The Steady State Theory says, in rough terms that the universe has been and will be more or less as it currently is, because there are parallel processes of creation and destruction.
So, galaxies exhaust but then there are new galaxies, and the thing, as a whole, remains unchanged, in a way. Sure, it's not the same galaxy, and nothing that was in the old galaxy remains, but if you avoid specifics, things are the same.
I feel the same thing happens in the free software universe. The two forces are reaction and features.
Reaction is the creative force. Most, if not all, free software is reactive. It itches, I scratch, scratching is reactive. There are many examples:
There is no free desktop! Let's create KDE!
KDE is not the right kind of free, sort of! Let's create GNOME!
KDE and GNOME are too heavy, bloated and whatever! Let's create XFCE!
XFCE is not as lean and mean as before! Let's create LXDE!
LXDE is (we'll find something) let's create WHATEVER-DE!
This even happens within a project:
Hey, we created KDE!
KDE 1.x has no object model and GNOME does, let's do KDE 2.0 on CORBA!
Hey, that sucked, let's do KDE 2.0 on DCOP and KParts!
Well, that's old an crufty, let's do KDE 3.0 cleaner!
KDE 3.x looks old, let's throw all the UI away and do KDE 4!
And I am sure those familiar with any software product that has lived long enough to go through major rewrites and upheaval can do similar lists.
However, sometimes, the complains just don't go away.
Netscape 4.x is slow and crufty! We rewrote it as Mozilla!
Mozilla (SeaMonkey) is slow and crufty! We rewrote it as Firefox!
Firefox is slow and crufty! We wrote Chrome using webkit!
And guess what people say about Chrome? It's slow.
So, the lesson there seems to me that writing a lightweight, generally useful, web browser is impossible. Why? Because of features.
As projects age, they grow features. Like the strange ear and nose hair men start growing in their 30s, features are a fact of the lifecycle. And with features come code, because that's how you do features.
And code is a liability, as (I hope) you all know. The more code you have, the more expensive it is to add things, and to be swift about improving your application. Most successful projects die, or grow senescent, hobbled by the weight of their features.
So how does a project stay young? I can think of a few ways.
It may have a benign (or evil, for that matter) dictator, with the right amount of hostility towards features (Linux). It may be so exquisitely modular that features don't couple with each other (emacs). It may reinvent itself every 5 years and throw everything away (KDE). It may have a very clear focus on one feature and a culture around it (Bacula).
And for each of those mechanisms, there are incountable examples of projects with too annoying dictators, projects overengineered to absurdity, stalled rewrites that never release and absolute focus on a feature noone cared about.
Sorry, the universe is a tough place.