In the run-up to the study results, the authors lament that only 1% of U.S. patents face re-examination. Viewed in a different light, one might think that patent examiners do a decent job in the first round.

Two years of litigation cases were studied - 1995 and 1997, with only 23 cases of 3127 reviewed terminating prior to the Markman ruling, which substantially changed the litigation claim construction process.

The study found that only five percent (5%) of cases filed went the distance to a court ruling; 80% settle; the remaining 15% resulted in a ruling, half of those in summary judgment, the other half (7.5% total) went through trial.

At least 98% of asserted patent claims are valid. Only two percent (2%) of at least one claim asserted were ruled invalid. Rulings of invalidity tend to occur early in the litigation process - up to 75% by summary judgment. The upshot is that the court is consistently willing to entertain invalidity arguments early in the proceedings.

By contrast, a finding of infringement necessitated a trial in about 75% of cases, the other 25% by summary judgment.

Around 20% of the verdicts resulted in damage awards, which averaged around $5 million, though an undisclosed percentage of cases settled with money changing hands.

Only 19% of cases through trial verdict resulted in a permanent injunction, and only 7% of those case from summary judgment.

A patent litigation case was on the docket an average of 15 months, a median of 10 months; that is, half the cases filed were over within 10 months. Summary judgment motions were attempted in just over 25% of cases.

The average number of documents filed in a case was around 65, and the median 25.

While no new hard figures on litigation expenditures were given, it appears, given Markman, that legal expenses are front-loaded, as the Markman ruling generally puts the writing on the wall as to outcome. The study cited an old figure of patent litigation averaging $1/2 million per asserted claim per side.

In reading the prose around the statistics, the authors clearly held a grudge about the cost to society of litigation vis-à-vis invalid patents, but the statistics of only 2% of cases finding a single claim invalid don't bear them out.

From anecdotal evidence of personal experience as a prior art searcher, a guesstimate figure of invalidity from work assignments reaches towards 15-20%, but since 80% of the cases settle, some at least with invalidity as a looming specter, the 2% study finding may be but a tip of an larger invalidity iceberg.

The full report can be downloaded from the Social Science Research Network.