Android is the most original-UNIX-like in it's development but I imagine you could extend it to systems in which large portions of source are available (since that is the kind of project C&B discussed) such as iOS and perhaps Java. Google Chrome. vBulletin. The Pine MUA. The ssh.com RFC 4253 implementation. Various parts of RHEL. And probably numerous "Free" programs that end up following the Cathedral model merely due to the culture of their maintainers.
We can think of many but the point is to make sure we're on the same page. It's not like all software projects in the world are neatly divided into "bazaar" and "cathedral". I don't know that Google Chrome is a "cathedral", for instance.
It's ridiculous that the author refuses to give a single example, instead opting to say basically "see! You people don't know what a cathedral is, just like I said."
I will cite as evidence two of my favorite programs ever:
WriteNow (for Macintosh) an excellent early word-processor written entirely in assembler. Unequaled for many years for its combination of stability, raw performance, and ease of use, ultimately it simply could not add new features (let alone make the jump to PowerPC) and died on the vine.
HyperCard, which was perhaps one of the most dazzling, innovative, and influential products ever to ship, which pretty much stopped evolving once its original programmer lost interest.
Today we have the phenomenon of the incredible version one product, usually developed by one person, which never really makes it to 2.0. These seem very much like cathedrals.
Most projects go nowhere, but once a project gains momentum it's going to split into pieces and each piece has a chance of "going bazaar" at each step.
Now, I've not been working in the software industry since before 1990, so I'm obviously not qualified to comment, but I'll say this: I can understand why a cathedral model might be warranted. Even using something like CMMI might be a good idea in some cases.
But for a lot of things, especially exploratory/experimental things, the bazaar model is really nice and can reap significant benefits. And the nice thing about the bazaar model is that if you want to follow a cathedral model, no one is stopping you! Go off and be your own little dictator with a "grand unifying vision". Come to think of it, that seems to be what many of the most successful open source projects are: one (or a few) people have a vision of an itch they want to scratch, and they pursue it with a bloody-minded persistence. The bazaar only comes in when someone forks or in the fact that anyone can compete or (try to) contribute.
And BTW, I do know who you are, and have a lot of respect for you, but in some ways this article (and your comments here) could be read as reactionary against the success of Linux and other more open OSS; the BSDs have always been more insular (or discriminating depending on your POV) and developed like cathedral models than Linux; oddly enough this has resulted in three distinct BSDs while there is still only one Linux kernel. I will agree that reading Brooks (and other computer history) is almost always a good idea; just MMM was enough to open my eyes to how little the industry has progressed (VirtualBox/VMWare? That's nice; IBM was designing full system emulators for hardware that didn't yet exist in the sixties).
The only example that jumps out at me is the original Unix, and I think you'll agree that comparing software from 30 years ago that does vastly less than ... pretty much anything out there these days is not an entirely fair, nor useful comparison.