DJALCHEMI

Recommendations for enhancing iTunes' sharing features

Since my posting on research into iTunes music sharing, I've got a copy of the full paper and found time to read it on a recent train journey. Last time I focused on how people manage the impressions that others get from their music collections, but the research also has interesting things to say about unanticipated uses of iTunes sharing, and implications for enhancing the sharing features.

The paper — by Amy Voida and four co-writers — points out how iTunes differs from the large-scale peer-to-peer applications (like the original Napster, KaZaA etc), which tended to anonymise music sharing. With the latter, a user downloading a track will typically have no interaction with the person who made it available for sharing. With iTunes, sharing is restricted to people on the same subnet, which often means the users know each other personally off-line. In the 175-employee organisation where the research took place, there were four different subnets, three of which were restricted to single floors in a building. This significantly alters the nature and dynamics of the sharing. For example, where the big peer-to-peer applications require thousands of users before they reach critical mass or tipping point, this research suggests that iTunes sharing can be viable and valued with just two users — in circumstances where they also share experiences and understandings in other parts of their lives.

All the research findings reflect different ways in which technical, musical and organisational factors (or, as the researchers call them, "topologies") are overlaid and interact with each other.

Unexpected uses

Uses of new communications technologies frequently evolve in ways that their designers did not anticipate. The enormous volume of SMS usage is one significant example. There have been personal and anecdotal accounts of interactions in online communities, or social networks as everyone calls them these days (John Seabrook's Deeper, Howard Rheingold's Virtual Community, Stacy Horn's Cyberville). But Voida et al's paper fits better into the genre of research that shows how organisational relationships get played out through the medium of network technology (Wanda Orlikowski's study, showing how competitive corporate culture can undermine the collaboration that a groupware system is supposed to foster, is a classic of the genre).

Some of the unexpected social and organisational by-products of iTunes sharing in the organisation in the study include:

• instant-messenger-type functions, in that the visibility of others' music libraries or playlists indicates whether they are still at work, their PC has just crashed, and so on;
• increased social awareness, not just in terms of presence, but in terms of appreciating facets of co-workers (like their disposition towards sharing, as well as their musical taste) that were previously unknown;
• frustrations about 'anonymous' sharing and uncertain identities, since iTunes can be configured to show your library under an alias (e.g. 'Prancing Horse's music' instead of 'David Jennings' music');
• and conversely concerns about privacy, which are reported to have increased when when the managers in the organisation started sharing their collections and browsing others' collections (I imagine worries like "Will my Butthole Surfers collection affect my promotion prospects?").

Design recommendations

The researchers review the design implications of several of their findings. This list is a distillation of some of the recommendations that stem from the research.

• iTunes should support multiple users of the same computer, to cater for instances where different members of a family (for example) have tracks on the same iTunes library. This would avoid sharers getting the 'wrong impression' of a users musical personality and preferences (a similar problem occurs with Amazon if you buy cookery books as gifts for friends and family while you have no interest in cookery: you will get recommendations for further cookery books).
• iTunes should provide alerts when users add new music to their libraries. The research identified cases where one user would browse another's library, and, if they found nothing to interest them at first, they would never go back, even though significant music might later be added to the library.
• Users should have visibility of how many people have listened to their music. There could be privacy issues with identifying exactly which users have listened, though there are precedents in cases like Bloglines, where subscribers can choose whether or not to be anonymous.
• iTunes should help manage shutdowns, and/or prevent them being so sudden, since the research found cases where listeners to others' music found their stream terminating with no warning, while those on the other end felt guilty about depriving their listeners.
• Where music libraries are withdrawn from a network, some 'trace' should be left behind, so that users have at least some 'grace period' to find the same music from other sources, rather than being left without even the track names.
• iTunes should be made a system-owned service (like FTP and Web services), so it does not depend on people having their PCs and iTunes applications running for their music libraries to be accessible.

The conclusion of the paper suggests that recommendations such as the last of these represent a possible coming together of technological innovation and legal/ethical considerations. The researchers believe their works shows that users' motivations for using potentially illegal services like getTunes (allowing download rather than streaming of others' music) may be "entirely reasonable". Whether RIAA and BPI members would agree with this is doubtful.

Social interaction on- and off-line

The most telling conclusion of the research may be that, "If it were not for musically-related social interaction outside of iTunes, these participants would not have discovered new music inside of iTunes". This may be an important corrective to those (and I admit that at times I'm one) who imagine scenarios where recommendation and word-of-mouth services like Last.FM and Webjay operate almost exclusively online, without reference to interactions off-line and in other media.

The new music that people listen to is not determined solely by their previous musical preferences, but also by apparently tangential and non-musical factors.

From the full paper details, you can also see the set of references cited in the paper, which include several other interesting articles that are available online.