Dan Corkill (Blackboard Technology Group, cork@cs.umass.edu) offers some additional advice based on experience. Commercial potential of proposed NSF SBIR work should be DIRECT and OBVIOUS. A proposal that emphasizes enabling technology and generally advances the state of the art, but without a direct commercial application, is unlikely to pass the reviews. (As an academic proposal, it might do well...) Researchers used to the academic review process should not think that the SBIR reviewers will support a generic advance, no matter what the potential commercial possibilities.

[The NSF brochure reads as if there were no conflict between fundamental scientific research and quick proof of commercial feasibility. Companies that want to do fundamental research sometimes support themselves on sequential Phase I grants from different agencies, with little intent to reach Phase II. They'd be in trouble if such intent were proven, though.]