I think an appropriate model for non-junk/junk status would be a long-tailed distribution, where the "non-junkness" of some DNA portion never gets to zero, and life is always a little bit better with a little bit more of the distribution (i.e., the area under the curve approaches infinity). These long-tailed distributions have been heavily discussed in relation to retailers like Amazon, which can have 1 million book titles in their "inventory", whereas a local retail bookstore might have perhaps 20,000 titles (I'm guessing as to the exact number, but they certainly don't have 1 million). Yes, there is human life when the number of books available is only 100 or 1000, but life is much richer with 100,000 or 1 million books available. So, although Sturgeon is "correct", when he says that 95% of everything is crap, he's discounting the value of having 20x the number of books to choose from. Another long-tailed distribution is that of patents in the patent system. If we're being charitable, we assume that every patent is valid, and that there is a perceived use for every patent. However, the technology for some patents "masks" the need for other patents, because so long as patent A technology is available, it may be strictly better than patent B, so patent B will see _zero_ use. However, if for some reason the technology in patent A becomes non-viable -- e.g., it depends upon some material which has suddenly become unavailable, patent B receives all of the attention and resources that were previously given to patent A. Whether patent B was "junk" before becomes merely a matter of definition. At 06:50 PM 9/8/2012, you wrote:

...

Further it seems that this test has been done, though removing only a portion of the junk DNA: http://www.ncbi.nlm.nih.gov/pubmed/15496924

--that was interesting. If the New York Timesy claim that our 3 billion base-pairs of "junk DNA" is filled with a "switchboard" with 4 million switches... which would be on average one switch per 750 base pairs... then this experiment where they removed a 1.5 Mbp and an 0.85 MBp region from junk mouse DNA, then the resulting mice grew up and seemed to be fine... would have removed an expected number of over 2000+1000 of those "switches."

Kind of like if you randomly zapped two swaths of 2000 and 1000 transistors from an early computer with 4 million in all. Would that leave the computer apparently working fine? Even if those computer designers had tried to design it to be robust against faults (which they mostly did not), still, this amount of fault tolerance would be pretty impressive.

So the 4M switches in our DNA if there, must either be very nonuniformly distributed, or very unimportant, or part of an extremely fault-tolerant circuit.

If, in contrast (to compare with the genes in DNA, traditionally regarded as non-junk), you randomly zapped one gene per thousand in your set of about 25K genes (total zappage: 25 of your genes) then what? Would you be fine?