Scientists are always open to practical approaches which actually work and bear up to testing.  To this end, I offer two metaphors for ways in which open-source technology/culture is carving out a niche at the expense of closed source, drawn from my experience as a working research scientist.

Linux vs Windows and commercial Unix
I realize this is an overused comparison, but it works.  Linux is by far a more reliable and stable system that works because of the linux community. When I started as a scientist in the 80s, unix based workstations were the gold standard for desktop computing.  Faster chips for pcs and Linux have lowered the cost and obliterated that market.  As you well know, Windows is looking nervously over its shoulder at Linux, and think tank types are even writing articles about the possibility of it overtaking Windows.   It has pretty well annihilated commercial Unix and is making enough inroads on Windows for think tank types to consider it overrunning it in recent articles.  The analogy of ``Linux is to the blogosphere as Windows is to the MSM'' is sufficiently well known here for me to go on to-

Open Market scientific literature vs refereed journals
 Just as MSM fact checking and editorial standards are supposed to confer a level of quality control over articles, the vaunted scientific practice of anonymous peer review is supposed to add a layer of quality control to our product, research articles.  Well, it is the best we have, and I do participate and subscribe to it, but the fact is that it is breaking down under similar pressures as the MSM journalism system:  fiscal pressures (the need to be productive in research to get research funding) and the exponential growth in both the amount and diversity of information. 

The refereeing process is, too often, broken by (i) negligence--busy researchers acting as unpaid refereesproduce sloppy reports or decline to review due to the cost in time and intellectual energy to do so; (ii) incompetence--just as for journalists in the MSM, some badly trained scientists do make it through the system; (iii) complexity--much of the cutting edge research is at the boundaries between disciplines (eg, physics and biology) and there are relatively few  practitioners skilled in judging such bridging efforts; (iv) the dark side--papers/proposals can and will be spiked by competitors, and sociopaths find ways to exploit the system for their own gain as noted below. 

When you add to this the substantial costs that journals charge for subscriptions and articles, you begin to understand why there is a revolt well underway.

The organic response within the community is something very much like blogs or Wikipedia or Linux.  There is a movement towards preprint servers which get articles out there and expose them towards the same kind of open source advancement and testing you find in the blogosphere and Linux world.  (Check out, for example,  the preferred server for mathematicians and physicists, arxiv.org.)

The driving philosophy of the open-source science movement is very much like Linus Torvalds and the blogosphere:  put it out there and let it stand the test of time.  If it is interesting, people will pick it up, test it as per Descartes or Bacon, and run with it.  If it is garbage, it will sit there and be ignored.  The truth is that there is a lot of garbage in scientific journals, or at least a lot of science not worth looking at.  Most articles even in the most prestigious journals garner only a few or no citations from other researchers in their lifetime.  Moreover, the system can be easily gamed by con-men and women; for a scientific counterpart to Jayson Blair or Stephen Glass, check out Hendrik Schon.  The premiere journal Nature once let through a ludicrous article purporting to provide a basis for homeopathy via memory effects in water, where the dilution process carried out would have led to less than one active molecule in the allegedly effective end solution. 

This said, it is unlikely that we will ever fully replace the journal system.  Far too much of the cultural infrastructure of science depends upon the final publication.  While superstring theory will advance far more quickly via the web than the journal evidence allows, superstring theorists still need published articles archived in journals to cement their ultimate reputation.  

But the preprint servers are forcing change upon the power structure of the journal system.  More journals are acknowledging the benefit of pre-publication  posting  on preprint servers, and a maverick band of biologists  forced Cell to lower its costs and change its practices while launching their own competitive on line journal (PLOS Biology).  

So, in response to Michael Tomasky, and in harmony with Digby, I say let the blogs roll on.  Linux, Wikipedia, and arxiv.org offer ample proof by analogy that the older powers and large world are better served by doing so.