Article I, Section 8, of the U.S. Constitution includes the following sentence about the patent system:
What follows is a statistical analysis of twenty four years of US patent bibliographic data, from which I argue that the patent system in practice is unconstitutional: i.e., it promotes marketing, which is not 'progress', and it is a registration process where no real determination of 'discoveries' is made, other than randomly in an environment where patent quantity is all that matters.
The intellectual corruption, then, is the publicly sanctioned but privately ridiculed fiction that the average patent is novel, unobvious, and enabled. It is time for everyone to stop pretending that the PTO has the resources and management capabilities to insure novelty, unobviousness, and enablement. Specifically I will argue that:
[NOTE 2: A more rigorous analysis would require access to the full text of patents, PAIR data, and random samplings of filewrappers. Due to the PTO's fraudulent overcharging for even just the fulltext data, $500,000 would be needed to acquire all of this data. In the near future, I hope to have time to approach a private foundation (or seriously interested company) to seek a few million dollars to do such a rigorous study. Any serious reform of the patent system will cost hundreds of millions of dollars, preceded by this multimillion dollar study of the current and historical workflow of the patent examination system. Anything less is a waste of time and an indication of a general lack of credibility in seeking real patent reform.]
All that said, here is the data and my comments.
Of interest is that over the last twenty years, the PTO has achieved a steady state issuance rate on the average of about 65% of the patents filed (I assume naively that it takes about two years for a patent to issue, so what is filed in 1997 issues in 1999), and that examiners issue about one patent a week. I argue that this is NOT a result of the PTO management policy, but IS the management policy of the PTO - a patent registration system with some minimal examination constrained by the examiners' two-point disposal system. It's why patents will always take over a year to issue - PTO management looks at patent applications in one year to figure out how many examiners they will need the next year. Given the inadequate upfront patent fees, PTO management can't but be running a registration system, since thorough examination is an unknown cost harder to budget for, as opposed to patent registration-like activities.
"% Patents to Large" measures the percentage of patents going to the largest 250 corporations lead by General Electric, IBM, Hitachi, Canon, Toshiba, Kodak, ATT, US Philips, Du Pont, Siemens and Motorola. That about half of all patents go to such large corporations explains why a) PTO management always talks about its customers and cares only about their views of quality, and b) why PTO management ignores everyone else. Why? When such companies are asked by Congress if they are happy, they will answer "yes" as long as the PTO is issuing quantity over quality to them. And as long as Congress hears "yes" and gets its campaign donations, it leaves the PTO alone, which makes the bureaucrats at the PTO satisfied, since Congress is the only outside influence with direct power over the PTO. It is a nice triangle that works as long as we have a patent registration system to support the needs of older-generation multibillion dollar corporations.
Last year's patent reform bill included amongst other things the creation of a Patent Public Advisory Committee (for which nominations just closed). Yet, with no effective budget to gather data (such as mine below) to verify whatever PTO management will present to the Committee (even if the large-company-dominated Committee cared to pursue issues of quality), it is unlikely this Committee will be a force for fostering patent quality. Not surprisingly, this same Congress that created this impotent patent advisory committee also underfunded studies of patent quality, for which the National Academy of Sciences is now seeking proposals, made worse by the NAS choosing as reviewers of these proposals the very same people who have ignored the issue of patent quality for too many years. So to patent litigators around the country, I say, do not worry in the least, these people are just going through the motions, and growth rates for patent litigation will continue unabated.
Number Number Patents per %Patents per %Patents %Patents
Year Patents Examiners Examiner Filed(Y-2) to Large Litigated
1976 70186 1150 61.0 68 0.94
1977 65205 870 74.9 64 1.19
1978 66084 930 71.1 64 1.15
1979 48839 850 57.5 48 1.38
1980 61812 870 71.0 61 1.37
1981 65766 800 82.2 65 1.20
1982 57878 820 70.6 55 1.25
1983 56862 1030 55.2 53 1.68
1984 67214 1330 50.5 61 1.38
1985 71668 1310 54.7 69 54 1.51
1986 70862 1310 54.1 63 54 1.39
1987 82961 1600 51.9 70 54 1.42
1988 77938 1640 47.5 63 53 1.16
1989 95565 2180 43.8 74 53 1.11
1990 90416 2270 39.8 64 53 1.16
1991 96550 2530 38.2 63 53 1.01
1992 97462 2440 39.9 59 53 0.89
1993 98356 2330 42.2 59 52 0.81
1994 101687 2400 42.4 58 52 0.85
1995 101430 2470 41.1 58 51 0.67
1996 109651 2420 45.3 57 50 0.70
1997 112019 2440 45.9 52 48 0.56
1998 147576 3210 46.0 75 47 0.36
1999 153588 3730 41.2 71 0.13
The lower numbers for the Patents Litigated column for the 1990s just means that there are large numbers of lawsuits yet to be filed, say 1/2% of the 1,000,000 patents issued in the latter part of the 1990s, i.e., about 5000 lawsuits (and patent busts) yet to come (many of which will be over crappy software patents - I hope the PTO doesn't change a damn thing :-).
With regards to 102/103 patent quality, over the long term, patent examination has mostly ignored the non-patent literature, with slightly less than half of all patents citing no non-patent prior art, and the average patent citing about two non-patent prior art items (while the length of literature bibliographies [such as the IEEE] are in the many dozens and growing). And, in citing less than a dozen prior patents, patent examination considers less than all relevant patents. In short, insufficient amount of prior art is being provided to justify patents being given a presumption of validity with regards to novelty, obviousness and enablement - the courts should stop respecting this intellectually unsupportable belief.
Indeed, the column (MechElec)/NewINSPEC is the ratio of new Mechanical and Electrical patents to new articles in the INSPEC abstract, one of the few databases that abstract most scientific and engineering conferences and journals outside the CAS/MEDLINE world. The ratio seems to indicate that the PTO thinks that about 1/3 of everything published in any one year is novel and unobvious. That is nonsensical.
Additionally, with patents citing few if any non-patent prior art, and non-patent prior art rarely if ever including patent citations, there is almost a complete disconnect between patent literature and non-patent literature. Such a disconnect makes it hard to accept that patent literature inspires and educates, since scientists and engineers in their real publications make little if any mention of patents as references. This is not surprising to patent searchers, who are asked to find patent references after the research and development has been done and the patent application is being prepared (to be augmented by patent references made by the patent examiner). In short, patent citations are extremely unreliable measures of information flows and patent concept maps fairly useless in many fields for determining key players. Any correlations are not causal, because there is no understanding of the statistics of who is actually citing those patents that are cited.
Part of this disconnect was due to the Patent Office's apathy throughout the 1980s and 1990s to provide convenient access to the public to patent abstract and fulltext data. How can scientists and engineers search for inspirational patent data they don't have access to? They couldn't and didn't. So that part of the Constitution where the public gives patent rights in order to get disclosed information that they can learn from hasn't been working too well. And between government funding of R&D (such as the DoD/DoE/NASA/NIH) and altruistic efforts (open source movement, public consortia to sequence the human genome), and even more simply the tens of millions of scientists and engineers around the world who invent if for no other reason than to overcome boredom, it is hard to accept that what drives much inventing is the constitutionally provided patent right. Rather, patents are driving the marketing of invented technologies, which, while beneficial and a legitimate reason to provide a property right, is NOT the property right specified by the Constitution - in short, the patent system is no longer constitutional.
Average number of patent and non-patent prior art references
(MechElec) Articles
%Zero IEEE ----------- citing
Year Patent Non-Patent Non-Pats BibLen New INSPEC patents
1976 4.81 0.34 76.5 14.2 0.35 <1%
1977 4.94 0.38 75.0 17.6 0.34 <1%
1978 4.98 0.44 71.6 14.1 0.34 <1%
1979 5.11 0.50 69.6 16.2 0.26 <1%
1980 5.51 0.60 66.4 18.4 0.33 <1%
1981 5.76 0.72 63.6 19.7 0.31 <1%
1982 5.90 0.84 61.2 18.7 0.26 <1%
1983 5.98 0.89 59.7 25.9 0.24 <1%
1984 6.10 0.97 57.5 19.1 0.26 <1%
1985 6.39 1.09 55.0 25.4 0.27 <1%
1986 6.48 1.12 53.9 21.0 0.26 <1%
1987 6.74 1.23 52.0 25.2 0.29 <1%
1988 6.99 1.37 49.8 29.3 0.25 <1%
1989 7.37 1.50 49.3 26.8 0.31 <1%
1990 7.37 1.53 49.2 26.0 0.28 <1%
1991 7.43 1.58 48.9 27.0 0.29 <1%
1992 7.81 1.67 47.8 28.0 0.27 <1%
1993 8.17 1.83 46.0 29.0 0.27 <1%
1994 8.72 1.99 44.9 30.0 0.27 <1%
1995 9.29 2.14 44.3 31.0 0.25 <1%
1996 9.81 2.26 44.4 32.0 0.26 <1%
1997 10.30 2.37 44.9 33.0 0.24 <1%
1998 10.52 2.40 47.3 34.0 0.33 <1%
1999 10.72 2.39 48.8 35.0 0.32 <1%
A measure of the decreasing quality is the decreasing number of months spent per claims measured by dividing the number of claims by the time from patent filing to the patent appearing in the Official Gazette. (For the most part, these pendencies are only for non-divisional, non-continued patent applications - determining effective pendencies of things like CIPs is always subjective.) The drop is due to more claims being sought, as average pendency hasn't changed much over the decades (again, PTO registration system policy). Thus given the low amounts of non-patent prior art cited (and missed prior patents) as a percentage of all available non-patent prior art, and the decreasing time spent per claim, it is not unreasonable to conclude that patent quality is dropping as the PTO is overwhelmed with patent applications. (NOTE: I assume that the ratio of independent claims to dependent claims has remained constant over the decades, so that I can then divide the total claims by pendency months to get an "average" amount of time spent per claim. Only analyzing full-text patent data can verify this assumption.)
Average pendency(mon.) Average Number Claims Months/Claim
Year Mech Chem Elec Mech Chem Elec Mech Chem Elec
1976 19.8 24.0 22.5 9.1 9.6 9.9 2.18 2.50 2.28
1977 19.4 22.5 21.0 9.2 10.2 10.1 2.12 2.20 2.08
1978 20.0 21.7 21.5 9.5 11.3 10.4 2.11 1.93 2.06
1979 21.5 21.6 23.0 9.6 11.3 10.5 2.24 1.91 2.19
1980 24.2 24.1 27.5 9.7 11.3 10.6 2.51 2.13 2.58
1981 24.4 22.9 26.0 9.7 11.4 10.5 2.51 2.01 2.46
1982 25.9 23.0 27.5 9.9 11.2 10.4 2.62 2.06 2.63
1983 27.1 23.6 28.6 9.9 11.5 10.6 2.75 2.05 2.70
1984 26.7 23.3 28.6 10.0 11.8 10.7 2.67 1.96 2.66
1985 24.9 23.0 28.3 10.3 12.2 11.0 2.42 1.89 2.58
1986 23.3 22.5 27.7 10.5 12.5 11.1 2.22 1.80 2.48
1987 21.5 23.5 25.3 10.7 12.6 11.4 2.00 1.86 2.23
1988 20.7 23.0 23.2 11.0 12.8 11.7 1.88 1.79 1.98
1989 19.7 22.5 22.1 11.4 12.9 12.6 1.73 1.75 1.76
1990 18.7 22.2 21.4 11.7 12.7 12.9 1.60 1.75 1.66
1991 19.1 22.5 21.6 12.1 12.9 13.2 1.58 1.74 1.64
1992 18.7 22.5 22.4 12.2 13.0 13.7 1.53 1.73 1.64
1993 18.6 22.3 23.5 12.4 12.8 13.8 1.50 1.74 1.70
1994 19.2 21.7 23.4 12.7 13.0 14.2 1.51 1.66 1.65
1995 19.6 21.6 23.1 13.1 13.3 14.6 1.50 1.62 1.58
1996 21.0 22.6 24.0 13.4 13.7 15.3 1.57 1.65 1.57
1997 22.4 25.5 25.5 13.7 14.6 15.8 1.63 1.75 1.61
1998 23.2 26.5 26.4 14.2 15.3 16.6 1.63 1.73 1.60
1999 23.8 27.6 27.5 14.6 16.0 17.1 1.64 1.73 1.61
Again I ask, why in the last twenty years has the PTO not published the above three sets of data (or something similar)? If I can do it with a PC, why can't the PTO do it with its mainframes and statistical analysis packages? Such data can be calculated both historically, and in realtime, for example, the number of months per claim for patents issuing in the month of April 2000, all easily presented on the PTO web pages. And the same question for the AIPLA, ABA, IPO, and IP law schools. How come none of these "professional" organizations have been and are publishing such data? I pose the same question to the IP magazines and newspapers, publications that get lots of advertising dollars from members of the AIPLA, ABA and IPO.
But the most blame is directed towards the PTO, with its Office of Patent Quality Review, Center for Quality Services, Quality Assurance Specialists, and office for statistical analysis. What numbers do they publish each year? Patent quantity numbers (who got the most, how many by each state, how many went to large corporations, etc.) Nothing relevant to patent quality, so their use of "Quality" should honestly be translated as "Quantity" - Office of Patent Quantity Review, Center for Quantity Services and Quantity Assurance Specialists. It is totally unprofessional for the PTO not to be publishing these statistics, and more detailed statistics based on the full-text data - it has no credibility when it comes to patent quality as defined by those skilled-in-the-art.
Why not? Because the patent bar and patent law schools have neglected the issue of patent quality over the last thirty years. I did a survey of the articles in three patent-law-rich publications, IDEA (the IP journal of the Franklin Pierce Law Center), JAIPLA (Journal of the American Intellectual Property Law Association), and JPTOS (Journal of the Patent and Trademark Office Society), from 1976 to 1999. Of approximately 1700 articles over this twenty-four year time period, only about two percent dealt with issues of patent quality (such as problems with prior art, problems with examiners, problems with checking for enablement, problems in claim analysis). It is clear then the apathy of the IP law schools and patent bar to this issue. Two of the very few critical articles appeared in JPTOS - "Is Rule 56/57 leading us to a registration system? Can automation help?" (Dec 1988 JPTOS), and "Is the United States automating a patent registration system for software? A critical review of information management in the USPTO" (Sep 90 JPTOS).
One explanation for the silence of the IP schools is due, indirectly, to Prof. John Barton of Stanford Law School, who in an article in Science magazine, 17 March 2000, page 1933, argues for patent reform. In the middle of the article is a graph of the number of patent lawyers (measured using ABA data) divided by R&D expenditures in billions of dollars, as measured by the NSF. The first sentence of his article? "The number of intellectual property lawyers in the United States is growing faster than the amount of research." No wonder IP schools too busy turning out IP lawyers don't have enough time to do critical studies of the patent system.
The Sloppy Claims column is a ancedotal estimation of the percentage of claims each year that are sloppily issued, by which I mean patents where one or more of the following conditions are true: the antecedent basis is violated, the issued abstract/claims have one or more spelling errors, uses claim language that explicitly violate one or more MPEP rules (like using "for example" in a claim, or use "means for" only once), or where too many "means" clauses are being used, or where the independent claim is excessively long (usually drafted by pro-se inventors who don't realize such claims are easily circumvented). These are sloppy patents in that the errors are unambiguous and semantic free, easily checked with simple computer programs automatically (I have a prototype of such a program at my Web site, which only needs convenient access to the fulltext data to become a real effective tool, a tool which sadly infringes a pathetic patent issued to Aurigin which claims, get this, automatic antecedent basis checking). If I have such a program, it should be quite trivial for the PTO to prepare such a program, and automatically apply it to all issued patents before the Notice of Allowment is sent out. The PTO doesn't, yet another measure of their apathy towards patent quality.
And to be snide, little if any of all of this patent quality data and critical commentary has its parallel in the publications coming out of Yale Law School or the patent staff of Xerox Corporation. Their credentials are lacking when it comes to being involved with patent quality studies (especially Xerox, which came close to being sanctioned for Rule 56 violations for one of its patents).
Patent Lawyers
Percnt Percent Critical -------------- Sloppy
Year Jepson Law Articles R&D (billion$) Claims
1976 12.8 1% (1/90) 54 ~3%
1977 13.1 6% (4/71) 55 ~3%
1978 13.7 2% (1/62) 53 ~3%
1979 13.7 6% (4/69) 55 ~3%
1980 14.5 3% (2/58) 54 ~3%
1981 14.4 0% (0/70) 53 ~3%
1982 14.1 7% (4/60) 53 ~3%
1983 13.1 2% (4/68) 52 ~3%
1984 11.9 2% (1/54) 48 ~3%
1985 11.1 3% (2/65) 47 ~3%
1986 10.5 0% (0/53) 48 ~3%
1987 9.6 2% (1/64) 46 ~3%
1988 9.2 7% (6/83) 50 ~3%
1989 8.4 2% (2/80) 53 ~3%
1990 7.8 0% (0/73) 58 ~3%
1991 7.2 1% (1/72) 62 ~3%
1992 6.6 1% (1/83) 64 ~3%
1993 5.8 0% (0/86) 70 ~3%
1994 5.3 1% (1/80) 73 ~3%
1995 4.6 1% (1/90) 75 ~3%
1996 4.1 0% (0/83) 72 ~3%
1997 3.7 1% (1/72) 75 ~3%
1998 3.1 1% (1/68) ~3%
1999 2.8 1% (1/67) ~3%
(39/1721)
I find the interest of the intellectual property groups of accounting and management firms just as bad. You would think that people trained in quantitative analysis would be applying such skills to patent data to better inform their clients. But you will find little if any critical quantitative analysis in the publications of these firms, even something as minimal informative as the low probability of finding a patent "Rembrandt in the attic", as opposed to the higher probably of finding a patent "pipe bomb in the neighbor's basement". That's the advice corporate clients need.
The following table shows the percentage of patents examiner by primary examiners solely, around 60% or so for mechanical and chemical patents, around 50% or so for electrical patents, reflecting the greater turnover of examiners in those arts. Again, the minimally fluctuating averages in the 1990s is more evidence of PTO management trying to maintain steady rates of patent issuances as a percentage of filed applications.
Patents examined only by Primary Examiners
% % % # # #
Year ME CH EL Mchncl Chemcl Electl
1976 52 59 67 17529 15172 7751
1977 59 66 73 19421 13793 8832
1978 65 72 73 21490 15638 8444
1979 71 76 74 17834 11939 6120
1980 76 80 76 24384 15792 7948
1981 78 82 78 25443 18292 8878
1982 78 84 79 21766 15617 9267
1983 70 81 77 19645 15020 8317
1984 62 76 63 21465 15496 8190
1985 60 72 57 21905 15113 8417
1986 57 72 53 20939 13764 8215
1987 55 68 50 23505 14759 9792
1988 57 64 46 21719 14056 8694
1989 59 61 45 26577 16928 10840
1990 61 60 50 26248 15888 11005
1991 56 52 48 25618 14529 11383
1992 54 46 47 23627 13536 11912
1993 54 47 47 21425 15031 13008
1994 56 49 45 23585 14131 14243
1995 61 54 44 24136 16137 14322
1996 66 63 46 27280 19700 17165
1997 67 68 48 26682 24081 18218
1998 69 70 52 35075 28234 29997
1999 60 68 53 34038 29127 29082
Data on the distribution of patents by general type. Over the decades, about a third of the patents have been chemical, with little change. But while mechanical patents used to dominate (about 50% in the 1970s and early 1980s), and electrical patents were 20%, they are both now about equal in their percentages, reflecting the growing importance of electronics and computing and diminishing importance/value of mechanical inventions (New Economy versus Old Economy).
Number % % % Number Number Number
Year Patent ME CH EL Mchncl Chemcl Electl
1976 70186 47 36 16 33414 25299 11473
1977 65205 49 31 18 32491 20713 12001
1978 66084 49 32 17 32938 21623 11523
1979 48839 51 32 16 25035 15636 8168
1980 61812 51 31 16 31859 19552 10401
1981 65766 49 33 17 32346 22063 11357
1982 57878 48 31 20 27844 18402 11632
1983 56862 48 32 18 27777 18352 10733
1984 67214 50 30 19 34168 20172 12874
1985 71668 50 28 20 36196 20776 14696
1986 70862 51 26 21 36543 19024 15295
1987 82961 50 25 23 42130 21447 19384
1988 77938 48 28 23 37480 21909 18549
1989 95565 46 28 24 44540 27359 23666
1990 90416 46 28 24 42480 26112 21824
1991 96550 46 28 24 45292 27616 23642
1992 97462 44 29 25 43450 28820 25192
1993 98356 39 31 28 39294 31423 27639
1994 101687 40 28 30 41635 28812 31240
1995 101430 38 28 32 39526 29406 32498
1996 109651 37 28 34 41241 31103 37307
1997 112019 35 31 33 39448 35039 37532
1998 147576 34 27 38 50365 40261 56950
1999 153588 36 27 35 56207 42532 54849
Corporations (large and small) continue to be assigned the vast majority of patents, over 90% of the electrical and chemical patents, but, reflecting that independent inventors are more likely to be mechanical tinkerers, only about 70% of the mechanical patents. Given that PTO application fees are a fraction of what applicants pay their lawyers, that most applicants work for corporations, and that the concept of a small entity beyond a few individuals is meaningless (is a Web startup of 30 employees with a billion dollar valuation a small entity?), why not get rid of the small entity distinction? It wouldn't affect the economy, probably save some individual inventors from wasting their money, and simplify PTO operations.
Patents assigned to Corporations
% % % # # #
Year ME CH EL Mchncl Chemcl Electl
1976 70 91 89 23592 23113 10218
1977 68 92 88 22186 19185 10611
1978 68 92 87 22571 19984 10121
1979 68 92 87 17272 14395 7180
1980 68 91 87 21780 17852 9110
1981 68 91 87 22247 20166 9954
1982 69 91 88 19468 16901 10261
1983 72 92 90 20263 16976 9680
1984 73 93 90 25148 18796 11669
1985 73 93 91 26749 19353 13460
1986 72 93 91 26583 17693 14033
1987 72 92 92 30729 19937 17846
1988 71 93 91 26978 20425 16964
1989 70 93 91 31440 25452 21555
1990 70 92 91 29906 24266 19901
1991 70 93 91 31896 25778 21660
1992 71 93 92 31045 26860 23200
1993 70 93 92 27767 29450 25561
1994 70 93 93 29261 26980 29091
1995 69 92 92 27623 27296 30081
1996 69 92 92 28810 28901 34678
1997 70 93 93 27774 32662 34905
1998 70 93 93 35501 37562 53234
1999 72 92 93 40931 39549 51480
More detailed data on prior art references. Insufficient number of foreign patents (particularly from EPO/PCT/JPO) are being cited. Electronic patents have and continue to do worst at citing non-patent prior art, though all three technology areas have inadequate levels.
Average number of prior art references
Mechanical Chemical Electrical IEEE
Year USA FOR NON USA FOR NON USA FOR NON BibLen
1976 5.40 0.07 0.51 4.42 0.46 0.36 4.62 0.19 0.16 14.2
1977 5.48 0.07 0.55 4.59 0.50 0.39 4.75 0.20 0.19 17.6
1978 5.53 0.08 0.64 4.77 0.55 0.46 4.63 0.22 0.23 14.1
1979 5.59 0.10 0.67 5.00 0.62 0.55 4.74 0.27 0.30 16.2
1980 6.02 0.13 0.77 5.30 0.72 0.67 5.19 0.35 0.34 18.4
1981 6.28 0.16 0.88 5.65 0.83 0.84 5.36 0.42 0.43 19.7
1982 6.55 0.20 1.01 5.67 0.94 0.94 5.49 0.46 0.57 18.7
1983 6.56 0.23 1.07 5.89 0.93 0.99 5.49 0.52 0.62 25.9
1984 6.82 0.25 1.16 5.90 1.05 1.04 5.59 0.59 0.70 19.1
1985 7.03 0.26 1.29 6.21 1.15 1.14 5.93 0.67 0.83 25.4
1986 7.15 0.28 1.36 6.27 1.28 1.16 6.00 0.71 0.85 21.0
1987 7.52 0.32 1.49 6.46 1.46 1.28 6.24 0.72 0.93 25.2
1988 7.85 0.35 1.63 6.66 1.58 1.45 6.46 0.78 1.02 29.3
1989 8.29 0.39 1.68 6.93 1.83 1.67 6.88 0.91 1.15 26.8
1990 8.34 0.43 1.68 6.86 1.95 1.75 6.90 0.97 1.16 26.0
1991 8.59 0.42 1.71 6.86 2.13 1.83 6.85 0.94 1.20 27.0
1992 9.07 0.50 1.80 7.04 2.33 1.91 7.32 1.03 1.30 28.0
1993 9.72 0.49 1.89 7.33 2.88 2.20 7.47 1.16 1.41 29.0
1994 10.36 0.59 2.00 8.03 3.14 2.47 7.77 1.38 1.52 30.0
1995 11.01 0.68 2.10 8.53 3.69 2.71 8.35 1.48 1.62 31.0
1996 11.25 0.70 2.10 9.21 4.75 2.97 8.95 1.63 1.70 32.0
1997 11.87 0.79 2.11 9.60 6.24 3.35 9.42 1.87 1.66 33.0
1998 12.16 0.87 2.14 9.57 7.05 3.36 9.83 1.99 1.71 34.0
1999 12.48 0.91 2.06 9.72 6.90 3.43 9.96 1.91 1.67 35.0
The conclusions above are most pertinent to Electronics patents (which should be presumed to be invalid), which is my area of expertise, a technology area where patent examination is far worse off than for Chemical or Mechanical patents. Also, I suspect these statistics will hold true for the European and Japanese Patent Offices if their data were similarly analyzed, especially as their volumes and practices converge to those of the United States. In short, no patent office around the world has really ever mastered the process of large scale patent examination.