As I have mentioned in previous posts, experimental economists create very simple serious games to study economic interactions. Experimental economists rely heavily on the notion of “induced value”: a good economic experiment must guarantee that players care about the things that the economic theory being tested assumes they care about. With that guarantee in hand, we can use the data to answer questions like “how does this feature of the environment affect behavior?” and “how well does this equilibrium concept (like Nash Equilibrium) predict behavior?” If we don’t know what people value, we can’t answer these questions.
In this post, I want to explore how one might extend a central element of EE from simple laboratory settings to complex virtual worlds.
Vernon Smith (who won the Nobel for being the father of experimental economics) laid out the elements of induced value in 1976, and clarified them in 1982. Modifying Smith’s discussion to reflect my own views, inducing value requires four features:
- Rewards. There must be some reward of value that participants can pursue within the experiment.
- Nonsatiation. Participants must prefer more reward to less (they cannot be indifferent, no matter how much they earn).
- Saliency. Participants must be aware of how game outcomes map into rewards.
- Dominance. Rewards must be large enough, relative to participants’ uncontrolled preferences, that those other preferences have the same size and type of effect on behavior in the controlled setting as they would in the real world.
Satisfying all four of these conditions allows the experimenter to “induce” value preferences in participant that the economic theory assumes hold in the real world. (Email me for a paragraph describing the differences between my view and Smith’s)
Virtual worlds do a very poor job of controlling preferences, primarily because they fail the “dominance” test. While participants in Entropia are indeed motivated to earn money, this is far from their only goal. They also seek the simple pleasures of virtual combat, along with the social pleasures afforded by collaboration and leadership. Second Life, as a world without goals, deviates even further from the tenets of induced value. Only about 300,000 of the roughly 1,000,000 SL subscribers spent a single dollar during the Month of April, 2007, and only about 35,000 of them had positive cash flow. Clearly, SL subscribers are using SL for pleasures that a researcher is unlikely to know, and is certainly not controlling. Student participants won’t be able to learn any more than researchers if their competitors are more interested in changing their avatar’s hair and chatting with friends than earning profits.
For the behavior of a single person to be generalized to real-world settings, one must induce value successfully for everyone that person might interact with, even indirectly (by interacting with someone who interacts with that person). Assume one person is trying to produce and sell goods, and you have successfully induced them to value what you want them to (money). If the sellers' customers are controlled by human participants, and their value is not induced by strong financial incentives, the experiment is still tainted, because the seller’s optimal action is influenced by whatever uncontrolled preferences the buyers have.
I see two ways to induce value in virtual world RPGs, each with its own problems. One way is to rely heavily on non-player characters (NPCs) who have preferences for goods that are controlled by the game developer. In the example above, one could then simply tell the seller the algorithms governing the buyers’ decisions. This approach hinders one of the particular strengths of virtual worlds—the ability to have large numbers of people interacting. Moreover, the artificial intelligence governing NPC behavior would have to be pretty sophisticated. We don’t yet have good rules for determining optimal behavior of traders in simple financial markets, so players would probably be able to exploit imperfections in NPC economic decisions.
An alternative solution is to induce value for every single player in the game by assigning a “consumption value” to every tradable good. When players consume the good, they convert it into cash (received whoever is running the game), but don’t get any other pleasure out of it. For example, one could imagine a game with a set of goods like flour, butter, eggs, sugar, milk, whisk and oven. The edible items would have limited consumption value, the whisk and oven would have none, but they could all be combined to create a new good (cake) that would have a high consumption value. This would achieve “dominance” because goods with high consumption value would still provide no direct entertainment value. (In contrast, weapons in Entropia can be sold for cash, but ultimately are consumed by people who like combat; the value is unknowable to the economist, because they can’t know which weapons allow combat to the be the most fun.)
While artificial consumption values sounds simple, they can still serve as the driver for complex economies and entertaining game play. Economies can be very complex because the consumption value of a good like cake can drive an entire supply chain, reaching back from the bakery to the food wholesalers to the farmers to the tractor builders and miners. Game play can still be fun because people would likely enjoy the many economic interactions occurring along the supply chain (but not enjoy them so much as to threaten dominance).
Artificial consumption value will pose special challenges to those in marketing and the behavioral sciences, who wish to study and understand people’s preferences. A platform for business research and education can and must support worlds in which preferences for consumption arise naturally. However, these worlds would have to be segregated from worlds that meet the standards of experimental economics.
References:
Smith, Vernon L, 1976. “Experimental Economics: Induced Value Theory,” American Economic Review, 66 (2), 274-79.
Smith, Vernon L, 1982. “Microeconomic Systems as an Experimental Science,” American Economic Review, 72 (5), 923-55.
An interesting topic, and one I think Thomas Malaby in "Parlaying Value: Capital in and Beyond Virtual Worlds" in Games & Culture Journal covers very well, and perhaps provides a good theoretical framework to examine many of these issues further.
David
Posted by: David Grundy | Jun 06, 2007 at 11:18
"(In contrast, weapons in Entropia can be sold for cash, but ultimately are consumed by people who like combat; the value is unknowable to the economist, because they can’t know which weapons allow combat to the be the most fun.)
While artificial consumption values sounds simple, they can still serve as the driver for complex economies and entertaining game play. Economies can be very complex because the consumption value of a good like cake can drive an entire supply chain, reaching back from the bakery to the food wholesalers to the farmers to the tractor builders and miners. Game play can still be fun because people would likely enjoy the many economic interactions occurring along the supply chain (but not enjoy them so much as to threaten dominance)."
It seems to me the second paragraph here is quite the way Entropia operates. Lets take a Mining Extractor as an example. In order to make a desirable mining extractor you need the Blueprint. In order to be sucessfull with the "blueprint" you need "skills". If you have suitable skills and the required blueprint you then need the materials required for a crafting attempt. Lets say two Screws, some Platinum and some enmatter. Those all funnel back to the miner. If your crafting attempt is successful you have created a mining extractor with a true trackable "value" both inherant and market driven.
It would seem to me, this system provides eveything you would need for in-depth economic analysis.
Posted by: Nate Randall | Jun 06, 2007 at 12:52
When I hear the term "induced value," I think I should write a scholarly paper about Random Unsung's Aging Hippie line of clothing in Second Life. All he did was take the tie-dye texture out of the Linden-provided free library in inventory, and slap it up in "appearance" mode, save it, and put it in a tie-dyed box and set the price to $5. Then he fished around on the Internet and got a few more tie-dyed textures here and there. Then he made boxer shorts, pajamas, etc. in appearance without even templates and put those in a box for $10.
Each week, some $100 worth of these awesome creations sell -- literally made out of thin air, literally made from nothing, made from stuff that anybody buying them could get themselves out of their own inventory for free. Often, Random tries to explain this to people. But they love the experience of buying a box called "Aging Hippie" and seeing the picture of Random and even meeting Random in his Birkenstock sandals and Pink Floyd t-shirt and wire-rim glasses.
So it's all in the experience and the packaging, and virtuality affords you an opportunity to accelerate, amplify, and replicate this in ways never experienced in RL.
Posted by: Prokofy Neva | Jun 06, 2007 at 13:16
I have a number of concerns about induced values in MMOs, but here I'll make an quick argument about why they're unnecessary.
I'm sure you know this, but for the sake of others who haven't studied as much economics, a great deal of economic study--theory, RL empirical work, and lab experiments--are concerned with comparative statics. Essentially that means that a theory may predict that under condition A, something (prices, quantity, whatever) is larger than under condition B. The details about how much larger are often ignored. Only occasionally do economists feel they need to know a specific predicted value of anything.
If the research question can be adequately addressed with comparative statics ("would a double auction result in more trade than a posted offer market?"), then induced values will generally be unnecessary. Simply knowing that things have some value should be enough--WoW players think that a "Major Healing Potion" is a thing of value, because it helps them die less and bash more. 'nuff said.
Are your research questions of the type where you need point predictions?
Posted by: Timothy Dang | Jun 06, 2007 at 16:34
OK, a bit more:
It's still difficult to imagine what sort of VW you're imagining. I
was going to suggest that there's no need to induce value via RL cash
payoffs when you could do so via game-world cash payoffs. However, if
you're trying to make a world in which there's no within-game
consumption value to any goods, then inducing value via game-world
payoff wouldn't work since game currency would be valueless in and of
itself.
It's certainly common in games to have a significant fraction of loot
items which are purely loot items--their only value is in selling them
to NPCs.
While I'm still doubtful whether the induced-value approach would be
particularly useful as a research method, I'm pretty confident that
what you describe above is actually impossible.
>> Dominance. Rewards must be large enough, relative to participants'
uncontrolled preferences, that those other preferences have the
same size and type of effect on behavior in the
controlled setting as they would in the real world.
There's two things which seem to me to be problems in the above
condition. First, that the ability to predict not merely the direction
of effect but the size of effect... well, it boggles me, frankly. I'll
correspond with you elsewhere about the difference between your
notions and Smith's (disclosure: I was introduced to experimental
economics in Smith's lab). One thing which seems like a problem about
it is that it sounds as if it is not simply a matter of induced value,
but also of the realization of what happens in the market/game/etc.
>> While artificial consumption values sounds simple, they can still
serve as the driver for complex economies and entertaining game
play... Game play can still be fun because people would likely enjoy
the many economic interactions occurring along the supply chain (but
not enjoy them so much as to threaten dominance).
This sounds like an impossibility. It can be a challenge sometimes
even in a laboratory experiment. In a complex environment where
players have more choices available to them (including choices of what
to do in RL instead of in the game), fun and induced value seem very
antithetical. Keep in mind that there's already an important minority
of MMO players who do it particularly with interest in being in-world
businessfolk, with no RL payoff.
Posted by: Timothy Dang | Jun 06, 2007 at 16:48
@Nate, you forgot - again - to tell us : how comes , that newcommer , without having any skills and using the lowest " mining tools " , " found " mineral resources worth....how much was that ?$ 5,000 ? And ofcourse, nobody ever heard of him/her , not before and not after the took the money. What was that avatar, 3-4-5 days old in game, right ? Was it MindArk's avatar ? Or a RL friend/relative ?
This is quite the way Entropia operates.
The distribution of " values " in Entropia is absolutely NOT related to avatar's skills or to the virtual items the avatar uses. I myself - and many others - i had " success " , using a total unskilled avatar and the lowest items ; i found " values " worth $ 100-200 ( considering the declared exchange rate/ covertibility of in-game coins into $ ) , then.....surprise : MindArk REFUSED to transfer the respective $ into my bank account. MindArk's explanation ? " According to EULA bla-bla, we are NOT under the obligation to approve any withdrawal request ; we're awared that your request is pending since 6 months ago , but please notice : MindArk never made the promise to change the in-game currency into real cash, but only stated that the player CAN and MAY ASK ; also, if you look at MindArk's official annual financial report , please notice that there is NO such a provision like ' this ot that amount of $ reserved for players' withdrawals ' "
Signed : " regards, MindArk Support team ".
This is the way Entropia operates. They never provided me a reason for the denial , and i've never got " my money " in my bank account.
But i know - and also you know - of that player who using an exploit called " trapped mob " got $ 8,000 and withdrawed the cash . MindArk had NEVER
addressed that dirty affair , never provided an official point of view to the " community ".The same , about the " hunter-bot " : the exploiters got the money, they are still in game and MindArk just keeps....the usual total silence.And the bribe .
Entropia is full of PROVED hackers , scammers and exploiters , and the only way to " earn " some cash there is to bribe a MindArk employee or staff. Or to be a RL friend.
I will not transform TN in a fanbois forum as your site is and i won't infest this site with a meaningless " war " with you , as it seems Prokofy have with csven. Don't feel free to answer to my post , but if you have the guts to contradict my assertions , do it at your site ; but we both know you won't ,and i can tell why : because my " allegations " are the truth and MindArk or/and you simply cannot denie the facts.
Posted by: Amarilla | Jun 06, 2007 at 16:51
I just thought I would weigh in on a few points, and ask a few questions.
Nate, I have not spent much time in Entropia. I know that MindArk pegs a fixed exchange rate for currency (10 inworld dollars to the US dollar), unlike Second Life (which allows a floating exchange rate. Do they also fix the value of items that require an extensive production process? (They could do this by buying and selling unlimited quantities at known prices). If they do this, then they have induced value. But if they set prices only for very fundamental items that are inputs to the production process, then they are not inducing value.
For example, I spoke briefly with a weapons manufacturer in Entropia. My understanding is that he charges whatever he can get for those guns, and that people buy them not only for their ability to serve as a factor of production (e.g., they can be used to kill creatures that provide valuable loot), but people also like them because they are fun to own and use. Their economic value is therefore detemined by players' tastes, and not controlled by MindArk.
Timothy, you make two points: that induced value might not be necessary to show directional effects of a treatment, and that artificial consumption values are incompatible with "fun."
On the first point, I wonder whether things have changed at the University of Arizona since Vernon Smith left for George Mason, as what you said about induced value would have been sacriligious 10 years ago, when no one would publish any experimental economics paper without very pure induced value--which meant not only having artificial values for goods, but also running experiments devoid of any context. (Players could only be named "A" and "B," not something meaningful like "auditor," for fear of contaminating the lab environment with baggage.) I always thought that such demands were overblown as long as researchers are holding all of those factors constant and searching only for directional treatment effects. Has that viewpoint become common now?
More directly to your first point, though, induced value is important even for understanding treatment effects. If value is not induced, it is possible for participants' subjective values (over how much fun a gun is, for example) to vary across treatment conditions. For examples, let's say my hypothesis is that more public information about the supply of guns will cause prices for similar guns to converge across disparate markets. But by making that information public, people decide it is less fun to own a common gun--not because it is less effective, but because it is less cool to own one. This changes the subjective value of the gun, which is not part of the theory.
On the second point (can a game with induced values be fun?), let me emphasize that inworld currency would be translated into real money, just like in just about every published experimental-econ study. Thus, the cake I described in my post would be converted into lab dollars, which could then be converted into US dollars. Having run experiments in my lab here at Cornell since 1991, I can tell you that this little detail makes games a LOT more fun. I usually pay an average of about $20/hour to my subjects, but how much they make depends on their performance. I have paid people $200 for an hour of (intelligent) trading in assets that are completely artificial. Believe me, the good traders have a great time!
This goes back to my last post on the glovebox and the glovatar. Obviously, it wouldn't be possible to have a game in which everyone is getting paid big bucks for playing. I envision a platform with virtual worlds that don't induced value, but provide reasonably entertaining and educational games that attract people interested in business. Researchers would then create glovebox instances/dungeons with induced values, and pay people to play--just as experimental economists routinely do.
David, thanks for the link to the Malaby article; I downloaded it and will read it tonight.
Prokofy, are those dollar figures you mention in US$ or Lindens?
Posted by: Robert Bloomfield | Jun 06, 2007 at 17:40
>> ...induced value might not be necessary to show directional effects of a treatment... I wonder whether things have changed at the University of Arizona since Vernon Smith left for George Mason, as what you said about induced value would have been sacriligious 10 years ago, when no one would publish any experimental economics paper without very pure induced value--which meant not only having artificial values for goods, but also running experiments devoid of any context. (Players could only be named "A" and "B," not something meaningful like "auditor," for fear of contaminating the lab environment with baggage.) I always thought that such demands were overblown as long as researchers are holding all of those factors constant and searching only for directional treatment effects. Has that viewpoint become common now?
My impression is that this has always been part of the "art" of experimental economics, and that what's acceptable has always been in flux. Even in market experiments, the whole thing could be put together in such a way that you don't have to identify who's a seller and who's a buyer, but such identification has always been the norm. In game theory experiments, bland context-free presentation has been the norm, but Schelling argued that was often a bad idea.
Field experiments are becoming more popular, which frequently use naturally occurring contexts, and sometimes homegrown valuations for things (David Reiley here at UA has done a fair amount of this with auction experiments).
Also, it seems more experimentalists have decided that presenting a "story" along with the experiment can ease subject's understanding, and that the improved understanding can be a greater benefit than the drawback of having them draw in unpredictable outside values.
I'll have to think about your other points.
However, your describing the glovatar again gives me a thought. It may be possible for the right kind of research question to run an experiment in an existing MMO without any great investment or cooperation of the game company. This could be done simply by having the experimenters run PC's in the world who can interact with the subject's PC's, the experimenter's PC's acting as the institutional environment. The experiment in question would have to be something that doesn't require as extreme isolation as the glovatar/instance. Then standard induced values could be used (or induced values with game currency).
I had thought about doing this some time ago, but dismissed the idea because I felt (as I've said here) that it would be a poor echo of a lab experiment, without showing any special effects of being in an MMO. However, maybe you could construct an experiment to your tastes in this way, as a proof-of-concept?
Posted by: Timothy Dang | Jun 06, 2007 at 18:24
"Do they also fix the value of items that require an extensive production process?"
All items have a fixed value. As the item decays the value goes down. Some items may be repaired indefinitly, others are one time use items. The fixed value of an item can be transferred directly into the game currency by basically dumping it into a trash terminal. The item goes away and its value ends up on your card. Now, the real rub is that most items do not sell for the MindArk fixed value. They sell for either a percentage of the value (eg 110%) or a + value, like Value +100. Even the materials used to make the items sell for more than thier fixed value.
"For example, I spoke briefly with a weapons manufacturer in Entropia. My understanding is that he charges whatever he can get for those guns, and that people buy them not only for their ability to serve as a factor of production (e.g., they can be used to kill creatures that provide valuable loot), but people also like them because they are fun to own and use. Their economic value is therefore detemined by players' tastes, and not controlled by MindArk."
Every item that provides a player an opportunity to profit or receive something back for his efforts, in your example a gun, has statisctics based on power, range, decay rate, fixed value and other things that produce its overall economy. The economy, as in Entropia you are dealing with real money every time you pull the trigger, is what drives the economic value of the item. And of course, this value is controlled by MindArk. They control the economy of the item, which can be vitally important to the whole game system and your own personal "return" if you will, as well as the drops of these items into the system.
Interestingly, most people I have known that play Entropia are on a never ending quest to improve thier own personal economy.
Posted by: Nate Randall | Jun 06, 2007 at 18:34
I'm out of my league here on economics issues, but here's a random thought. If helpful, great. If ignorant, please ignore...
Why not predicate value based not on in-game money, or even a link-back to RL money, but on time spent in an activity? For me (middle-aged, middle income dude), dropping (or earning) US$20 on a thing-a-ma-bob may mean much, much less than to a college kid.
But... neither of us will spend much time doing something that's not fun. And if it is fun, we will probably both want to do more of it.
If you somehow measured the amount of time folks spent in game doing various stuff, and then provided ways to block access to the funner, more popular stuff, and means to "earn" the right to do the funner stuff... would that help with the dominance issue? I mean, 20 minutes of fun in RL is the same 20 minutes as in a virtual world doing something equally fun.
By "fun" of course, I mean something you choose to do for whatever reason; could be substituted for interesting, educational, sexy, etc.
Seems to me that the time I spend reading/commenting on this blog could be spent doing other things. But I choose to do this for X amount of time... make it easier or harder for me to do so, or give me more ways to participate or intervene in others' participation, and you adjust the value of my time spent here.
Just a thought.
Posted by: Andy Havens | Jun 06, 2007 at 23:06
Robert Bloomfield>> More directly to your first point, though, induced value is important even for understanding treatment effects. If value is not induced, it is possible for participants' subjective values (over how much fun a gun is, for example) to vary across treatment conditions. For examples, let's say my hypothesis is that more public information about the supply of guns will cause prices for similar guns to converge across disparate markets. But by making that information public, people decide it is less fun to own a common gun--not because it is less effective, but because it is less cool to own one. This changes the subjective value of the gun, which is not part of the theory.
To me, the above is a good reason to do experiments in MMOs. If the coolness factor affects outcomes, that's a real effect, and one I'd like to observe. Plenty of investors open things like funky tapas bars rather than fast-food franchises, and not just for financial motivations. I suppose I see what you're saying, though. It would be different than an abstracted lab experiment.
On the issue of fun & induced values, I have also seen subjects in lab experiments having fun. But subjects having fun because they can tell they're doing well and they know they're winning money (while other subjects aren't having fun because they know they aren't doing well or winning money) is very different than what you'd find in a typical MMO environment.
Possibly the fun & induced value concerns come around to self-selection issues. There is some degree of self-selection into any experiment, and that should be a concern to any experimenter. But there isn't usually self-selection into a treatment. While subjects have the right to walk out of an experiment any time they want, it must be an extremely rare occurrence, even if they find the treatment they're in not to their taste.
Put the experiment into an MMO and the self-selection problem starts to loom large. People who enjoy a high information environment stay in the high-info environment, and those who enjoy low information environment stay there. The enjoyment may come straight from the "game" aspect or from the monetary rewards. Those who don't enjoy some environment, leave it. Those who got stuck being a buyer when they only like being a seller, leave. You might not allow the players to move from one environment to the other, but they can always quit the experiment.
It might be possible to fix that problem. If all treatments are offered simultaneously, then the subjects needn't know about them in advance. If the fixed fee for participation is high enough, then they'll be less likely to quit even if they aren't having fun (of course, they might watch TV or open up a browser window and do something else while the experiment runs its course). High fixed fees have their own problems. If the game players are in the broader MMO with playing the instance/experiments as a major motivator, then "blacklisting" those who tend tend to quit mid-experiment would also provide incentive to stay.
Unless you control adequately for self-selection, you'll be biasing your population toward those people who have fun in the given environment, and so probably biasing it towards those who value the fun too highly versus the induced values.
Posted by: Timothy Dang | Jun 07, 2007 at 05:32
I am going to second Andy's comment, once you start to discuss the structure of the experiment I am out of my league, but I think he is onto something when he suggests measuring time.
One of the huge advantages of VW environments is the log files. With good analytical tools, the kind that we see in the financial sector, it is possible to measure value terms of time, price paid, and item usage (utility?) and perhaps even desirability, based on analysis of auction house searches.
I would think that it would be possible to establish base line values for just about everything in a VW and then change some type of reward to induce dominance. Log files could be used to both analyze economic behavior and verify that dominance had actually been achieved.
The people that run Whyville do this kind of thing all the time. Their experiments are not as rigorously controlled as you might need for a peer review paper, but they often mess around with the economy. In support of Roberts point about the good trades having a great time, Whyville changes the economy to make (or keep) the VW fun for the kids playing the game.
Posted by: Tom Hunter | Jun 07, 2007 at 08:23
Robert,
The dollar figures are in Lindens. The Linden is worth $3.70 US to 1,000 Lindens, the course changes up and down every day.
I can only say that you will likely come around to doing your experiments in Second Life, because it is open-ended, yet you yourself can set your own controls on your own land.
Prokofy
Posted by: Prokofy Neva | Jun 07, 2007 at 11:27
Multiverse or the newly available OLIVE by Forterra might be options as well.
Posted by: Nate Randall | Jun 07, 2007 at 12:22
Fascinating discussion.
If you were using a VW as an education tool - to generate discussion about some concept or public policy - you would not need to be so strict about controlling for value.
Posted by: PJ | Jun 08, 2007 at 14:43
Time-use surveys as measures of value, I think, are particularly apt for measuring item value in MMO's. I think this is especially true because the rate at which in-game gold is valued is not strictly universal. Over the course of month, using logs as Tom suggested combined with intensive interviews, would probably be the best way to examine value in MMOs. Even then however there are going to be a variety of difficulites:
1) Value is likely highly correlated with one's chosen class (when dealing with games like WoW).
2) Value is likely variable based on one's skill at playing the game (which will again probably vary with one's class).
3) Players will likely mix activities: Socialization/grinnding, or socialization/playing the auction house, etc. It would be hard to analyze such activity into discrete components.
Nonetheless, time-use is probably preferable to using in-game currency as a measure of value simply because the value of the latter will be difficult to control, and it is possible that the value of the currency itself will vary from player to player (and probably from class to class).
Posted by: Isaac Knowles | Jun 08, 2007 at 16:19
Can someone direct me to game theory experiments that are content-laden?
I know market entry experiments are the 'bland, context-free' type that Timothy Dang talks about. I wonder if 'market entry' experiment is a misnomer, because lab markets used in such experiments do not have an active demand element in them. I also wonder if researchers would still find convergence to equilibrium if a demand component were introduced.
Why link rewards to a lottery or participants' performance in trivia quiz /darts? Why not link their reward to their 'sales'? Ah, there we go - the hypothetical lab market has NO product! What can they sell?!!
This is where I see great potential in MMORPGs - if you were to put your participants in an MMORPG environment, how many of them turn out to be as entrepreneurial as Random Unsung? Now, that's what I would call a market entry experiment.
Posted by: Lax Natarajan | Jun 09, 2007 at 03:22
As mentioned by one of your commentors, economic indicators are an important part of how we manage the Whyville virtual world for kids.
There are now a number of academic education research faculty publishing peer reviewed articles on the educational use of whyville (google Dr. Yasmin Kafai). Numedeon Inc, the company that runs Whyville, has encouraged this type of research.
We are equally interested in other forms of research including economic research. We already know that there is a great deal that can be learned from the economic data we have in whyville, as well as from our ability to monitor and change the economy.
There is also a very interesting connection between Whyville and entrapenaurship. See for example this recent article on another educational collaboration we are involved with:
http://www.santacruzsentinel.com/archive/2007/May/04/local/stories/06local.htm
There is a great opportunity here.
Dr. James Bower
CEO Numedeon inc.
[email protected]
Posted by: jim bower | Jun 10, 2007 at 11:55
Lax Natarajan>> Can someone direct me to game theory experiments that are content-laden? I know market entry experiments are the 'bland, context-free' type...
It's hard to answer the question about game theory experiments, simply because so many experiments are game theory experiments, whether discussed as such or not. So, for instance, "Voluntary Contribution Mechanism" (VCM--i.e. public goods) games are usually pretty clear as to roughly what they're about, even if they're not expressed in terms of something like public TV.
I did some searching this morning, figuring I could counter your characterization of market entry experiments, but now I'm convinced you're right. One problem with experimental economics papers is often the information about context is hidden in the instructions which are only "available from the author on request", so it's hard to be sure. Some market entry experiments are framed as such, but with the buyer-side of the market only being simulated. Some experiments are put only in terms of "choose X or Y".
Still, this seems like something which should first be addressed by better laboratory experiments on market entry, like having both buyer & seller sides of the market. I think interesting things could be learned about market entry in MMOs, but it also raises huge problems for analysis. The big one which jumps to mind is: How can you tell if a seller is suffering losses when you have no idea what their opportunity costs are? That kind of problem is (I think) what inspires Robert to want to use induced values.
Posted by: Timothy Dang | Jun 12, 2007 at 14:02