Page 29 -

where λ defines if the patent is licensed (licensed = 1, unlicensed = 0), b is the y-axis
intercept, and the k’s are constants. We observe significant differences in the signals of
these patent populations—calculated results and stylized representations of the data for
signal and aggregator Web hits results are presented (Figure 9C and 9D, respectively).
While both scale linearly with patent citations, licensed patents receive more signal Web
hits per citation than unlicensed patents. Licensed patents also appear to accumulate
aggregator Web hits to a greater degree per citation than unlicensed patents. This
acceleration of accumulating Web hits in licensed patents may allow this measure to
appear more quickly than citations because of the delay in patent issuance [42], thus
representing a harbinger of technical importance.

Scholars use patent citation analysis to retrospectively assess the patent record to
understand modes of collaboration, analogous to the work done in this study. However,
this Web presence indicator may allow innovators, scholars, and research managers to
obtain more dynamic and real-time intelligence regarding the commercial value of
patents. For example, two unique patents—one commercially used and the other not—
are indiscernible in terms of vintage and number of citations but can exhibit significantly
different Web presence signatures. Since the mechanics of Web presence accumulation
are materially different from those underlying citations, this novel measure provides new
information on the value of a patent. Populations of licensed and unlicensed patents are
known to be different given the ex post information about whether they have generated
commercial interest, and we demonstrate that the populations exhibit different signatures
in the aggregator and signal Web hits that are statistically different. The differences in
the signal Web hits are the expected differences in the commercial impact of the patents
while the aggregator Web hits are likely an amplified measure of the technical impact of
the patents, typically measured by citations. With further refinement, aggregator Web
hits may be able to mirror analysis that currently uses citations to predict technical
impact, while adding signal Web hits as a measure of commercial importance.
Combination of these metrics may allow for easier and faster assessment of patent
impact. While this measure is noisy and requires refinement, these preliminary data
indicate that Web presence analysis may be able to help determine or predict commercial
value of patents as distinguished from technical merit.

In an attempt to extend these observations that signal Web hits are a measure of
commercial importance, we performed the same Poisson regressions using signal hits that
we performed on the patent citations to test differences between the factors producing
commercially relevant, rather than technically relevant, innovations.

3.3 Breakthrough Sources
Technology importance (citation based) and commercial importance (Web presence
based) findings are summarized by technology in the following sections. Effect sizes are
expressed in terms of absolute citation or Web hit count differences, and we report both
observed and simulated variance of like patents. Observed mean values for the patent
group of interest (b) are compared against all available control groups (a) for the contrast
variable of interest. We report all significant findings (regressed or simulated p < 0.1);
greater coefficient size and agreement between the two statistical methods indicates
increased certainty. For each reported effect, there is an unreported significant finding in

22