Proposal for better ReverseScrobble.

[GNU]

Hi,
Just today I found out your wonderful tool, and its ReverseScrobble is
what I searched for a long time. In fact, I planned to code a small
tool that does that by myself (I am a graduated Software Engineer),
but gladly you've already done most of the hard job:)

Anyway, It works well, but, probably because of last.fm limitation, if
I have two songs from the same artist (very common situation is having
both "album version" and "live performance version"), the
ReverseScrobble gives them both the same play count, and only one of
them gets the "last played date" while the others are just without
this piece of information. This is very annoying, especially with very
popular songs of an artist ("Comfortably numb" by "Pink Floyd"), that
may come in many variations (For example, I have 6 different versions
of "CN" mentioned above).

Well, as a fellow programmer, I don't just complain - but I propose a
solution as well, you can alter it the way you think is right:

* First of all, we need to "rank" the various albums that hold the
"duplicate" song. This rank will help us divide the play counts
between the various versions. To determine the priority, you can use
some solutions: a) Random. b) Using data from last.fm (using the API
to get Album popularity by artist) c) Using data from Musicbrainz
(using PUID count)
* After we ranked the versions, you can divide the play counts between
them by some degrading function. I can suggest some, if you like..
Now, just for the discussion we can set something like:
1st - 50%
2nd - 25 %
3rd - 15 %
etc. etc.
* For the last played info, this is less defined, since if last.fm
doesn't give any data about it, you can assign practically any date. I
would suggest using real dates the song was played (I understand that
you collect weekly based charts of the last.fm account, so you may
assume that you'll have some play dates for each song).

Of course, if some info is unknown (i.e. if some versions lack album
information), you can resolve it in any way you like (My solution
would be to rank them as last in the priority).
I know my solution is way from perfect (since it still doesn't reflect
the real listening history), but I think it's much better than the
solution today.

In conclusion, If I had enough time I would be more than glad to help
you out, but since I have very little spare time nowadays, I can only
help you out in ideas and design solutions..

Hope my suggestion will be accepted, since this matter is very
important to me :)

[Mike Delpach]

Hi,

Thank you for your feedback. I fully understand the issue you arose and the
workaround you brought up due to Last.fm limitations sounds solid. I am much
like you struggling for time these days. I can't wait myself to see this
being implemented. I wished there were more contributors to this project so
we all could benefit. This will definitely be in the todo list.

Cheers,
McoreD