Do NOT use the 7+-2 principle as a guideline for building navigation menus. The context of the Miller experiment was entirely different. IIRC it involved rote memory, which is not the case for navigation menus.

A more pertinent number is 4. A paper last year found this to be the upper limit of recall of disparate visual items. The phonological route and visual route do not have the same capacities, and navigation menus are more visually-driven. The other point I would cite is the subitization limit, or the intuitive sense of visually-presented quantities, is efficient up to 4-5 items. Granted, nav menus can stay open, therefore require less brainpower to maintain in memory, which can push the upper bound, but strive for 4-5, rather than 5-9.

Oh yeah, also, good job for citing the Magic 7 and giving me 30 items to remember... (it was helpful though, thanks)

"...when searching for specific information on a website, [people] focus only on the parts of the page where they would assume the relevant information could be, i.e. small text and hyperlinks."

Hm, perhaps Google's success in advertising isn't because the ads are more relevant to the content, but because they look like relevant content.

I bet you all were "early adopters" of banner-ad blindness. You're probably blind to Google ads now, too. (I almost never click on them.) So... how long will it be before Google-ad blindness kicks in for the general population?

That is an extremely helpful article.