« VWs, Data Mining, Advertising and Privacy | Main | Themis Group Projects Rapid Growth in Online Game Revenues; Teams with IGE »

Jan 06, 2004



As we are having a promotional evening:

Can I add then that I spent the part of the holidays that I was not in SW:G or ATITD adding more information than you can shake a stick at to my web site.

It now includes a massively expanded games studies bibliography (this is not complete yet but I am adding a reference to every paper presented at every games studies conference so far and conferences with a games studies track):www.ren-reynolds.com/bibliography.htm

There is also and an expanded links section – now including a section on artists that work with games which has links to over 50 artists: www.ren-reynolds.com/links.htm



This is a very interesting paper; however, I have a few questions and disagreements to put forth.

A number of points are made to support the idea that the game market will not support monopoly or oligarchy, including these:

“One of the major attractions of life mediated by avatars is the anonymity it affords, and anonymity requires a person to have exit options: other worlds to escape to...”

Most games allow multiple characters per account, and every game I’ve heard of allows players to ‘start over’ at will. Doesn’t this provide adequate anonymity?

“...content will almost always be subject to some kind of congestion effect.”

Doesn’t the practice of running distinct instances of the game world on multiple servers solve this problem? As implemented today, this creates another problem by limiting social contact across server boundaries, but I suspect there are ways around that.

“...there is a marginal cost to creating and maintaining game content.[9] Moreover, there are no economies of scale on the supply side to match the increasing returns on the demand side (Liebowitz and Margolis, 1994).”

I don’t understand the intent here. Don’t all game companies face the same marginal cost of content development? Or is this meant to say (as the footnote seems to imply) that customer service costs rise as the amount of content increases? I don’t have ready access to the Liebowitz and Margolis paper, so perhaps that’s what I’m missing. More generally, doesn’t the non-rivalrous nature of content lead to positive returns to scale?

Tangentially, when I was thinking about market structure, I concluded that monopoly might actually benefit players. Consider the simplistic notion that the game industry is essentially a market for game content. If a monopoly controls this market, they will obviously exploit monopoly pricing. However, because content is non-rivalrous, each player will theoretically benefit from the entire industry’s content output. Now, assume that a second company enters the market. Price competition results, but does player welfare improve? The total supply of content has not changed, it has merely been split between two games. A single player can now access only *half* the industry’s output, since (if present games are any indication) they will have time to play only one of these two. From the player’s standpoint, the content supply has been effectively halved. Price competition will drive game prices down, but the ‘drop’ in content supply will push them up. The total quantity of content production will increase, but the proportion of industry content that one player can consume will drop. The welfare effect on players is thus inconclusive. These effects are even stronger when comparing duopoly with larger markets.

Anyway, I agree that the diversity of player tastes can likely not be met by a single game; I suspect that ultimately, however, a very small number of games will predominate.

I had some more important points to make about the nature of game goods and player wants, but I’m tired now. Perhaps later.

P.S. Is the bibliography correct? I recall references to Eco and Borges in previous works, but I don’t seen any here.


Kelly > ...is this meant to say (as the footnote seems to imply) that customer service costs rise as the amount of content increases? More generally, doesn’t the non-rivalrous nature of content lead to positive returns to scale?

Yes, and, no.
Online game management and servicing is labor-intensive. It's not like OS software, where you make one copy and effortlessly produce enough for millions of users. Providing "online game" to a new user becomes more costly as the number of users rises. Jessica Mulligan is way more vocal about this than anyone: you're not a software developer, you're a service provider. That's a different industry structure, completely.

Second, online game content is rivalrous like a freeway: up to some level of use, yeah, there's no rivalry. But people tend to use these things past that point, a lot. That's why there are constant complaints about kill-stealing, and why EverQuest guilds (SORRY to cite EverQuest yet again, Richard) develop community calendars of when they can take down which dragon.

There's a wrinkle, of course, in that the game is actually negatively-rivalrous in one sense: it's more boring if there aren't enough people. A freeway isn't like that, of course. It's best when it is empty. But game worlds are like parties or clubs: adding people increases the benefit for awhile, but eventually there are too many people and the benefit of a new member starts to go negative. Meanwhile, the cost of servicing a new member starts out positive and just rises. Somewhere in there, the marginal benefit of a new member equals the marginal cost. At that point, the gameworld reaches its optimal size. (All this is from old Buchanan/Tullock theory of club goods, 1960s-era econ theory.)

The main point about market structure is: there's an optimal size. And it's less than 6 billion. Online games are not natural monopolies, they aren't OS's or IM systems. They have some network effects on the demand side, but those are eventually mitigated by congestion effects and supply-side costs. So that's basically my argument for why EverQuest and its ilk will not eat the world.

Kelly> P.S. Is the bibliography correct? I recall references to Eco and Borges in previous works, but I don’t seen any here.

Yeah, this is when I started trying to be more like a normal economist, putting in fewer references outside the discipline. Bah. Never again.


NB: IANAE, but I did stay in a Holiday Inn Express a few weeks ago ....

Ted: "In an avatar economy, however, the government can effortlessly peg many prices at any value. Since the goods are digital, they can be costlessly created and destroyed. Hence price ceilings create no excess demand, and price floors no excess supply. It may make sense to control some prices.

* * *

In an avatar economy, however, increases in per-capita wealth – which make it easier to accomplish various quests and missions – will lower the challenge level of the game, potentially making it a less interesting puzzle." (emphasis added).

Can those two statements be reconciled? In the second paragraph, Ted recognizes a certain "cost" associated with the creation of wealth (in game assets, I assume). But is it a cost in the economic sense? Let's consider it to the extent of my limited knowledge ...

Keep in mind that developers are not in the business of selling items--they are in the business of selling "fun" and items are an input to the final product.

For the moment, let's ignore the Ted's second paragraph and assume that "fun" is proportional to number of in-game items (i.e. that a developer is indeed in the business of selling items). Is the statement that items can be "costlessly created" economically true? Marginal cost might be zero (I suppose you could argue that that is "costlessly created"). So might variable cost. But average cost isn't--it's the total cost of the developer's operation divided by the number of items.

Remember it's the cost of producing "fun," not just items.

And this makes sense: the average cost curve continues its downward slope because there are no variable costs to eventually dominate the fixed costs and cause the average cost curve to bottom out and begin to rise again.

That the average cost curve is ever-declining suggests economies of scale and that the developer is a natural monopoly. But that doesn't feel right. Not at all.

So ... there's a problem. Ted's second paragraph states the problem: too many items diminsih the value of the game. But is that an economic cost?

I think it is. Consider items as inputs to the final product: fun. I think it intuitive that each input will exhibit diminishing returns. Indeed, returns for a given might begin to (probably do) decline after levelling off. There will be a cost involved: the amount it would cost to maintain the level of fun. Seems to me that would implicate both variable and marginal costs associated with creating items.

Elsewhere in the article, Ted recognizes that these costs can be significant.

And this feels right: introduction of easily-acquired, powerful items can effectively wipe out considerable amounts of "fun."

Jeff Cole


First, congratulations to Dr. Castronova! It's always exciting when someone is able to get published and bring even more attention to the economies of virtual worlds.

That said, far be it for me to pass up an opportunity to play armchair economist.

"the government can effortlessly peg many prices at any value."
While I agree that there are cases where virtual world governments, ie the developers, are able to 'peg' prices to in-world currencies, I have yet to see one that has been able to 'peg' any in-world content to any outside world currency for more than a few short months.

"assumes that work causes disutility"
There are places in real world economies were 'work' is not viewed as 'disutility'. Clearly in an industry like Academics people pay to be given work and even tested. Even the GMAT costs some $225 these days, something that is very labor intensive but seems to have quite the value, and follows the basic 'pay to work' model. There may be other examples, like the best golf course isn't always the easiest one. In fact, few golfers think that golf is about getting the best score, but rather about how you feel you did vs the other players (PvP) in the game or vs the difficulty of the course(PvE). It would be easy enough to design a course where everyone eagled on every hole, but I doubt you would get top dollar from the players looking for a place to challenge their skills.

"Growth can be bad"
Again going back to academics, although we have super computers and spell check, there are probably few math or English teachers that would think that their students would learn more if they were given access to these tools on test day. Even in a market like Las Vegas, for good reason they discourage the utility of silicon technology within the casinos, and much of the time under the same premise that the use of this technology would make the activities there 'less fun'.

"takes the population of humans as fixed"
Since the day that Adam and Eve left the garden, populations have been in flux. However, more to the point of the argument, I think most virtual worlds are closer to the economic principles of cities than countries. The macro economic numbers of a city include not only the activity of people that never leave the city, but also people that live in the city and work elsewhere and people that live elsewhere and work there. On the far end they also include the occasional truck driver that stops for a cup of coffee. Again, this pattern is very close to the economic activity patterns of virtual worlds with members appearing and disappearing now and then.

"assumes that their tastes and initial abilities are fixed"
So, i might disagree with this statement on two accounts. First, my marketing background tells me that the Earth economies don't always assume that people's tastes are 'fixed.' In fact, most of new business development in any economy is based on the assumption that tastes change. Second, if anything, virtual worlds have a government more likely to assume that 'initial abilities are fixed'. They go so far as color coding your abilities for you with things you are likely to be able to kill in one color and things you are unlikely to kill in another color.

"well as how many different people to be"
Again, I think there are analogies in the real world that draw similarities. I'm not sure how different a single Mom with two kids that works as an executive in a Fortune 100 company would be that much different from someone in EverQuest that has the ability to switch from one community role to another. Healer, Fighter, Crafter, and Merchant could describe the available skill sets either person.

All in all, I think there are probably more parallels that can be drawn between VW and RL economies than dissimilarities, and it has always been these parallels that I have enjoyed in Dr. Castronova's work.



Jeff> Ted> Since the goods are digital, they can be costlessly created and destroyed.
Jeff> Ted> In an avatar economy, however, increases in per-capita wealth – which make it easier to accomplish various quests and missions – will lower the challenge level of the game.
Jeff> Can these two statements be reconciled?
Jeff> Keep in mind that developers are not in the business of selling items--they are in the business of selling "fun" and items are an input to the final product.

Yes, that's the key distinction. The items are intermediate goods in a sense. Fun - user well-being - is what the VW is supposed to produce. And fun is the thing that cannot be costlessly created. Creating a meaningful mix of rewards and challenges is the craft of game design, and it's a hard thing to do, it can't be automated, it needs a human touch. However, that having been said, creating ITEMS is, actually, costless. How much coding time would it take to have a bottle spawn at the same point in a VW every 15 seconds? Essentially zero. How many resources are required to erase an entry in the user database? Zero.

My point here was to contrast this feature of VW items to those on Earth, where no item is trivial to produce and none is trivial to destroy either. Price controls fail on Earth only because oversupply or overdemand always create market pressures that work against the legal price. My point is that in a VW, you can costlessly eliminate oversupply and overdemand for ITEMs. Not for fun in general. I am not arguing that fun can be costlessly created.

I think some of the confusion comes from misunderstanding the target audience for these statements I amde in the paper. My target was not the game design community. It was academic economists. More on that below...


This statement was of interest for a couple of reasons:

"Moreover, there are no economies of scale on the supply side to match the increasing returns on the demand side (Liebowitz and Margolis, 1994). Production of game content and its maintenance are both labour-intensive activities."

Firstly being able to put the same code onto multiple servers looks like an economy of scale to me. Its true that within one virtual world this economy of scale is impossible to realize but most of the existing virtual worlds are on multiple servers. This may not be ideal but its what is out there now and definilty helps the profitablity of companies with many subscribers by giving them economies of scale.

Second the marginal cost to maintaining (and expanding) game content without increasing returns is debateable. Game companies do add content with expansions and they do get a one time payment. Subscription pricing may also make it appear "that that there is a marginal cost to creating and maintaining game content." without adiquate compensation to the game company but that may be more a function of the pricing scheme than the actual underlying economics of adding and maintaining content.

I would argue that there is a lot value in virtual worlds that has yet to be realized by the companies that build them and that as they discover this value some of the assertions that Liebowitz and Margolis seem to be making ( have not read the paper) may be challenged.


Bruce> That said, far be it for me to pass up an opportunity to play armchair economist.

No Bruce, you're playing armchair sensible social theorist, and you're doing a good job of it. Every single thing you wrote is a pretty level-headed denial of the little sub-headings you picked out. And each one of those, it turns out, is treated as High Truth in the twisted world I was trained in. I think you might be surprised to hear this, but in arguing against these things, you are being a fairly rabid anti-economist.

Just to pick one example: I know of no theoretical or empirical work in the field of Labor Economics - not one - which treats work as an activity that can raise human well-being. In technical terms, economists always and everywhere assume that work imposes disutility. That axiom is held as a universal truth in my discipline and imposed as doctrine on every graduate student. (Why? Because without it, mathematical models of work behavior would be impossibly complex.)

It also makes no sense at all, as you rightly point out.

So, to re-iterate what I said above: all of these things - price-fixing, the joy of labor, fixed human poplations/identities/tastes, and so on - are part of the byzantine architecture of contemporary economic theory. The ediface has long since lost its appeal, in my view, so its many odd butresses ("labor hours has to have a negative sign in the utility function, otherwise you can't derive a unique optimum") are worthy targets for the wrecking ball.

Operating a wrecking ball is itself costly, of course. Why bother to change economics when just abandoning it (or just conforming to the rules of the mandarins who run the place) is so much easier? Well, VWs give me, personally, a reason to re-model. Many of the more crack-pot ideas in modern econ only appear as such when viewed through the lens of virtual world studies. Thank you, Mr. Boston, for highlighting some of the more inane ones. And if anyone with an economics background wants to interface with people in the design community (as I hope to do), they had better get these ideas out of their heads. Further: if an economist wants to contribute to the construction of a virtual world, he'd better find (or found) a new economics. And while I can't do that myself, those last paragraphs in the paper were trying to point in that direction: Here are some things that just don't fit, as is made obvious by a cursory examination of synthetic worlds. That means, these are problems for economic theory to solve. And I really do hope somebody solves them, soon.



Why distinguish between item-as-item and item-as-fun?

OK, so the creation of item-as-item doesn't add to the marginal cost of creating the item-as-fun. But why is that important?

Certainly there's an economic marginal cost associated with creating each item-as-fun. Isn't there?

Jeff Cole


Let me jump in way over my head (I've heard of economics). Logically, I'm seeing what appears to be level-distinction confusion in some discussion.

VW economics are being discussed on 3 levels which are interconnected but nonetheless seperate.

Level 1: Real World, "on earth" - RL money is made from subscriptions and retail sale of games/expansions.

Level 2: In Game - Game currency is made by participating in the game in various strategic ways.

Reasonable success in 1 along with the oft-maligned disutility of work in 2 will create:

Level 3: Crossover - Julian Dibbel and IGE (sometimes) make real world money selling virtual items.

Supply, demand, consumers, "government" all mean something else depending on which level you're referring to. Statements like "creating items is free" can get confusing when taken out of their level context. Keeping that in mind and ensuring that statements you are debating are level-dependent will help out a lot.

That being said, I (quite ironically) dislike this rigid distinction. I think a more rewarding view is that there's nothing "virtual" about in game economics. By identifying exchange rates, GNP, etc, Professor Castronova inspired (to me, I don't want to tie the good professor to anything radical nonconsensually) the line of thought that EQ has a real economy, and EBay does not create but merely converts value into our more familiar currency. It's a foreign economy that operates in a somewhat unique way, especially in terms of population and internal governance.

Random thoughts/illustrations: If I found 100 euros I would convert it to $US; but it would have always had value simply based on the fact that others want it. If I knew a VW item would stop being created, I could spend $US to buy 20 of them, wait one year, and cash out; this investing in a foreign market is as real as anything done on Wall Street.


Dave Myers entry on Ted's article:


The comments to this entry are closed.