the average rating problem
-----BEGIN PGP SIGNED MESSAGE-----
Pedro Bessa wrote on 24/11/11 10:44:
> The people who install a software think the software's description
> goes towards their objectives. If the software does what its
> author says that it does extremely well, those people rate it 5. I
> see lots of software whose description doesn't even go towards my
> objective, but are still rated 5. I mean, software that I'm not
> interested in using above software that I'm interested in using in
> the lists. That's really weird!
Weird compared to what? The same happens with ratings of books, or
hotels, or restaurants, or anything else.
> We must keep track of usage interest. If a software is read by 100
> people and 90 from among them choose to install it, that means 90%
> are interested in using it.
> To keep track of what apps were read, we can 1: mark read, mark
> unread, save marks, load marks.
The problem with that approach is that the effect would be both
invisible and obscure, so no-one would bother.
> To keep track of what apps were read, we can 2: don't show more
> than one app at the same time, show mini description, expand to
> full description, collapse to mini description and a link to the
> next app in the current category.
> In my opinion, 2 is better than 1.
That would require uploading data to a server whenever you viewed the
screen for an application, and whenever you installed an application.
It's an interesting idea, but again the effect would be invisible and
obscure, so I don't know how we'd interest people in opting in.
For the next version of Ubuntu Software Center we plan something
simpler: recommendations based on what you have installed. If you want
to see recommendations, you'll opt in to uploading your inventory.
People will opt in because it has a useful, near-instant effect.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
-----END PGP SIGNATURE-----
ubuntu-desktop mailing list