Truer words. I think there are some interesting questions to ask here though.
Mozilla isn't terribly different, it also sort of languished for years, it wasn't even possible to build it when it was "open sourced" the need Mozilla filled is a big part of the problem, it's a hard project to get into though. OpenOffice, Ingres, Firebird, and if you take a longer view projects like Zope, Chandler, and other projects that started out with either a corporate idea and code base or just really grand and public ideas have seemed do less well ('fail' is a strong word, there is code to use, it's just not highly used.)
Did Mozilla sort of take off because of the need? Or was it the combination of need and reduction in focus? Firefox is quite a bit less than netscape communicator or the full seamonkey mozilla (no email, no editor, no chat, etc..).
Ingres and Firebird are both easier to explain and harder to explain. There has been a tide of MySQL backlash in some circles and from all indications it looks like there is some kind of trouble in Sun with MySQL. Remarkably, most seem to have gone to Postgresql from MySQL or to like sqlite, hsqldb, and derby. Firebird, Ingres and throw SAPDB in to the mix are really good databases but you'd never know that from the community. I'd have expected at least a few more pockets of people trying them out, choosing them and making noise about it. Ingres does have that look of decay about it though, you can't see through that.
Zope and Chandler? I don't know what to say or where to target their wrong turns. Chandler seems to just have been over hyped from the start. Zope on the other hand produced a remarkable stack a decade ago, well documented, fairly easy to use, just completely missed the market and then was unable to adapt over time. I wonder how it could have been if they had produced something like a "zope in a box" that was a single one shot package that got a full Zope with database running, kind of like rails, just a run a few commands and you made a database backed zope app... things could have been different, I remember jerking around with different python postgresql connectors trying to figure out which worked and which was supported..
OpenOffice is a good piece of software, I'd say it's a very important and needed piece of software. For reasons I don't understand at all, geeks just don't like it. (They don't particularly like java or mono either, at least a large and vocal chunk of geekdom doesn't.) Part of it is the giant amount of time it takes to compile openoffice. Part of it is that there was some engineering that went in to it and whenever that happens there is a great debate, small things tend to take more effort. Part of it is this belief that it's incredibly slow, those rumors never go away and on a machine capable or running a modern distribution well it seems to run pretty well. There are probably some other little things too, language choices, quality concerns (they won't just include any patch) part of it is the audience. GIMP doesn't appear to be thriving, it's chugging along. There is not great opensource CSS editor either, go figure. These are products that are needed for the platform to reach a more broad audience but they are also products that don't really appeal to geeks, I hardly ever use a word processor or graphic editor. Somehow it seems the community has to either navigate this chasm or it's reached the limit of what opensource can really do.
The other part of it might be that people really just don't contribute but bit patches and it's a thankless job running such a project. The Linux kernel might be the exception but when you really cut through it all, there are a lot fewer contributors than you might think and there are a core of 10-20 guys that run the whole show. I tend to think it's a gap for the community to cross though, the community sort of thinks easy to use or wysiwyg or office like apps are the land of microsoft and apple and copying them is taboo (look at mono, another incredibly fine piece of software that people will fight to keep out of a linux distribution)
I don't really understand it. It's somewhat puzzling though, kernel hacking clearly isn't for every hacker, you'd think there'd be this pool of application hackers that would be pushing openoffice in different directions.