37signals comes down against a decree of no more remote working. I am shocked (shocked that there is gambling going on in this establishment).
Much like hiring practices this is and will continue to be a divisive issue. This one however seems to be largely driven by personal preference: if you want to work from home, you can't understand why anyone is against it. If you don't, then you do.
I fall in the camp of not necessarily being against remote working but it's not something I want to do, I understand why companies don't want it and I don't think working remotely scales.
The last is the most important point. Many take the view that they can be much more productive with flexible work schedules and when people don't bother them with trivial stuff and when they don't waste time on the commute.
Firstly, working from an office doesn't not exclude a flexible work schedule. I've gone through phases where I don't get into work until 3pm. A guy I know decided he would take Thursdays off and work on Saturday instead. My manager forgot I was going on vacation for a few days and I didn't get a call or email asking where I was until midday Tuesday (having not shown up on Monday).
As far as not being bothered, this can certainly be true. This is actually part of the reason why I often wear headphones. Noise-cancelling headphones do a pretty good job of filtering out a lot of noise but just the act of wearing headphones I tend to find acts as a psychological barrier from others bothering you.
As for commute, this much is true and is a particular problem in the Valley. For me, my commute is a 7 minute walk to work (in NYC). YMMV. :)
But a lot of people make the same mistake with this issue that they make with hiring practices: they tend to think that the individual case matters. It really doesn't. If a hiring process weeds out some qualified candidates, it doesn't mean it's broken. Whether or not it's broken is determined by whether the company finds a suitable candidate, how long it takes and how expensive it is. False negatives don't matter (to a company) as long as you get a positive result.
The same goes here. The individual thinks they can work from home and they may well be right but there are bigger issues. For teams--particularly large teams--to work together requires a certain camaraderie that is orders of magnitude easier to manage when physically colocated. This also applies to teams that are geographically split. All other things being equal, a team in N sites will perform better than a team in N+1 sites.
The other issue, and this particularly applies to large companies, is one of culture. Culture like team cohesion is orders of magnitude easier to spread when physically colocated. For a company the size of Google (my employer), IMHO this is far more important than any individual perceived benefits about permanent remote work.
Google has (IMHO) done a remarkably good job of maintaining cultural consistency such that transplanting an engineer from one team to another is relatively seamless. This isn't just about common tools either.
Chris Dizon wrote a great post called Twelve months notice [1]. I like it because it articulates the primary difference (in my experience) between those who work remotely and those that don't. In my experience, those who work remotely tend to be far more in the transactional work category. While there's nothing wrong with this and I think it's particularly suited to freelance consulting, IMHO it is at odds with running and maintaining a large engineering organization.
But I know I'm not going to convince anyone. This another polarizing issue. Just be aware there are bigger issues than the individual not wanting to drive to work.
EDIT: Also, I disagree with the position that this "punishes" good employees. Yahoo wants to build (or rebuild) an engineering culture. They've decided this is easier to do with engineers physically colocated. This doesn't preclude workers from working from home on an occasional basis (eg waiting for the cable guy). It simply means the default is you come into the office.
[1]: http://cdixon.org/2009/10/23/twelve-months-notice/