 |
 |
Some critics have said that due to the low citation numbers of non-patent references in software patents, the burden of showing non-obviousness and innovation has not been met and these patents should be considered invalid. Additionally, the "state of the patenting art" is marked by increasingly high number of patents in technical areas of minimal revolutionary innovation.
This study was undertaken to investigate the practice of citation of non-patent references in patents over the course of the last 27 years in areas of decidedly non-technical nature. In addition, the growth in the numbers of patent issued and numbers of citation of any prior art was assessed. Comparison was made to the major criticized area of software patents
The following Patent Classes were included in the study:
| Class | Description |
| D18 | 365 Designs - Printing and Office Machinery |
| D01 | Designs - Edible Products |
| PLT | Plants |
| D04 | Designs - Brushware |
| 365 | Utitlity - Static Information Storage and Retrieval |
The software data for comparison was taken from an assortment of classes, primarily Classes 364 and 395 as compiled by Gregory Aharonian, Internet Patent News Service, in '1997 Software Patent Statistics - Feb. to July' and '1995 US Software Patent Statistics'
Tremendous growth in numbers of patents issuing in the software arena is noted in the Aharonian articles. This growth was noted as having no relation to to the minimal revolutionary innovation in the software industry. When the software growth rate is compared to the growth rate found in the studied classes, it is apparent that the growth rate is not confined to the area of software patents, but appears to be on a trend similar to other classes of patent. Some of the classes studied are arguably in areas of even LESS revolutionary innovation.
|
|
It appears that there has been a surge in patents issuing in all Classes studied, especially in the late 80's. That growth has slowed somewhat in the 90's. Obviously data for 1998 is not complete yet.
This trend may be rooted in the economy rather than in science and invention. Inventors and investors who are developing and commercializing products in a competitive marketplace are likely to want to protect their investment. One very legal form of protection is patenting. A patent can ensure a corner on a market or at least a healthy head start to commercial success. It is quite likely that the rise in the number of patents is not linked to the joy of pioneering invention, but rather the joy of acquiring cash.
The studied classes yielded an average number of references cited per patent currently in the range of 9. The data indicate that the average citation per patent was roughly 4 in 1972. Therefore, the average has been steadily, although slowly, increasing with time.
This result is not surprising. Information of all kinds is increasingly available through the internet and other forms of electronic or published media. This rate of increase of citation may not be increasing as fast as some would desire, but critics must remember that information does cost. Most businesses doing the patenting still have to turn a profit, Rule 56 or not. And Rule 56 does not require all information on a topic, just that which is material. The mountain of information must be managed and balanced to provide a sufficient amount of information which does not prohibit patent filing by being too expensive.
In fact, this increase in average number of citation per patent may even be an indicator of LACK of innovation in an area. It would seem that true pioneering patents would be uniquely short on citation of prior art. Declining rates of citation or classes of very low rates of citations may be the hot beds of invention.
 |
 |
It should be noted that outliers have not been removed from these data. In small classes such as D01, outliers have a large effect. However, once the classes were totaled, the effect of outliers appears to have been minimized.
 |
 |
Interestingly, the average percent of non-patent references appears to have remained relatively constant over the period studied, varying in the range of 12 - 14%.
The Aharonian articles indicate that the average number of non-patent references in software patents averaged 3.1. If it assumed that the % of non-patent prior references for the software classes are consistent with the data studied here, the average number of prior art references cited in software patents on average is about 22 references. Compared to the patents studied here, that is an incredible amount - and the software patent writers should be commended.
Also, the data indicate that plant patents cite 20-30% non-patent references. Apparently the plant patent writers are hanging around with the Caterpillar folks named in the Aharonian article - and cranking out non-patent references in a most fertile manner.
The Classification Manual for the PTO was reviewed and classes for analysis were selected. Classes 365, D01, and a plant class were selected for analysis. The classes were chosen to be as widely divergent as possible. Due to the large size of the utility class compared to the other classes selected, additional classes were added. Design class D04 was added, and, due to the small size of all classes in the Plant category, all Plant classes were included for analysis. All patents for each class were accessed using LEXIS during June and July of 1998. Printouts of the name, date of issue, patent number and references cited were obtained for each patent in the selected classes. Note that these data will suffer from any shortcomings associated with the database which is their source.
The numbers of patent and non-patent references were counted by hand and the data was entered into a spreadsheet. These data were then totaled for analysis.
In summary, the classes studied have demonstrated several interesting points. First, there has been a marked increase in the number of patents issuing annually. After a peak in the late 80's, the rate appears to have dropped of in the 90's, although more patents issue annually now than in the 70's and early to mid-80's. Second, the number of references cited on average has increased over time, from an average of 4 references per patent in the early 70's to an average of 9 per patent currently. Finally, the % of references cited which are based on non-patent prior art appears to have remained relatively constant over the last 27 years at a rate of about 12-14%.