The Kasbah system has had two trial runs, both one day events inside the MIT Media Lab. A version was run in mid-October 1996 involving Media Lab students, staff, and faculty. This trial was in preparation for a full deployment as part of the Digital Life symposium [2]. On October 30, 1996, two hundred people (largely from technological industries) were given books, souvenirs, and play money to participate in a full-day Kasbah marketplace. Kasbah was one event among many taking place during a day-long symposium at the Media Lab. Participants were encouraged to create agents to sell the items they were given and buy items that they would like. Over the course of the day some 250 transactions were made. A large number of the participants were actively involved in the system, paying attention to the market and how their agents were doing.
This second Kasbah trial had two main purposes. One goal was simply to show off Kasbah itself, communicate the idea of an agent based marketplace. The other goal was to encourage participants to socialize, to have them use the marketplace as a way to meet each other both in sharing a common experience and in particular interactions of making trades. The ways in which the various objects in the Kasbah system succeeded or failed to accomplish these goals, as evaluated based on observation of the participants, yields interesting lessons about how to build an effective virtual marketplace.
There are numerous components that went into the Kasbah demonstration. A full list of every single object is surprisingly large; here, only the most interesting aspects of the system will be described. Following the categorization above, the foremost objects are the items of trade themselves. Each participant was given three objects randomly chosen nine possible types. The items were disposable cameras, canvas bags, boxes of chocolates, lunch pails, coffee mugs, watches, and three different books. In addition to the items themselves, participants also were given fifty ``bits,'' play money in the form of paper currency.
The trial run of Kasbah was unusual in that it took place in one building during the course of one day. The experiment took place within a physical space shared by a large number of participants. The trial was unlike the eventual plan for Kasbah of an Internet-wide deployment where participants will not interact closely and may not ever meet. Most Kasbah activity took place inside the Cube, a large room in the Media Lab shared by several projects. Some of the activity also spilled out the entrance of the Cube, into the lobby for the symposium. A central bank service was located in the lobby entrance. The bank dispensed items to participants at the beginning of the day and served as a middleman to help complete trades when the two parties could not easily meet.
An important set of objects in a computer mediated market are the objects that comprise the interface to the computer system itself. Traditional markets don't need interfaces, at least not explicitly recognized as such; the world is the interface. A virtual marketplace needs explicit interfaces from the physical world into the virtual. In order to externalize the workings of the Kasbah marketplace a large number of interface objects were deployed. These interface objects included badges for people with barcodes, computer kiosk stations with barcode readers, web pages (virtual objects, not physical), pagers, two large computer screens, a tickertape display and a virtual tickertape on the computer kiosks.
The unique component of the Kasbah system was the computer agents themselves, the programs running in the virtual marketplace to perform transactions. Their status as object is problematic. The agents were certainly not physically manifested and were not even clearly representations of objects. In folk speech about computer agents we speak of agents more like people than objects. Kasbah was designed to help personify the agents. Agents were given human sounding names like``Sherlock Holmes'' and ``George Smiley'' and their actions were described with some degree of intentionality. The issue of personification of agents is a large open research problem in software agents; the role that the interface objects played in that personification is one of the interesting lessons from the trial run. In that the Kasbah agents are replacing the traditional social aspect of negotiating a sale, their status and effect on the social system of the marketplace as a whole is important to understand.
