Nathaniel Borenstein wrote:
On Dec 22, 2009, at 12:45 PM, Michael Thomas wrote:
And it's not like this sort of thing is anything new anyway: lots of
vendors have "report as
spam" widgets that get bolted onto the side of your favorite MUA. A
little standardization
would be nice though as it would decouple that UI hassle from the
actual job of filtering.
Absolutely -- the report-spam UI will almost certainly be better if
it's integrated with the MUA and agnostic regarding the spam engine
receiving the report. The only major open question I'm hearing is how
much information that report should contain. Clearly it should be no
more than the number of bits that the user himself can be relied on to
provide, where our differing opinions might be resolved via user studies.
It might also be worth considering offering 1 button to most users,
but 2 buttons to users who understand the distinction well enough to
change a default in their MUA in order to get 2 buttons instead of 1.
I conjecture that the users who would take that action would have a
much lower error rate than the average user. In that scenario, most
users would send back a single bit "unwanted" message, but
sophisticated users could send back two (or even more?) types of
"unwanted" message. That might be the cleanest data we could hope to
get. -- Nathaniel
I think the problem is that if you open it up to more than one bit, it
begs the question of what the
actual number of bits such a button is. I'd say that it's probably got a
lot of bits -- far more than is
likely that any user could be bothered with.
Want/don't want is nice in its simplicity, and I suspect it's about as
much as you can expect from users.
However, there's probably a lot of data that MUA's have at their
disposal to see how you react to mail.
Like, oh say, educing the duration that you viewed a piece of mail. Or
whether you replied or forwarded.
Or whether you have a habit of deleting particular kinds of them
en-mass, and other kinds of behaviour
based data.
I think that if we stopped with this absolutist campaign of "spam/ham"
(most of us are not on some
paladin's quest against the evils of spam, after all) and focused more
on the context sensitive job of
prioritizing mail, we'd all be a lot better off.
Mike
_______________________________________________
Asrg mailing list
Asrg(_at_)irtf(_dot_)org
http://www.irtf.org/mailman/listinfo/asrg