UK Patent Office "Technical Contribution" Workshop - London 2005
In February 2005 the UK Patent Office announced it would be holding workshops to answer the question What is a "technical contribution"?. Software developers, academics and patent professionals were invited to attend the workshops to contribute their views.
What follows is my record of the workshop held at the Law Society building in Chancery Lane, London on 7th April 2005. The author is Stuart Cunningham and he may be contacted here.
Other coverage of UKPO workshop events has been made:
|The Law Society, Chancery Lane, London|
After collecting my name badge and a cup of tea I attempted to mingle. A well-spoken gentleman introduced himself as Peter Hayward, a Divisional Director of The Patent Office. I introduced myself as a software developer with a large British company of more than 30,000 employees and was eager to ask him a few questions I hoped he would answer.
I wanted to know whether patents approved by the European Patent Office, such as those listed in the FFII's Software Patent Horror Gallery, were enforcable in the UK. He answered that it depended upon whether the application had applied for protection in one country or several and that if the applicant had paid the relevant fees, the patent was enforcable in the UK even though the UKPO's own examiners had not examined the application.
Peter Hayward did express some reluctance to use the term "software patents". I pointed out that this was the language used by the EPO and that the EPO included the terms patenting of software "as such" in their literature. He replied that it was an unfortunate term and that he would prefer not to use it.
I asked what I thought was an innocent question along the lines "Do you think it would be fairer for the Patent Office to be funded directly by the government regardless of how many patents were granted?"
I was surprised that this elicited a strong response from Peter, saying that this was an unfounded criticism he had been hearing frequently of late and that there was absolutely no incentive for the UKPO to grant more patents than should be granted on their merits. He added that they were only now coming out from under a large backlog of patent applications and he wished there were fewer applications rather than more. He also said that it was the funding model used by many other patent offices. I read later that the USPTO was funded by the tax payer until its funding model was changed in the 1990s after which it became a "profit centre" making more money from fees than it cost to run the office.
He was shortly to contradict his response, however, when I recounted an anecdote about a colleague's patent application. My colleague did not wish to have his patent granted, but rather apply and have it published so that it would exist as prior art, then abandon it. The advantage being that the application fee was small, around £200 instead of about £10,000 for a granted patent, yet the prior art was out in the open and would prevent competitors blocking his use of the technology in question. Peter replied that this approach was widely used and that IBM had their own publication for this purpose (IBM Technical Disclosure Bulletin). He added with a smile that if everyone did this (instead of having patents granted) the UKPO's funding would be at risk.
My next question was about the difficulty of reading patents. I recounted how two of my colleagues familiar with filing patents, both with PhDs and more than 15 years experience in their areas of expertise, were unable to identify their inventions from the text of their granted patents. Here, Peter Hayward sympathised, saying that the language used was due to the history of case law. I suggested that for computer implemented inventions the clarity and precision would be greatly improved by using one or more of the follow forms of expression:
I asserted that software developers would be familiar with at least one of these options and would find it far easier to search the patent databases. Peter said that the situation is unlikely to change since the existing language is what patent attorneys use and is established in case law.
Another point I raised was how apparently obvious methods were allowed to be patented, citing an example from the FFII Patent Horror Gallery. I said that in a given narrow field of computer science there may be only a thousand experts in the world and asked what the chances were that the UKPO have an examiner in that select few. "I'm not saying there are no bad patents" replied Peter and added that they have around 200 examiners who must cover all technical fields. "Their jobs must involve an awful lot of reading of journals and other material" I offered and Peter agreed that reading relevant journals was indeed one of their duties.
I asked whether the process of overturning a patent in the UK was as difficult and expensive as it is in the US. Peter didn't answer this question but instead asserted that the situation is far better here than in the US since there is a significant period between publication and the granting of a patent where prior art can be brought to the attention of the Patent Office. "It costs you nothing" he continued, explaining that all you need to do is write to the Patent Office pointing out the prior art.
As the workshop was called to order I was left with the impression from Peter Hayward that the UKPO felt they were doing a very good job with limited resources but that change was not welcome, while the mention of the word "reform" evoked hostility.
Once we were seated at our colour-coded tables, Sean Dennehey, Director of Patents, introduced himself and the two other facilitators; Jeremy Philpott and Peter Hayward.
There was a total of 28 participants arranged on 6 tables with either 4 or 5 people on each. Out of the 28, only 3 were women.
Sean made it clear that the scope of the workshop was narrow and to concentrate on the definition of "technical contribution". Our task was to assess four different definitions of "technical contribution" by applying them to five ficticious patent applications (called Case Studies). We were to decide for each case study whether it would be patentable under each of the four definitions and record whether our decision was unanimous within our group. We were told not to get diverted by the obviousness of the patent claims, but to pretend that they were fictional inventions which were truly novel and not obvious, since the purpose of the workshop was to concentrate on the four "technical contribution" definitions being tested, labelled A, B, G and K:
A "Technical Contribution" means a contribution to the state of the art in a field of technology which is new and not obvious to a person skilled in the art. The technical contribution shall be assessed by consideration of the difference between the state of the art and the scope of the patent claim considered as a whole, which must comprise technical features, irrespective of whether or not these are accompanied by non-technical features. B "Technical Contribution" means a contribution made by a claimed invention, considered as a whole, to the state of the art in a field of technology. "Technical" means belonging to a field of technology. New teaching about the use of controllable forces of nature under the control of a computer program, beyond the implementation of the data processing procedure itself, is technical. The processing, handling, representation and presentation of information by a computer program is not technical, even where technical devices are employed for such purposes. G "Technical Contribution" should be applied to a putative invention when considered in its entirety if that claimed invention is concerned with new teaching in controlling forces of nature in a field of technology. The processing, handling or presentation of information by a computer program -- or abstract representations of programs and algorithms -- should not be considered technical, even when that processing, handling or presentation of information is part of a larger putative invention that makes a technical contribution, and even when technical devices are used in the processing, handling and presentation of information. K "Technical Contribution" means a contribution made by a claimed invention, considered as a whole, to the state of the art in and limited to a declared field of technology. "computer-implemented invention" means any invention the performance of which involves the use of a computer, computer network or other programmable apparatus and having one or more prima facie novel features which are realised wholly or partly by means of a computer program or computer programs. The processing, handling, and presentation of information by a computer program is not technical, even where technical devices are employed for such purposes. Implementation of a software solution on a platform or architecture not itself supported by the claimed invention is a separate contribution and is excluded from the claimed invention.
The full text of all case studies used in the workshops is available from the UKPO site.
As we started on the first case study it became clear what the breakdown of the group at my table was: two software developers, two IT managers or IT executives and one patent attorney (who was an engineer in a previous life).
This case study was typical of what many software developers would call a pure software patent. The claim was for an "apparatus for digitally enhancing a photograph taken in low light conditions by applying a digital filter..." where the apparatus consisted of a means for "creating a two dimensional array representing a photograph", applying an "operator matrix" to enhance low-light photographs, and a way to output the result of the "operator matrix". There was no mention of any novel CCD technology or display technology.
Since we felt that definition A was vague about what constituted a "technical feature", we were unanimous that this invention would be patentable under definition A.
Definition B made us think about the terms "forces of nature" and "processing, handling, representation and presentation of information" present in that definition. Initially there were different opinions on "use of controllable forces of nature". I felt the term was unambiguous since there are four fundamental forces of nature:
Once I explained that I believed "forces of nature" was intended to indicate any interaction with the physical world, since that can only be done using one or more of the four forces, our group agreed on that interpretation.
Someone then pointed out that any computer uses the forces of nature to move electrons around in semiconductors. I replied, "yes, but those patents have expired now", and explained that such physical interaction is only relevant if it is novel and being claimed in the invention. Case study 4 wasn't claiming a new way of moving electrons, or any other physical interaction with the world, so there was nothing "technical" about this invention on the "forces of nature" grounds.
In any case we agreed that the "information processing" clause of definition B excluded this invention. Definition B says that information processing is not technical, and we all agreed that case study 4 was purely information processing and so were unanimous in rejecting it under definition B. Definition G was very similar to definition B but with a messier rewording we felt. Eventually we agreed that the reasoning we used for definition B applied equally for G and rejected this case study under definition G. Definition K had some new language but also mentioned the "information processing is not technical" clause, so we all excluded the invention under definition K.
This invention concerned software "tools" running in virtual machines. The claim was for an application programming interface (API) which comprised "means for receiving parameters from an application" and "means for instantiating a tool object". Since definition A was so permissive we agreed A would allow this invention.
Initially we all felt that this invention was purely "processing, handling, representation and presentation of information" so therefore not patentable. Then one member of the group said he wasn't so sure and suggested that "instantiating a tool object" may lie outside the "information processing" scope. A discussion of just what information is ensued, diverting into areas such as entropy and information conservation of black holes, before being brought back to earth with our working definition of information being "anything you can store in a computer memory".
Then we were all able to agree that Definition B and G excluded this invention from patentability. Definition K made us feel uneasy since it discussed a "computer-implemented-invention" and a "software solution" in imprecise terms, muddying the water unnecessarily. On balance we agreed it probably would deny this invention.
This was the first invention where we did not reach concensus. The claim was "a method of manufacturing an integrated chip to perform a particular calculation". It consisted of an algorithm to process a specification of a mathematical function to produce a chip floor plan as the result. The last step of the method was "manufacturing the chip by using the chip floor plan" and it was this step which caused the dispute.
Without that last step we all agreed that it was a purely "information processing" patent without use of "forces of nature". Half of us argued that "manufacturing a chip from a chip floor plan" was an invalid claim since the invention did not claim a new way of manufacturing a chip from a floor plan, it only claimed a novel way of producing the floor plan. The others argued that the whole point of producing the floor plan was to make the chip, so the actual manufacture of the chip was made possible by the "information processing" steps in the invention. After several attempts we could not resolve this argument so we recorded half of us allowing the invention under all definitions and half denying it under definitions B, G and K.
This was very similar to case study 10 since it described a interface between two software programs. Since we had hammered out how to apply the definitions to no. 10 we agreed that a similar analysis applied: definition A allowed it, while B, G and K did not. This invention described an algorithm and API between two networked online game programs, the object being to reduce the time delay for a two-player game such as Backgammon. One member of our group made what I felt was an appropriate comment "It won't work".
This invention was a method of dynamically allocating radio frequency bandwidth using an algorithm to optimise efficient use of the bandwidth. The context was mobile phone telephony where demand for bandwidth varied within a cell and across a number of cells.
We did not reach a concensus for this case study. The contention arose from whether "dynamically allocating bandwidth" involved the use of the forces of nature or not. Half of us argued that the invention did not describe a novel way to interact with physical forces to change the output of antennas - the case study implied that changing bandwidth per se was prior art. The fact that the decision of how to adjust the bandwidth was arrived at by a novel algorithm was irrelevant, it was argued. The others argued that although bandwidth allocation had been done before, this invention was a new way to, eventually, interact with physical forces to change the radio emissions.
Time was running out so half of us recorded patentable under all definitions and half denied it under definitions B, G and K.
After one hour had elapsed our decisions on patentable versus unpatentable were collected and tallied on four large sheets at one end of the room.
Sean Dennehey went around each table, asking for a volunteer to summarise how the table felt about each definition and to add any other comments. It became clear that other tables had had very similar arguments as our own. Case Studies 11 and 13 had made people discuss in depth what "forces of nature" and "information processing" meant in definitions B, G and K.
At least two tables became bogged down on the question of hardware implementation versus software implementation which definition K invited by mentioning a "software solution". One table had argued over whether an FPGA was hardware or software. Our table had avoided this discussion, probably because we agreed early on what "information processing" was. For us it didn't matter whether a general purpose computer was executing an algorithm or whether a custom piece of hardware computed the same algorithm. Either way it was purely "information processing".
By the time table 4 was reached, definition B was in the lead as the favourite definition, with A being the worst. Table 4 consisted of four expensive suit wearing patent attorneys. Since the organisers had arranged participants into groups at random, this prompted Jeremy Philpott to remark "So much for the laws of statistics". I found this remark jarring since most of us present would know from undergraduate mathematics that we would expect "clumps" in random selections, not uniformity. As the attorneys announced that definition A was their favourite definition, hoots of laughter erupted around the room. Seeming somewhat disconcerted by this reaction the spokesman for table 4 explained that he didn't think it was appropriate to attempt to define "technical contribution" since technology may change in the future and force us to keep changing our definition. He said that it should be from judgements and the case law was that the definition should evolve.
For our table I summarised our opinions as follows. Definition A was too vague and didn't add any information on what "technical" meant so we considered it a poor definition. One member of our table (not the patent attorney) suddenly recanted saying he liked definition A because it was vague. Unperturbed I continued, explaining definition B was not so much our favourite but was the "least poor" of the four definitions. Definitions G and K were similar but with less clear language we felt, especially the second and fourth paragraphs of K.
Sean ran a tight ship making sure every table had their fair say within the time constraints, even cutting of Jeremy Philpott during a followup question he was asking to table 4. After the last table Sean took a straw poll on whether we thought each case study represented a patentable invention:
|11||16||4 - lowlight photograph image enhancer|
|6||19||10 - application programming interface|
|17||8||11 - manufacturing chip by algorithm-generated floor plan|
|2||24||12 - time delay reduction in networked game|
|13||12||13 - dynamic bandwidth allocation using algorithm|
Upon a request from table 1, Sean took another poll this time on industry background.
|"Law-related"||1 (I noted that all four on table 4 abstained)|
Sean thanked all of us for participating and announced that the results of all the "technical contribution" workshops will be published after the May 2005 election.
After the workshop officially closed, a number of animated discussions took place. I could hear several complaining that the event was too narrow and didn't permit them to express wider views especially on the patentability of software "as such". I noticed that Jeremy Philpott had become surrounded by software developers pointing out examples of apparently absurd patents granted by the UKPO. At one point, a developer suggested that it would be helpful for granted patents to be accompanied by a summary of the examiner's qualifications or perhaps a resume. "I find that comment deeply offensive" replied Jeremy causing an uncomfortable silence in the assembled crowd. Shortly afterwards he was rescued by Sean Dennehey who announced that we must make way for the next workshop as there were several being held that day.
I was optimistic that the workshop could only serve to improve the current situation. One patent attorney remarked to me that he would prefer any improvement, such as definition B, since the current case law was full of contradictory judgements on the meaning of "technical" making his job very difficult. I felt that the granting of a legal monopoly to an individual or company was a grave step and required the establishment of clear boundaries. I learnt that the situation in the UK is better than the US but that the 20,000 or more software patents already granted in Europe are causing us to slide slowly towards the US. At least in the UK there is no well-funded effort to patent art as there is in the US (see www.patenting-art.com).
|Copyright Stuart Cunningham 2005.|
This work is licensed under a Creative Commons License.