Idle thought from yesterday evening

[simont]

Sometimes, one band writes a song, or publishes an album, which has the same name as a different band. For example, last night lark_ascending played me some music by ‘Van Canto’, including a song called ‘The Mission’. And Leonard Cohen had a song called ‘Sisters of Mercy’.

It occurred to me last night to wonder how common this is. In particular, my most immediate thought was this: if one were to define a directed graph on bands with an edge from one band to another iff the former had a song or album named the same as the latter, would the graph be cyclic? That is, does there exist a band called A with a song/album called B, and a band B in turn with a song or album called C, and … eventually some band X with a song or album called A? My own knowledge of music is less than encyclopedic, but I wonder if any music-trivia buffs among my readers might enjoy having a try.

(Standard conventions for directed graphs apply: it's cheating for a band to link directly to itself, but two bands linking to one another are a valid cycle. Also, I'm prepared to be a little forgiving on the matter of whether or not band, song or album names have a leading ‘The’. Finally, I don't demand that the name matches be coincidental: if the song name directly inspired the band name or vice versa, as in the second example above, that doesn't disqualify the link.)

Comments

[fanf]

Yes, you need to forbid self-links because of eponymous albums :-)

In general self-links are permitted in directed graphs (e.g. in automata) so I don't think this is a standard restriction.

[simont]

Hm, possibly you're right. In undirected graphs an edge from a vertex to itself is usually disqualified (which is what I was thinking of), but Wikipedia seems to agree that the usual definition of directed ones is different.

[tackline.livejournal.com]

Is it not the case that for most (computer science) applications, self-links are trivially reduced?

[songster.livejournal.com]

Don't see why we need to forbid self-links. The truly interesting question is "What is the longest cyclical graph". It may be that the answer is 1, but hopefully there's something longer out there.

[simont]

You can see it that way if you like, but I don't want to see a self-link being used to pad out the length of a nontrivial cycle!

[mooism.livejournal.com]

Can we just say that a band can’t appear more than once in a cycle? Depends whether you see figure-8 cycles as cheating, I suppose.

[simont]

I think it's a question of improbability: the impressiveness of any result is directly related to its unlikelihood of appearing on the assumption that graph edges appear at random with an appropriate probability distribution.

Self-links are known to be common (eponymous albums and/or songs). Hence the base probability model on which to judge impressiveness should reflect that, by considering self-links to appear with a significantly larger probability than non-self links. So including a self-link in a cycle doesn't increase its impressiveness by very much (so I'd be inclined to consider it a cheap trick), and finding a cycle entirely composed of one self-link is definitely too easy to be a challenge. But two nontrivial cycles sharing a vertex would be a more impressive find, so I'd accept the resulting figure-8 as a worthwhile result.

[bugshaw.livejournal.com]

You could have fun with Peter Gabriel and his five eponymous albums.

[sparrowsion]

Five? I'm missing one….

[bugshaw.livejournal.com]

I was wrong, of course. I think Smash Hits had a running joke about albums called Peter Gabriel when So came out. (I should have written to Black Type and suggested they call his first four albums Doh, Ray, Me and Fa)

[valkyriekaren.livejournal.com]

The Sisters of Mercy lifted a lot from Leonard Cohen (Some Girls Wander by Mistake), as well as other artists including Bob Dylan (the lyric "stuck inside of Memphis with the mobile home" is taken from "stuck inside of Memphis with the Mobile blues again").

[gerald_duck]

My hunch is that the most promising starting point is the Sky album Mozart. Mozart didn't compose a piece called Sky, but so many of his pieces have names that are in turn the names of bands or artists that it's reasonably likely one of those will have a track or album called Sky.

[bugshaw.livejournal.com]

<CopiousSpareTime>I wonder if last.fm will let me scrape their data</CopiousSpareTime>

[zotz]

Freedb or Musicbrainz might work too.