Does innovation truly require more collaboration than in the past?

Some interesting info on this in a piece from Physics Today entitled Too many authors, too few creators. How could you tell if more collaboration is truly needed? This piece uses patents as a baseline for comparison. Under the patent system apparently listing someone as an inventor on a patent will invalidate the patent if it's found that they didn't make some substantial contribution - this doesn't apply to academic publications.

What does an analysis of the changes in the number of authors per patent versus the number of authors per paper between 1965 and 2011 show? There's an increase in authors per patent but not as pronounced as for academic publications. Do different academic journals differ much in this regard?

Both Science and Cell, for example, do ask that all authors of an accepted paper “state their contribution to the paper,” but they do not list any criteria for actual authorship, nor whether specific types of contributors should be relegated to an acknowledgement section.

Of the publication venues noted, the number of authors per paper in Science fairly closely follows the number of authors per patent. (Cell isn't broken out in their table). The trend in authors per paper might be attributable in part to the publish-or-perish paradigm, but perhaps something like the approach taken by Science would help curb this somewhat.