be willing to pay x gold for an amulet with +y dex and +z vit'. Without the human getting a read on the economy the bot is useless.
And regarding blizzard's developers being able to come up with a value proposition that matches their players. You're right. It's theoretically possible. But doesn't the fact that it has never, ever happened in the history of game development tell you something about understanding complex systems?
Persons will sell items at a cost that their customers can afford. It's not as if persons would go through the trouble to farm gear and then sell it at a rate no one can afford.
yeah, I have exactly 15M lying around right now so I'll avoid lowballing you on that one. I have 30M worth of overpriced offhands that nobody wants, though... :?
The entire premise of your AI bot army suggestion is so absurd you'd be laughed out of any game development studio, but you don't even have to go further than this line right here.
Posts
either way the +90 int is a 1200 dps increase
I'll list it for 50 mil just for the lulz
I don't even see the point in % life for melee wizards anyway. I can spam diamond skin why do I care about health over of mitigation?
I don't think they could've simulated the AH but I don't think they really needed to. This discussion began because the developers and testers couldn't get through A1 Inferno, and the justification for that has been that "they didn't have an AH," not that they suck. That's a bad justification IMO.
You know what gear is possible with ilvl 61 items and you know what stats are good. You don't need to give out 100% max rolls for ilvl 61, but you can create gear that would be in the 60th or 70th percentile (and you can find out what those values specifically are) and try it out to see if your testers/developers have a gear issue or if they're just bad and don't completely understand their own game.
Simulating the AH would have other benefits but for the purpose of where this discussion started, it's not necessary. I mean, maybe the Barbs and Monks had a reasonable justification for not getting very far with the release and pre-beta versions, but the DHs and Wizards? Like Delph has shown, you can roll through A1 in sub-55 gear easily and a weird build. :P
Assume everyone will trade, since that's what we're trying to model.
We don't need this number, since we're simply trying to model what X players will generate in Y hours of farming.
This wouldn't be difficult to model. You already have playtesters going through the game solo. Just record how much gold they amass, and average it out.
Again, this isn't relevant. We're just trying to discern how much gear players will generate. It doesn't matter if they post it to the gold AH or the RMAH. For our purposes, we assume that everyone trades and buys gear in the quest to get the most gear they can as fast as they can.
Since that's the game.
Again, that's nothing more than another X characters farming Y hours model. Not difficult.
Why is this unknowable? You just sit down and figure out how to get the highest DPS for a particular class. Stacking attack speed generates more DPS than stacking, say, int, so your demonhunters stack attack speed.
I'm not sure why something they playerbase figured out is impossible for game designers to know.
You don't need a supercomputer. Fucking Koreans do this shit every day.
I mean, unless Koreans have super-magical future technology to which Blizzard does not have access.
And you aren't programming bots to play the game. You're factoring in the formulas for loot generation, and inputting the stats for:
1) Number of elite packs killed.
2) MF%
Why is that impossible? It's known now. So why, in principle, was it unknowable then?
Yeah.
The post I recall said something to the effect of, "We didn't expect players to have this level of gear this quickly."
And my question is, "Well why the fuck not?"
You can discern how much loot can be generated by X players in Y hours.
You can discern the means by which they would trade it.
Why was it so mysterious?
I mean other than, "We didn't expect people to use the AH to trade items, because obviously there's no insentive to do so."
Koreans are not simulating the economy. They are giving bots very simple commands so they can operate within an existing economy. If you can't understand the distinction between using bots to operate within a human system and using bots to simulate a human system then you don't understand the problem. The only reason the bots are capable of doing what they can do is because a human can log in and say 'hmmm the market seems to be willing to pay x gold for an amulet with +y dex and +z vit'. Without the human getting a read on the economy the bot is useless.
And regarding blizzard's developers being able to come up with a value proposition that matches their players. You're right. It's theoretically possible. But doesn't the fact that it has never, ever happened in the history of game development tell you something about understanding complex systems?
https://medium.com/@alascii
I think we may be arguing past each other.
My position is that Blizzard could simulate the amount of gear that X characters with Y MF% would amass in Z hours of farming.
Your position, as I now understand it, is that they could not simulate how much a particular item would sell for.
My reply to that position is that it's irrelevant: Persons will sell items at a cost that their customers can afford. It's not as if persons would go through the trouble to farm gear and then sell it at a rate no one can afford.
You seem to be primarily concerned with the economic factors. I'm saying we don't need to consider those. We just need to know how much loot is in circulation when X characters with Y MF% farm for Z hours.
https://medium.com/@alascii
To discern the amount of gear in circulation within various scenarios.
When 10,000 people with 200 MF farm for 100 hours a week, X gear is produced.
When 100,000 people with 150 MF farm for 70 hours a week, Z gear is produced.
It's my reply to their, "We didn't expect people to have this high of gear this quickly."
I'm not saying that Blizzard could have predicted this exact future. I'm saying that they could have simulated various futures, and designed for one of the more middle options. Somewhere between "everyone plays solo and never uses the AH" and "everyone always farms and always uses the AH."
The degree to which they were baffled by the AH a week after release seems to indicate that they didn't do any theorycrafting that hypothesized an active farming community. And my point is that you can theorycraft that shit. Could they predict the exact future? No.
But they could have laid the foundations for a world in which, a week after release, they look at the AH and go, "Oh, look. It's scenario 17."
The entire premise of your AI bot army suggestion is so absurd you'd be laughed out of any game development studio, but you don't even have to go further than this line right here.
How do they do this?
You make it sound like it's a matter of whipping up a formula and looking at the answer on the other side of an equation. Maybe because you play this game in such a robotic and predictable way you might think that this is a reasonable analysis tool for a larger audience, but it's completely inappropriate.
You have to know so much information about effectively unpredictable things that the only way it would work is if your entire player base is comprised of wind-up toys. Where do people end up farming? What are the popular farming strategies? How much MF are they able to get away with successfully? How fast can they clear that farming run? How long do they farm in a given session? Account for the distribution of players between 30 minutes a day casuals and 10 hour a day no-lifers? How do you account for exploits and balance oversights and bugs and the hundreds of other completely unanticipated variables?
You can come up with variables all day long that are effectively impossible to predict with any kind of confidence that would wildly affect your results. The variance would be so wild that your resultant "data" would be completely meaningless. How do you pick a "mid-range" solution when your parameters are completely made-up and you have no idea where the middle will actually be or what kind of pattern you're looking at?
Can you find a single example of a well respected development studio using a bot army to simulate intelligent players in any task more complex than highly directed server stress tests?
okay so you think blizzard is the joker from The Dark Knight, rather than a game developer in the real world. Carry on then.
EDIT: scosglen said it better.
https://medium.com/@alascii
I prefer the way you said it.
I mean you won't get it perfect but you'll at least be in the ball park
I don't actually consider the gear quality that the top players have to be a problem though. No offense J, but you're statistically insignificant. I find it to be a much greater problem that the AH is required at all, when it should be a feature rather than a core pillar of design
Ow, my feelings.
Could they have playtested Inferno with gear considered poor, mediocre and excellent? Of course they can. For whatever reason, they didn't. It's not necessary to emulate the entire playerbase's effect on the AH to do that.
I mean, you know what I mean. The top bleeding edge of the playerbase isn't and shouldn't be the focus
It seems like they could look to games such as D2 and WoW to discern a statistical range for what percent of players will be the 10 hour no-lifers, against the 30 minute casuals. It's not as if that data is completely unpredictable. I mean, unless you think that D3 is so qualitatively different from every other game that its gameplay is completely untranslateable to other gaming contexts.
As to the particular farming strategies...since gear drops are defined by act, you wouldn't need to lay out a definate path...but you could. Look at where elite packs spawn in each zone, and discern the optimal farming strat given that elite configuration.
The only weird variable in all of this would be discerning the MF% and kill time for your farm bots, since their gear and MF would be obtained via the AH that this is all meant to simulate. So your system would have some self-reference in it. That's not necessarily problematic, though. All you have to do is discern the rate at which MF% gear is obtained, and then build that into the formula. You start out farming with 0 MF, then you get upgrades slowly and that increases MF, which increases drops, which Etc.
That doesn't seem crazy.
Again, you won't be able to predict what every particular player would do. But does it sound absurd to assume that, say, 1% of 4 million people farm for 50 hours a week?
Well now I have to do another run before I sleep.
/grats!
That's fine.
Just throw us a frickin' bone now and then.
How do you know that they didn't do this, and their assumptions turned out to be wrong, because (e.g.) they didn't anticipate so much high end gear getting out there so quickly due to people running goblins over and over?
To my knowledge, they've not said "Never in our wildest dreams did we think that people could ever have this much stuff at this time." They've said things like "We didn't think the community would have this much gear at this time." This is consistent with them having multiple scenarios, but having assessed a set of scenarios as being the most likely, and the actual scenario did not fall in that range.
EDIT: The point is that any assessment they do on these lines is going to have huge error bars on it, and we don't know that they didn't do the best they could and the 'best they could' just turned out to be inaccurate for any number of factors that they couldn't have predicted because if they had predicted them, they'd have fixed them pre-launch.
http://imgur.com/a/ecxFB
Not quite sure on pricing for the big STR one.
goddamn, how much do you charge for rolls like that? I'm really, really tempted to throw a figure at you for that second one, but I'm relatively low on cash.
actually, no
persons would do that
Skynet obviously.