I'm a 1st-year Learning Technologies PhD. student in the School of Education. I did my undergraduate study at Xavier University and the University of Paris IV (Sorbonne), and earned my master's in Social Anthropology from the London School of Economics.

General interests include:

- new ways of forming relationships through distributed information and communication technologies (ICTs),
- the concept of a language community,
- semiotic approaches to the study of human development,
- and the importance considering culture in any systematic study of people.

I would like to better understand the differences between on and off-line primary school community building. With this in mind, I'm planning a collaboration with colleagues in Developmental Psychology and CPEP to study the social cognition of late primary school students in a statewide, online learning community.

---

DesignAssignment

Discogs: Design Successes

I. What is Discogs?
II. Design success 1: Charting taste
III. Design success 2: Information architecture
IV. Brief conclusion and suggestions



I. What is Discogs?


Discogs.com (or “Discogs”) is a user-built webbified database for cataloging, rating and discussing music publications. The project was started by Portland-based Kevin Lewandowski in 2000, and in five years has grown to comprise ten of thousands of artist, release and label web pages, in addition to useful features such as shopping and discussion forums. There are about 2,000 members, a small percentage of which pay twelve dollars annually for “advanced” features such as fuller search and browsing capabilities. Though originally conceived for aficionados of electronic music, the popularity of the site has spurred old and new users to add information about publications from other genres such as jazz, hip-hop and rock.

The website’s success- defined as retaining and attracting new members who build its database- has a few basic explanations. First, the site offers readers highly valuable information. Electronic music fans were the first to use the site because the genre is dominated by artists recording on short-lived and financially strapped independent labels. Given the lack of mass appeal and circulation, information about such artists, labels and releases can be very difficult to obtain. The opinions of site members, discussed below, is also extremely valuable, as the “if you like this, try this” dynamic can lead even the most seasoned experts of a particular genre to discover artists, labels and releases anew.

The information obtained by members or just browsers is valuable (in Kollock’s term it’s the intended “public good” of the website), but Discogs offers another important incentive to membership: the potential to establish openly recognized mutual interest, or trust, which provides the basis for the development of stronger social ties and exchange of important goods (though the importance of this is debated, cf. McKenna et al. 2002). Trust is a powerful motivator for browsers to join Discogs. Music inspires many people to exchange knowledge and artifacts thereof, and in online communities, the terrain of trust can be uncertain (in Sassenberg’s terms, the Discogs community mostly constitutes a “common bond” group). The architecture of the site allows for an efficient and well-integrated flow of opinion, data, and discussion, which facilitates relationship-building, though such may still be considered relatively “weak ties” (Constant et al. 1996). Overall, good information architecture adds social value to the Discogs website: it offers the opportunities for fans to share their passion for music and to establish trust.
PaulResnick: I would think that the primary attraction would be based on common identity (collector of rare music) and that bonds form only later. (AndrewBabson: Yes I meant his "identity" term). So this would be a "common identity" group. Also, the incentive of "openly recognized mutual interest", basically common identity, is quite distinct from the incentive of building "trust", which enables social ties and exchange of goods. AndrewBabson: Just to clarify, by mutual "interest" I mean mutual "advantage".


The following design features support the above key assets of the website.


II. Design success 1: Charting taste


Labels are brands. A brand can embody generalizable attributes that differentiate one product from another. As such, they aid consumers by eliminating guesswork and narrowing the field of buying options. If one knows artist A has released a record on label B, then one might like to know about other artists and releases on label B. This can be difficult with independent music, as many releases may have to be special-ordered, or are out of print altogether. An opinion system, then, is almost invaluable: before one goes through the effort of hunting for a release, one can browse to see what other members think (Miller et al., forthcoming). Because Discogs attracts highly committed and knowledgeable fans of electronic (and now other types of) music, this opinion can be powerful.

As an example, let’s say you notice that DJ Record released EP #5 on Label X in 1995 under a pseudonym “Record Man.” Going to the page for “EP #5”, you notice that the score, on a 1-5 scale, is 4.8, based on votes by 25 members- a succinct and useful marker of overall quality from minimal user investment (Dellarocas 2004). You also notice that in the “comment” section, three users have contributed descriptions of the tracks on the record, as well as information on the label. The artist name, label name, and member names of the contributor and raters are all linked. For example, if I want to know how the 25 members rated the record, I can go to a page with all of their names, and then visit one of their “member pages,” to find out more about them and their musical taste. More about the member pages below; but I would like to emphasize that the ratings and comments of members are voluntary and have only the slight moral reward of adding points to members’ rankings (cf. Wasko et al., 2000). As such, the opinions tend to be well-defined and nuanced- very valuable information in the highly fragmented domain of independent, “underground” music publications, where obscure labels, pseudonyms and unnamed tracks are the norm.

All members have their own profile pages. What is interesting here is that, while personal information is listed on the left, all of the comments each member has logged are listed on the right side, under links to the respective release pages. What adds another dimension to the user’s profile is the twin feature of “collection/wantlist.” On each release page, there is a simple set of “Xs” on the right hand side, denoting “In my collection” and “In my wantlist.” When a member first joins Discogs, a fun activity is browsing through all of the releases and adding them either to one’s collection or wantlist. What the Discogs database then does is compile these lists, make web pages for each list, and, importantly, make each list downloadable as an Excel .csv file. For many users, this is collectively the killer app that makes Discogs worth using- all free of charge.

Each profile presents a snapshot of the user’s unique musical taste, and facilitates usage of other Discogs features. Email functionality on the page allows communication, while the link to discussion boards allow for group discussion of music, on topic such as track identification from recorded DJ sets, upcoming events, and trading/selling. For the latter, Discogs has recently opened a shopping section, whereby members can hawk their wares: equipment, records, etc. By allowing for users to show what they like, what have and what they would like to have in their collections, it opens the door for establishing trust with other members and thereby functional and/or personal relationships.

PaulResnick: In this section, you have not described any design alternatives and then made claims about why the social science findings suggest that one alternative would be better than another, or even why having a particular feature is better than not having it. For example, you might argue that the collection/wantlist tool turns the creation of the group database from a public goods production problem into a private goods production problem: peeople create their lists to have for their own consumption, but everyone else can get some benefit from them. This is the same story behind del.ico.us.


III. Design success 2: Information architecture


Discogs would not work so well as a tool for information and communication exchange if it were not well organized. Members can efficiently find what they are looking for musically, find out what others think about a release/artist/label, classify how that release fits into their collection and then refer back to that information while communicating with other members. In sum, the most important information is linked, and this cross-referencing capacity is a major design success of Discogs.

It is crucial in this process that each release added to the database is vetted by the dedicated albeit shadowy Discogs volunteer moderators and editors. Members receive points for contributing information about a release that is accepted into the database. The release then has its own page, with links to the artist and label pages. There is also the ability to add an image file of the release, which allows other browsers to see all types of labels for the release, which is particularly interesting for collectors of older music.

As mentioned above, it is also important that members can compile their collection and want lists by simply clicking the respective “X”s on a specific release’s page. Each release is then added to a web page listing all releases that the member owns, and another listing all the records a member would like to own. This saves the member hours of cataloging time and resolves the crucial data hygiene issue in one feature. In addition, if a release is not on Discogs, one can add it- providing it’s done correctly of course (there is a new “help” section that guides members on how to add releases).

Finally, members can click a link to generate an Excel spreadsheet of both collection and wantlist- complete with the most important fields such as title, artist, label, Discogs release number and year- for personal storage, printing, etc. If the information weren’t well organized and vetted for cleanliness, this service would immediately lose its value, and be another case of “bad data in, bad data out”- no one would derive value from storing bad or poorly organized data.

Taking a step back, one notices that there are multiple fields of information for indexing a wide range of values, which provide, for the first time, a full picture of a highly fragmented set of information. Multiple channels and levels of access to this information provide the browser and to a greater extent the member with highly valuable information and efficient means by which to use and exchange it with others. Because it is a database, Discogs’ public, webbified information is also able to be linked and referenced outside of the community, for example in emails or links on other sites. In this way, Discogs’ information is not just for members as co-builders; it’s a truly open and public source of valuable information for music fans worldwide.

PaulResnick: Is there anything about the information architecture that is good because it embodies an understand of some social science finding that we discussed in class or that you know of from elsewhere? Good information architecture seems like a good idea, on general usability principles, but isn't informed in any way by what we covered in this class.

IV. Brief conclusion and suggestions


In conclusion, Discogs features two design successes: presentation of member taste, and information architecture. There is some room for improvement, however. First, there are no music samples available on the site, which is inconvenient. In addition, the basic, non-paying search function is not very good. And finally, if one makes a typo or would like to edit a comment, it cannot be edited. Given that this short paper concerns design successes, however, these topics will be left open for further discussion. Discogs is a user-built goldmine of highly usable and valuable information whose assets will only grow with more contributions.


References

Dellarocas, C., Sanctioning Reputation Mechanisms in Online Trading Environments with Moral Hazard (July 2004). MIT Sloan Working Paper No. 4297-03.

Constant, D., Kiesler, S., and Sproull, L. (1996). The Kindness of Strangers: The usefulness of electronic weak ties for technical advice. Organization Science, Vol. 7, No. 2 (March – April, 1996), 119-135.

Kollock, Peter, The Economies of Online Cooperation: Gifts and Public Goods in Cyberspace, in Communities in Cyberspace.

McKenna, K. Y. A., Green, A. S., & Gleason, M. E. J. (2002). Relationship formation on the Internet: What's the big attraction? Journal of Social Issues, 58(1), 9-31.

Miller, N., P. Resnick, R. Zeckhauser (forthcoming) Eliciting Informative Feedback: The Peer-Prediction Method. Management Science.

Sassenberg, K. (2002). Common bond and common identity groups on the Internet: Attachment and normative behavior in on-topic and off-topic chats. Group Dynamics, 6(1), 27-37.

Wasko, M. M. and Faraj, S. (2000). "It is What One Does": why people participate and help others in electronic communities of practice. Journal of Strategic Information Systems, 9, 155-173.

Comments

LaurieBuis: All semester, I've thought that this was an interesting community... I do have a question though for you. You state in the beginning of the paper that "An opinion system, then, is almost invaluable: before one goes through the effort of hunting for a release, one can browse to see what other members think (Miller et al., forthcoming). Because Discogs attracts highly committed and knowledgeable fans of electronic (and now other types of) music, this opinion can be powerful." I keep thinking about self selection and how this impacts the user ratings of different music releases. It seems to me that through self selection that the ratings of music releases would be highly biased. The people who review and comment on a release will be people who are presumably interested in that type of music or particular artist and thus might rate the release more favorably. What kind of impact do you think this has on the overall usefulness of the member ratings for releases? Is there any way to control for this?
PaulResnick: Could you use the user ratings to help create common identities, or help people create bonds?