DVDs might have a 100 year life, I don't know. I'd be very surprised to find that *recordable* cd/dvd media on average has anything like that.
When optical storage was a new thing, platters were being produced in cheap plastic for 'normal' use and glass or more stable plastics for archival use.
My recollection is that the archival quality media indeed ran at a nominal 100 year life, however on average cd-r is pretty chancy see: http://www.mscience.com/survey.html
Here's a link for a 2500 word article on what's necessary to achieve reliability in CD-R: http://www.mscience.com/longev.html. How many people do you think actually observe this kind of detail in making cd archives? --- I have no knowlege that DVD-R is considered yet more reliable than CD-R
Tape archival quality *is* well understood. Here's a brief survey of actual experience, indicating that many users find individual tapes may only be good for 2-3 write cycles. http://www.sunmanagers.org/archives/1992/0319.html. This squares with my operating experience.
Exabyte claims 30 year life for 8mm helical scan media. They also claim 50,000-pass mechanical life (which very few people would say works in practice, possibly in highly ideal conditions?).
But in any event Exabyte absolutely warns users that the archival life claim is based on a single-write use.
I.e. you can't apply thier specs simultaneously. So iff one observes very exacting procedures (and expense) long-term archive is feasible on tapes, and probably on DC-R but if practically speaking few people will apply the correct procedures it's probably a moot point.
Finaly, once on archive-media data is no longer available for random deletion/edit/reorganization, so while you're right that ability to throw stuff away is important, archival systems don't help with that.