I'm willing to bet that people who lightly use IM and haven't used Jabber before would find entirely different clients pleasing than people who are hardcore Jabber users. And which kind of user do you think is going to be more involved in voting on their favorite client? Right, the hardcore Jabber user.
But whatever, I have a distaste for "popularity" rankings in general. I think it's a mistake to think that the "community" has one view as to which clients are best, and that those views are what's best for first-time users. As long as you realize that popularity is its own kind of bias, then I guess that's fine. You guys were talking of eliminating bias, though, and I wanted to make sure that you were aware you weren't doing any such thing--you were simply shifting the kind of bias.
I'd take the bias of one person who has really thought through which client would be best for a first-time user over the ratings of the unwashed masses any day.
If you'd rather have a community-generated "average" of what's good, then that's fine. Just don't pretend you're "avoiding bias".
I'm glad people are finally putting things together for a community site--sorry if I'm being a pain.
Julian
On 4 Oct 2004, at 11:21, Richard Dobson wrote:
Any decent community site is going to have some kind of bias.
Of course but that bias should really reflect that of the community that supports it rather than the people creating the site, which is where the rating system comes into its own allowing the community at large to select a more balanced view of what are the best clients.
I don't think "eliminating bias" should be your top priority. I think "making it easier for people to get started" should be your top priority. You don't need to fear "hiding" other or new clients.
Of course not which is where the only showing the top few rated clients to the new users comes in, using the ratings you would hide the not so good clients automatically, but it also allows the site to be a bit more dynamic and represent the views of the community at the moment rather than those of the writers at the time in the past when the short list was compiled by them.
I'd go so far as to say that having 4 clients in a list may be a bit much for someone just starting out. I'd much rather have the user go through a quick form where we find out which IM clients (if any) they have experience with and recommending a client from there.
Sure, which is where the more complete list of clients comes in, you can filter the list of clients to the top say 3 rated clients with an ICQ like interface (if they are an ICQ user) or if they also use MSN a list of the top 3 rated clients with ICQ or MSN like interfaces.
For example, I'd only recommend Psi to someone who has heavily used ICQ
(in fact, I think most people who still use ICQ nowadays wouldn't like
any clients /other than/ Psi). For someone with zero experience with
IM, I'd never even show them the possibility of Psi. Just look at
hardware web sites--most of them have a selection process where you
tell them what you're looking for or what experience you have and they
recommend based on that.
Tailor the data presented to the user.
Sure this is where a more complete list of clients comes into its own, also if a user finds they dont really like any of the top rated clients they have the option of just clicking a button that gives them more options rather than being stuck with the only 4 that they are told about, its far better that they can find out about all the clients they can in the one site rather than having to hunt around in google looking for clients that are included into the recommended few.
On the other side, however, if the user doesn't give that data, we should still only present a quick 4 or so clients.
Of course which is exactly what I said, for the newbies give them a short list of the most highly rated clients, say 3 or 4 of them, and as you say we can always tailor that list further based on a simple questionaire.
I guess my point is, really, don't worry about the bias. There's always
going to be bias. The number 1 concern should be in making this
experience easier. That means there has to be some bias.
Yes but as I say, it is far better that it is the bias of the community at the current time (i.e. dynamic through rating the clients) rather than the bias of the writers at the time.
As long as the list of clients can be changed, I see no reason to avoid
bias at all costs.
Of course not, but if there is a simple more community spirited way of avoiding it its worth it, and you still solve the problem which is helping new users easily and quickly find a client that is good for them, but have the added benfit of a more balanced view of what really is the best client for them by having community involvement.
Richard
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ jdev mailing list [EMAIL PROTECTED] http://mail.jabber.org/mailman/listinfo/jdev
