On 27 March a judge in the Eastern District of Texas handed down a written order dismissing a patent infringement suit filed by patent troll, Uniloc against Rackspace, confirming that the patent involved, describes an invention that is purely a mathematical formula and so is not eligible for patent protection.
Procedurally the case is unusual, particularly for the plaintiff-friendly Eastern District of Texas, in that the case was dismissed prior to a full hearing. So naturally Rackspace have presented this as a significant victory. It surely was significant, as Uniloc alleged that this patent would have been infringed by the Linux operating system. But how reassured should we be? Does the case represent a sign that the US court system is starting to tackle the problem of dubious patents?
The “Invention”
The patent involved, US 5892697, relates to the way in which floating point numbers are represented. In computing, the floating point numbering system is analogous to the scientific notation for representing numbers. A number is represented in format having a sign (positive or negative), a body of significant bits (sometimes called a mantissa) and an exponent representing a power to which the body is raised. This representation is used because it is more memory efficient way for storing large numbers.
The invention purports to improve upon a standardized way of processing floating point numbers, as found in IEEE 754. Essentially, according to the standard, a floating point number is loaded into a memory register and arithmetic operations are carried out on all its fields, with the end result being rounded. The invention claimed in the patent was to round the numbers first before doing the necessary arithmetic operations.
It’s always helpful to consider the actual wording of the patent claims in cases like this. Claim 1 reads:
A method for processing floating-point numbers, each floating-point number having at least a sign portion, an exponent portion and a mantissa portion, comprising the steps of:
converting a floating-point number memory register representation to a floating-point register representation;
rounding the converted floating-point number;
performing an arithmetic computation upon said rounded number resulting in a new floating-point value;
converting the resulting new floating-point register value to a floating-point memory register representation.
The Decision
It has been judicially recognised that laws of nature, natural phenomena, or abstract ideas are in themselves not patent eligible subject matter in the US. Practice in the US has developed with respect to software related inventions has developed to establish a so-called “machine or transformation” test when considering the question of whether an invention should be considered as patent-eligible subject matter. According to this test, a process may be patentable if it “(1) is tied to a particular machine or apparatus, or (2) it transforms a particular article into a different state or thing.
The court held that claim 1 fails the “machine” test because it does not recite a machine – only steps for manipulating numbers and performing computations; and that it fails the “transformation” test because it was seen as relating to “mere manipulation” of the data which does not result in a “meaningful” transformation.
The court went on to consider whether claim 1 recites a mathematical formula and therefore should be excluded from patentability as relating to a “law of nature”. They held that claim 1 uses numeric conversions and arithmetic operations in a prescribed procedure to solve a mathematical problem, and is therefore an unpatentable mathematical formula. They also noted that “the conversion of floating-point numbers has applications across fields as diverse as science, math, communications, security, graphics, and games. Thus, a patent on Claim 1 would cover vast end uses, impeding the onward march of science”.
Analysis
It seems the patent would not have covered the methods mentioned in IEEE 754; but Uniloc were alleging that the Linux OS makes use of the improved method described in the patent. That factual question was not decided upon in court, but the judge is right when he said that the patent would have covered a vast array of end uses. So, the Linux community can breathe a real sigh of relief that the threat of this patent has been removed (subject of course to any available appeals).
However, an incredibly large number of patents of dubious quality or broad scope remain in force, and the patent troll industry is thriving. Apparently Uniloc alone have dozens of other patent suits pending, and in his blog post, Rackspace’s general counsel mentions that their patent litigation costs have recently become significant. So, this patent may unfortunately be a drop in the ocean.
From my experience of prosecuting patents in the US, I wonder if the outcome might have been different if the claims had been worded in a different fashion. For example, claim 1 refers to “converting a floating-point number memory register representation to a floating-point register representation”. This is expressed in general terms. However, if the claim had been given a bit more structure, for example, if it recited something like “number conversion processing circuitry” and had a step of “operating said number conversion processing circuitry to convert said memory register representation to a floating-point register representation”; then I think the court may not have thrown out the case with such alacrity.
Further, there was some odd “double think” logic employed by Uniloc. There is an historic case (Benson) that was thrown out as being a mathematical formula because it included mathematical operations such as shifting and adding. Uniloc argued that just reciting an “arithmetic computation” meant that no specific mathematical formula was taught. I found this line of argument frankly nonsensical and thankfully the court did not get caught out by it. But it illustrates the difficulty that there is in these types of cases in rising above the semantics of the patent language and focussing on the actual technical merit of the claimed contribution. If this had proceeded to a full jury trial, who knows how the jury might have reacted.
So what does this mean for companies in Scotland and beyond? Well, the good news is that the practice of the UK and European Patent Offices is a good deal stricter (although still far from perfect!). These offices require a “technical effect” to be present before considering an invention as patent-eligible and so this cuts out a lot of the dubious patents that get granted in the US. If you look at all the media stories about rubbish patents and vicious patent trolls, you will see that the vast majority concern the US.
However the fact remains that the US is still a prime market. Indeed, when I speak to clients about their international ambitions, the US almost always figures. At a pragmatic level, a company should understand the basics of the IP system and assess the risks involved. It pains me to say it, but when a patent troll pops up it can often be wiser to simply settle quickly and move on. Perhaps this means patent licence fees should be budgeted for if problems are anticipated.
And of course, filing your own patents should also be considered. This does not in itself give you freedom to operate – a patent gives the negative right to stop the activities of others, not the positive right to carry out any activity yourself – but in a country where patents are so much part and parcel of everyday business it makes sense to come into the market with some weapons of your own.