Less is more. Twitter is the embodiment of this ideology. 99% of normal people don't care about Twitter clients.
Back in the olden days of Basic Auth, you would email one of the API devs at Twitter and get a special source token to put in your code for client attribution. Then, Twitter said you couldn't get a source token any more and had to use OAuth to specify your client source. This was one of the main carrots they dangled in front of developers to get them to switch to OAuth before the Basic Auth shut off; "Use OAuth so that your client source will show on the website."
It would be one thing if it felt like a snub, but now it feels like a bait and switch.
Disclosure: I've written several twitter apps and worked on the Twitter API team as a contractor for a bit last year.
Do you think? I'd have expected most people came to twitter.com and after some time (and possibly frustration) learnt of other ways to access the main features (ie 3rd party apps).
No, you either don't spend enough time with these 'normal' people or you are simply too biased.
As someone in University right now watching his 'normal' friends and colleagues join Twitter, and even just other 'normal' people in general -- very few to none use a Twitter client, they all use the Web. The one exception that I've seen some use are Blackberry people and ÜberTwitter.
I think one reason geeks like us use clients is to have a constant flow and connection of updates coming from Twitter. 'Normal' people on the other hand use it in a more 'get-in-and-out' sense, they get on and post what they're doing, read what the last few things their friends wrote and then get back on to Facebook or whatever.
(Outlook and IM all week at work is bad enough, thanks.)
A couple of weeks ago they released a graph suggesting that 78% of all users use twitter.com (http://blog.twitter.com/2010/09/evolving-ecosystem.html)
If you look carefully, they explain that this 78% is the percentage of people that have logged into twitter.com at least once in the last 30 days. Unfortunately this is not very useful since there are a multiple things that can only be done on twitter.com and not in 3rd party clients, either because the functionality is missing from the API or is broken (they started to fix some of these though). So even if people use a 3rd party client most of the time, they would still log into twitter.com every once in a while and thus will be counted as a twitter.com user.
A more useful statistics would be the percentage of users who use any given client as their main way of interacting with Twitter. Of course twitter.com would still be the biggest client, but not with 78% of all users, more likely somewhere between 40-50% (based on stats by twittersource.info, tdash.org and stats of Hungarian users I've collected).
As for the original post, we're running a web-based client and twitter.com is one of the most important referers so I'll have to agree that removing the "via client" attribution will hurt developers in a big way. (Yeah, I know, they are not removing it just making it a lot less prominent)