It's really very straightforward. Devise a testable hypotheses. Set up a few very simple experiments. Preliminary results are publishable and can generate grant money. Mind you, it's what all scientists have to do. Why should IDers get a free pass?
Pretty neat way to ditch the whole issue, no?
On one level, it's a way of keeping junk out of journals. Junk still slips through--it's not a perfect system. On another level, it's how science is done.
I have no idea how to construct a testable hypothesis for an algorithm from inception (I suspect that A-G is onto something here...). Do you have any ideas how to go about doing that?