First off, and I invite the archivists in the discussion to pitch in: archives and libraries are very different things.  Libraries tend to catalog each item.  Archives may have 1 catalog entry for "the collected papers of Henry Smith" and it's up to the researcher to weed through the boxes of papers to find the document he or she needs.
Both archives and libraries cull their collections.   You actually raise an interesting point in that regard, because libraries spend much more per item cataloged just to catalog it -- often more than the price of the book -- and so when they discard a given item, it's an expensive proposition.  But never mind all that -- both libraries and archives discard things.
When I refer to link rot, I'm talking about the deliberate adoption of a new URL naming scheme, without regard for static resources whose address need not change.  Let's not confuse demolishing a house with changing its street address.

On Thu, Sep 3, 2009 at 12:57 PM, John Gorentz <[log in to unmask]> wrote:
While I'm generally on the side of stable URLs and at least not letting them rot through carelessness and neglect, David does bring up some valid points.   I am reminded of something I once heard Le Roy Barnett (former head reference archivist at the state archives) say on the phone to a caller.   He explained to the caller (as only Le Roy Barnett could do) that the archives are more in the business of destroying information than preserving it.   He gave some figures which I don't remember about the proportion of governmental records that they destroy vs the amount that they preserve.   Maybe that's a standard line in the archive business, but for me it was a new way of thinking about it. They are selective about what they preserve.  

John Gorentz

At 12:13 PM 9/3/2009, Richard Wiggins wrote:
If you think ISBNs are used by any library as shelving numbers, and if you think a research library uses Dewey, and if you think libraries switch shelving schemes in midstream.... then I'm not going to be able to make you understand why changing URLs willy-nilly is common, but foolish.
URL changes that occur because you've hired a new director or installed a new technology are inexcusable.   Period.
And if you insist on changing URLs in a site re-org, you can still build a smart Error 404 handler that redirects to the old content.
Link rot is a willful, arrogant act.
On Thu, Sep 3, 2009 at 11:44 AM, David McFarlane <[log in to unmask]> wrote:

And why should I expect URLs to persist?  Shelf numbers for books do in fact change -- as time progress, libraries may change from, e.g., the Dewey Decimal system to Library of Congress numbers to ISBNs, and who is to say what further scheme will come along to supplant those?  Must we forever lock ourselves into old organizational schemes when better ones come along?  And the street addresses of homes or businesses do in fact change all the time, every time a family or business moves.  It is then up to the family or business to arrange for mail and visitors to get redirected to the new address for a short time; anyone who shows up at my address of 20 or 30 years ago is just a fool, and I feel no obligation to provide redirection from those old addresses.  And if an organization's URL scheme turns out to be a nightmare, should they remain locked in to it rather than revise it to one that better fits their and their customers' needs going forward?

And after all, aren't URLs and all they represent just a *human* undertaking, and humans notoriously unreliable and fickle, and their systems subject to constant improvement?  I am not saying that I don't want URLs to persist, in fact I would like that.  But I know better than to *expect* such persistence, that's all.

-- dkm

Why?  Why shouldn't URLs persist?  Do LC shelf numbers for books change?  Does the street address of your house or business change?