• Christian Langkamp


Defining Friendship Amid a Global Pandemic

By Christian Langkamp

Image credit: Tallulah Fontaine

Lunchroulette ( is a project run with the aim of digitally connecting people who are likely to become friends. It relies on a set of questions formulated based on research, which consider the relative importance of different variables that might be important to a friendship. These come from a variety of sources, such as philosophy (Aristotle and Cicero emphasising the virtuous character and character development), theology (biblical quotes, as well as explanatory texts such as Olyan, putting the focus on hospitality, kindness, patience and loyalty), modern philosophical thought (Lewis, Alberoni and Nehamas; referring to a meeting of minds, independence of people and the importance of mutual passions and interests), feminist theory of care in friendship (acceptance and emotional intimacy), as well as various self-help books on friendship (Rath, Millington or Nelson; including fun, sociability or excitement/challenge). Whilst some themes are reoccurring across the literature (honesty, loyalty, enjoyment), other themes are more particular to different philosophies, age groups or cultural groups. Capturing friendship behaviour as a whole - even after some clustering - thus required around 50 variables, which are included twice: once as gift love (‘I am good at X’), once as need love (‘I really appreciate when my friends are like/ think like/ do Y’).

The second ingredient of the project entails a detailed register of potential 2nd level interests, which provide the breeding ground of commonality. Essentially, this is trying to increase the probability of the mutual insight effect occurring - the moment where you might think ‘Oh my god, that is my favourite thing about [insert favourite pastime] too’. Indeed, it seems like this is one key part of the Oxford experience, the realisation that for each curious little hobby or political/social passion you may care about, there are others out there who share your sentiment. A second level interest could for example be ‘coastal rowing’ to the 1st level ‘rowing’, or ‘People and candid photography’ to the interest of ‘Photography’. There is a wide range of possible options (Wikipedia alone lists about 200 hobbies or interests), of which usually people pursue 2-5 intensively and another 5-15 as general interests. For this reason, the questionnaire does take some time to go through (20 – 30 min), but this allows the resulting matches to explore common ground in depth.

The inspiration for the design of the project was based on my own experience of Freshers Week in non-COVID times, which allowed us to socialise intensively with everyone in college within a short span of time, through college meals, club nights and JCR hangouts. As exhausting as this experience seemed at the time, it meant that by the end of my first weeks at uni I had already met around 100-120 people, out of which I naturally gravitated towards five friends with whom I ended up spending most of my remaining 3 years there. Indeed, part of this process relies on a numbers game - by being exposed to enough people with diverse interests, there is a sizeably positive probability for reaching that ‘Oh my god moment’ with someone. Of course, our friendships eventually went through a few transitions and additions, as people developed other interests, or as we got to extend our circle beyond college, but, retrospectively, the bonds made in the first two weeks at college were remarkably stable over time.

Unfortunately, this process was compromised to a large extent in 2020 – this is why the set-up of the LunchRoulette platform relies on delegating the initial ‘filtering’ stage of any friendship to ‘the machine’, which then supports possible matches. After a match is generated, mutual contact details are shared with a pair of possible friends, and they are encouraged to get in touch and meet virtually or in person (if possible). This algorithm is not designed to generate ‘perfect’ matches, but to facilitate the process of information gathering that in the past would have occurred within large groups, and narrow down the choices to get an estimated ‘success rate’ of around 40 % (i.e. if you meet 5 people, then you are likely to hit it off with 2 of them). And as a bit of a fun Oxford gimmick, it can also calculate your matching scores with figures such as Aristotle, Cicero, C.S. Lewis or some religious figures (based on profiles calculated from extracting information from letters, quotes, religious texts etc.).

Of course, staples of the Oxford Fresher experience such as subject drinks, fresher groups meeting in the Christ Church Meadows or rowing taster events have been incredibly helpful ways to discover “your people” for most students here. But for those confined within student housing, this virtual alternative based on various psychological, social and literary theories of friendship may be one way to use the confinement time to reach out to others outside of your “bubble” and plant the seeds of a new friendship. is a research project run by Christian Langkamp, who did Undergraduate Mathematics at Lincoln College several years ago and has since been working in the industry. He is now back as a visiting fellow to research friendship and how to maintain it into old age.

Disclaimer: The data collected by LunchRoulette will also be used for research (as stated in the GDPR statement), and there is an expiry and deletion process.


Recent Posts

See All