Concepting
Relationships between people are often intricate and different in depth (Figure 1). Knowledge about how people are socially connected is a valuable asset in social networking in professional and personal life. Having common acquaintances may be one of the oldest and most common ways of getting to know new people for personal, business or other purposes - and the best way to start a conversation with strangers. But how do you identify those common acquaintances from the start?

One valuable source of information could be the list of contacts people store on their mobile phones, including phone numbers and other information. Traditionally this data is used easify calling, messaging and other forms of communication, but it could also be said to mirror the social network of any given individual, albeit in a crude fashion. If people could easily compare each others' lists of contacts, common acquaintances and interests could be discovered, which otherwise would go un-noticed.

This is the basic idea behind many pieces of social software, e.g. Linkedin.com, but due to their PC-centricity, these tools do not support social networking when best needed, namely in ad-hoc conversations out of office/home.

The core feature of Scent application was to allow two mobile phone users to automatically compare the phone numbers stored in their lists of contacts, and to see a list of the stored contacts they had in common. Via short range radio, Bluetooth, such comparison could be done within a range of up to 10m and without having to connect or align the mobile phones. By using existing list of contacts already stored on the phone, setting up the application and start using it would require little effort on users part.

Prototyping: Scent features
Scent application was implemented on Series 60 platform, running on one commercially available phone model in spring of 2003 (Nokia 7650).

Create identity expression. Before start using Scent, users had to define a name and image for themselves, enabling others to identify them in the Sniffing process (see below).

Sniffing. Before comparing phone numbers, the user had to identify the correct interaction partner, by scanning the environment, so called 'sniffing' (figure 2). Via Bluetooth, the application found other Scent users nearby (if any), fetched their identity expressions and listed them. Also non-Scent users were listed.

Figure 2. Sniffing the surroundings for other Scent users.

Request to exchange Sniff cards. Once a user was identified (Patrick in Figure 2), a request could be sent to compare contact details. Figure 3 shows how Patrick responds to the request and how the comparison results are shown on the devices of both users. The comparison was based on phone numbers entered into the phonebook

Note that only matches were shown (not the whole phonebook) and that the names/labels displayed originates from one's own phone. Both of these features ensured that users only show as much as they get to see from the other party, and that personal labels from the phone book were not revealed to the other party (e.g. "Anna Colin (cute)").

The comparison results with any given user was called Sniff Card since it was possible to save those comparisons in a collection, similar to business cards.

Figure 3. Requesting to exchange Sniff card with another user and displaying results.

Guestbook. Once sniff card had been exchanged, user could also post entries to the Guestbook of the other party (Figure 4).

Figure 4. Guestbook.

Public Sniff. If users really wanted to open up for contact, they could allow any user to compare phonebooks without mutual consent, so called Public Sniff (Figure 5).

Hide names. If a contact stored in phonebook was private or sensitive, user could choose to hide those names before engaging in a sniff card exchange (Figure 5). In this way user could protect themselves from unpleasant surprises when engaging with another user.

Figure 5. Public/personal sniff and hiding names.

Viral distribution. A user could easily send the application to another user phone-to-phone via Bluetooth, so called viral distribution. In this way a Scent user could easily introduce the application to another user and exchange Sniff cards immediately on the spot. In social application of this kind, the saturation of the software in the environment is crucial for user adoption ('If nobody else uses Scent, why should I use it?').

Scent user study
Scent prototype was provided to Nokia employees for user trial. The software was made available and users with compatible phones were invited to download and try it out. The prototype was equipped with a logging functionality that enabled us to collect very detailed information about what features were used and how much.

539 users installed Scent on their phones. The trial period lasted approximately 8 weeks. The use span, i.e. the time over which each individual was engaged in using Scent, was 16 days. The SMS use logs indicated 39% received the application from another person's phone.

Scent facilitated short-distance social exchange through the mobile terminal. Scent was by many seen as a catalyst of social interaction. Scent-based social interaction manifested itself in several ways, ranging from passive lurker-like action to rich social interaction making it suitable for users with a variety of social preferences. Aspects related to identity expression were also prominent, with 45% of the population updating it at least once.
Scanning for co-present users was a process associated with curiosity and being approached by other Scent users was well memorized an event. The phonebook matches created a feeling of surprise when an unexpected common acquaintance was identified. Some users raised privacy and security concerns. These were alleviated, however, by the fun and non-serious feel to using the application. It was also suggested that having something in common with the Scent partner would increase trust and dispose of privacy concerns.

Although the study was associated with limitations, such as relatively low penetration of the application, technical problems, and inability to use Scent out-of-office, it could nevertheless be concluded that Scent was well received and enriched face-to-face interactions.