The Lagrange points (there are 5 for any restricted three-body problem) are positions where a satellite/spacecraft or even small celestial body (asteroid) can theoretically be stationary relative to two larger objects (such as Sun and the Earth-Moon system). Therefore the Sun-Earth L2 point was chosen for several observatories as it faces away from the Sun and the Earth-Moon system. It sits at a distance of 1.5 million kilometers.
L2 based missions include: WMAP, Planck, Herschel, JWST, PLATO, etc…
The above short list of missions (to be) placed at the L2 point immediately points out it might get crowded out there. We all know there’s a lot of space out there but these spacecraft are situated in stable periodic orbits around the L2 point itself.
So what’s the risk on interference or even collisions of L2 based observatories?
Moreover, the problem gets worse in time as these spacecraft become inoperative and start to buildup as “space junk” at L2…

AFAIK, they aren't stable orbits, but are quasi-stable loops around the L2 points, at least in the real world and not mathematical 3-body problem space. They still need some station keeping propellant to keep from drifting. Without station keeping, an object (space junk) will eventually drift away.
I don't think there's any valid cause for concern of collisions, space is really, really big and these quasi-stable orbits (a.k.a. halo orbits) can have significant (tens of thousands of km and more?) circumferences. Satellites are miniscule in comparison to the scales involved.

Space junk shouldn't start to build up as the Lissajous orbits around the L1,2 & 3 points are not perfectly stable and require significant station keeping to remain there.

The orbits are at L2 are of the order 100,000km I think, so it will be a while till there are any crowding issues.

There seems to be some abuse of the terms "stable" and "metastable" in the literature on spacecraft placed at the Lagrangian points, but the truth is that L1, L2, and L3 are all unstable -- not metastable -- and L4 and L5 are metastable, not stable. That means that an unpowered object at L1, L2, or L3 will drift away over a period of months/years -- in the case of the Sun-Earth points, that most likely means "drift away into solar orbit," so the space junk problem isn't an issue. You could get junk at L4 and L5 though -- perhaps we should call it "Trojan Junk".

Unstable is like a pencil standing on its point. It will fall unless it's absolutely perfectly centered, and even then the tiniest breeze will make it fall. Metastable is more like a coke bottle standing on its base. The wind won't knock it over, but bump it hard enough with your elbow, and it'll fall. Stable means it's as low as it can go -- like a coke bottle lying on its side. You can formalize this in terms of potential vs. kinetic energy, but the idea is the same.

--Greg

Well the area around the L2 Lagrangian point got "less crowded" as on 8th September 2010 sent, a 20-minute burn brought the WMAP spacecraft out of L2 and into a heliocentric orbit. WMAP completed 9 years of CMB observing, Planck is continuing the research...

Interesting. Any source for this info?

www.dmuller.net/planck gives the current distance between Planck and Herschel. Currently 390,000km ... that's like from the Earth to the Moon, so plenty of space to share.
EDIT: BTW, the Planck orbit is somewhere in the magnitude of 800,000km IIRC

Spaceflightnow has picked up the story:
http://spaceflightnow.com/news/n1010/06wmap/