Discovering people

David P. Anderson
16 October 2023

Some internet apps provide tools for people to efficiently discover other people, for various purposes. Examples include OKCupid, LinkedIn, Grindr, and so on.

In the world of classical music there are many reasons to want to discover people:

  • A pianist wants to find (local) string players to play chamber music with.
  • A performer wants to find a composer to write a piece for him/her.
  • A composer wants to find a performer to play a piece they wrote.
  • A composer wants to find someone who knows about score-editing software.
  • A performer wants to find someone who knows about recording techniques.

As with other search types, individual taste comes into play. The pianist wants to find string players who have similar (or at least compatible) taste, and are able to play music of a similar difficulty level. Thus, the notions of efficiency and completeness apply to people-discovery mechanisms.

Before the Internet, people discovery was through in-person interactions (parties, meetings, work, practice room hallways) and word-of-mouth. These are incomplete (they can't connect me to that Kazakh composer) and only somewhat efficient.

There are a few existing Internet-based mechanisms for musical people discovery:

  • Meetup lets you (for a price) publicize in-person events of any sort; e.g. going to concerts or amateur music groups. This is incomplete and inefficient, but better than nothing.
  • Organizations like Associated Chamber Music Players have web interfaces (often funky) for browsing their members (limited by location, instrument, and playing level) and messaging them. ACMP focuses on amateurs. See also Chamber Music America (very limited and funky), which focuses on professionals.
  • The social features of some music apps can be used for people discovery. For example, MySpace has generic and crude mechanism for finding people (you have to log in with Facebook to use it). Note: MySpace doesn't seem to work anymore.
  • Idagio apparently lets users take online music lessons (for a price) from a set of professionals (selected by Idagio) - an extremely limited form of discovery. IMSLP has talked of doing something similar.

Music Match

Music Match is intended to be a complete and efficient tool for musical people discovery. Its features:

  • Users can describe themselves as performer, composer, teacher, and technician. Each of these profiles is detailed; for example, a performer profile includes preferred genres, instruments and per-instrument playing level, whether you perform in public, whether you typically get paid to perform, and so on. You can supply your location (ZIP code / country), a textual description of yourself, and links to your works or recordings. You can supply an audio 'signature' representing yourself as a performer or composer.
  • You can search for users on the basis of any subset of the above attributes.
  • Search is persistent. If you do a search, and (possibly years) later a new user satisfies the search, you get notified.
  • You can communicate with other users through private messages, and on public message boards.
  • There are various social features: you can 'follow' other users. When you view a user, you can see who they follow and who follows them.
  • There are various 'notifications': new search results, new messages, new people following you, etc. These are shown on the web site, and optionally in emails.
  • You can create 'Ensembles' - orchestras, choirs, chamber groups, etc. These have profiles analogous to performer profiles. Ensembles can be marked as looking for new members. Users can search for ensembles.

In its current form, Music Match serves only to connect people. It doesn't host scores or recordings; it doesn't handle payments to teachers.

Music Match doesn't currently have a critical mass of users. I've proposed integrating its functionality into music apps that have large existing user bases (such as IMSLP and Groupmuse), but these discussions didn't go anywhere.

If the Music Preference Service existed, it could be used by Music Match to increase search efficiency. Search results could be ordered by 'taste similarity'. Recall that a user U has a taste profile L(U) describing what U likes to listen to or play. If U is a composer, they also have a separate taste profile C(U) describing the music they compose.

The use of these profiles would depend on the type of search:

  • If performer U is searching for composers V: order by the affinity of L(U) and C(V).
  • If composer U is searching for performers V: order by the affinity of C(U) and L(V).
  • If performer U is searching for performers V: order by the affinity of L(U) and L(V).