>Sure, this is inconvenient for the user who creates an account and
>immediately forgets their username/pass and didn't save it in their
>browser (really?). In that case, create a new twitter account - they're
>free! ;-)
I'm not sure I'm following you - or rather, not sure if you're saying anything different from what I'd (perhaps poorly) said.
I'm explaining the logic of today's conversion specialists. That's what they call people who spend all day trying to figure out how to convert a browser into someone who signs up for, say, a newsletter - or twitter.
These folks use A/B and multivariate testing to tweak the User Interface to see which approach to a sign up form converts the most users so you can "build relationships" with them. :)
They've probably already tested which approach converts the most users and they've determined, as I think you're saying, that a small minority of browser abandon the completion of the sign up. The majority simply sign up under a new twitternym. The other approach, two fields in which to type an email, to weed out typing errors (or at least a lot of them), is worthy, but doesn't convert as many browsers to twitter users. If you believe that your goal is to convert twitter users, then you're gonna go with the method gets the most conversions.
We do this sort of research in my business, and it's surprising how moving this block of text here or removing a form field over there, can have a huge impact on how many people convert. In one case, there was a suggestion to use a red background on the form. According to the logic of what is called "genius design" where the designer knows best based on some majikal intuitiveness about what users supposedly want, a red background would be horrible. Too flashy! You look like you're trying too hard.
And yet, the difference in conversions was 40%. People favored the flashy, trying-too-hard red background! Now, if you directly asked a statistically random sample of users, they would tell you that the red background turns them off. But that isn't actually how they behave. Which isn't surprising to social scientists. :)
shag
-- http://cleandraws.com Wear Clean Draws ('coz there's 5 million ways to kill a CEO)