improved representation of bias

[Caleb Cushing]

On 12/12/08, Jud Craft <craftjml at gmail.com> wrote:
> That does bring up a very good point.
>
>  What the heck does "90% fuzzy bias" mean to a user, anyway? :D
>
>  In the case of "over 100%" though, I think it's pretty simple:  the
>  two may be mutually combined.
>
>  If you set 75 rock/60 pop, that means (at least I would believe) that
>  for any song in the random playlist, 75% of them are rock, and 60% are
>  pop.  That may mean that a few of them are both pop AND rock (I'm
>  guessing the overlap, since you can't go over 100, is in fact 35%).

considering I was referring to genre and not tags, a track can't have
more than one genre, I don't believe amarok is smart enough to dissect
a genre pop/rock in an id3tag meaning rock, pop, a pop/rock are all
mutually exclusive.

>  No clue what proportional vs. fuzzy means, but that has been
>  introduced in posts on the Amarok blog before.

I know what it means... actually it's ends up being pretty much what I
suggested for 'radio' play.

> In Amarok's defense, it's not like iTunes tells you how it figures out
> it's shuffles, recommendations, or Genius lists either.

the benefit of open source is we strive to make something better?
besides this isn't a random shuffle it's weighted and we should be
able to understand the weighting if we can configure it.

it might be possible to show stuff based on a points system rather than percent.

-- 
Caleb Cushing

http://xenoterracide.blogspot.com