Valve recently hired an economist to wrestle with the large data set of electronic commodity exchange humming along on their servers. The second article on the blog, on the establishment of numéraire (reclaimed, keys, Bill’s, Buds) and arbitrage gave me a fairly sickening feeling. The exchange rate is well documented. The establishment of these two hats as currency relies on the fact that there is a finite number – Bill’s hats only being given out for Left 4 Dead 2 preorders through Gamestop, and earbuds being rewarded during activation of Team Fortress 2 on a mac computer during a set period of time. http://tf2finance.com/supply/ gives an even more in-depth survey of the state of the currency. There’s even a black/grey market.
The Arbitrage article tries to characterize how close the Valve economy is to equilibrium and notes how potential arbitrage windows open and close. Considering that new items are generated at frequent intervals, it becomes increasingly clear that the economy is very simply being driven away from equilibrium at a fixed interval, with predictable depreciation. From my personal forays into trading, both TF2 items and games. It quickly becomes apparent that each item has an identification number that tracks them through trades, exit and entry into and out of the native economy, they’re “lost” upon exit to the grey market but the tracking number and subsequent trade outside the walled garden, allows the establishment of a “true” market value even after things blink out of Valves near-all seeing eye.
It’s cool that they’re trying to glean some sort of economic insight from this, as it is in some form or another actual money, but people do not act rationally and this system doesn’t necessarily offer a true simulation. In my mind, this is similar to the World of Warcraft Corrupted Blood Incident, and the intentional echo preceding the Wrath of the Lich King expansion. While economics gives the drive, the push and pull of the engine, I think ecological or epidemiological models would be better models. That being said, how could these systems be analyzed and investigated? First off, Steam has a built-in API which allows the automated access of source data, which could provide sample backpacks. Second, the TF2 item drop rate is well documented, both rate of individual events, as well as the percentage of those events. The sheer amount of statistics is quite significant (but I might be sending Valve an email asking for some more realistic numbers depending on how this goes. A one time census and sampling of “open” backpacks was conducted, which gives a decent snapshot of the economy.
Using all this data, an initial population could be established with items being realistically dispersed, the influx of additional items could be characterized, trading behaviors and arbitrage could be simulated then compared to actual trade behaviors, and exceptional cases could be implemented (level set collectors, scammers, Paypal, trading for games, unusual hats, etc.)
So now that a methodology has been defined, one of the questions is what is their motivation. Steam Workshop’s taken hold (8 games), Steam Trading is in place for two games at the moment (DotA2 is in beta). Based on Valve’s history – Shifting to F2P, incorporating community content through steam workshop, next creating the inventory metagame. Steps could be taken to prevent predatory tactics, improvements could be made to the monetization of the F2P model – increasing the number of people who make 1 purchase, increasing the frequency at which purchases are made, and moving some of the currency from the community to the company, while improving the general happiness of the participants.