Deutsche Version

The MP3 Patents

The MP3 patents are frequently given as examples for “positive” software patents. The reasoning is that the extraordinary effort behind this innovation deserves protection by a patent.

This article takes the MP3 patents as an example how software patents harm our economy. In doing so we will see that the “extraordinary effort” which “deserves protection” is not real in the MP3 case. In fact, when it comes to broadness and triviality, the MP3 patents are on the same level as other software patents like the progress bar patent by IBM, the context menu patent by Philips, or the online gift patent by Amazon. On the contrary, the harm to our economy caused by the MP3 patents is even higher and clearly visible.

Protection for an Extensive Innovation?

When looking at the MP3 compression method one has, at first, reason to be awestruck by the enormous lot of work behind it. The audio-psychological perception which enables us to drop systematically parts of the acoustic information is most certainly based on many years of scientific work. It seems self-evident that this kind of effort would never have been financed without the prospect of a 20-year monopoly on its application, i.e. a patent.

At this point already, let me add the following remarks:

• The Fraunhofer-Gesellschaft is a non-profit organisation, financed to a level of 40% by public revenue.
• Scientific research cannot be thought without free exchange of knowledge. The perceptions of the Fraunhofer-Gesellschaft are built on those of other scientific organisations.

Anyway, let's adopt that a 20-year monopoly was an adequate reward for that part of the work done by the Fraunhofer-Gesellschaft itself. In this case a patent which protects this effort against third-parties' parallel efforts might be appropriate.

Unfortunately this is not what the MP3 patents cover.

A glance on the patent specification of one of the MP3 patents reveals the following principal claim:

1. Digital coding process for transmitting and/or storing acoustic signals, specifially music signals, comprising the following steps:

• N samples of the acoustic signal are converted into M spectral coefficients;
• said M spectral coefficients are subjected to quantisation at a first level;
• after coding by means of an entropic encoder the number of bits required to represent all the quantized spectral coefficients is checked;
• when the required number of bits does not correspond to a specified number of bits quantization and coding are repeated in subsequent steps, each at a modified quantization level, until the number of bits required for representation reaches the specified number of bits, and
• additionally to the data bits the required quantization level is transmitted and/or stored.

All this might sound terribly complicated and technical for a layman. However its meaning is simply:

• First one chooses how much bandwidth/disk storage one wants to provide.
• One compresses the music using well-known algorithms (spectral coefficients, entropic encoder) taking a loss of quality by simply dropping part of the information (quantization).
• One checks whether the result fits in the bandwidth/disk storage previously chosen. If not, one repeats the compression with another quality level until it fits.

So the patented idea is not to specify the quality and to get the size from the algorithm, but to specify the size and to adapt the quality.

Complex Science – Trivial Patents

The psycho-acoustic perception which took many man-years of world-wide scientific work and many millions of public revenue to retrieve is not mentioned at all at this point. This only comes later, in claim 11.

11. Process according to any of Claims 1 to 10, characterized in that continuously audibility thresholds of quantisation errors are calculated in accordance with the established findings in psychoacoustics, and that the spectral values are corrected so as to preclude the audibility of noise.

Like before, ths patent does not cover a extensive scientific discovery, but the idea to apply the psycho-acoustic perceptions – taken as already available – in data compression. It is a matter of opinion whether to regard this idea as “ingenious” or “obvious”. Anyway the patent is not on the fruits of their long-time work, but on a matter of seconds: a basic idea for mathematical/logical instructions.

But in fact this consideration is obsolete. In order to infringe the patent it is not necessary to infringe Claim 11. Claim 1 is enough to constitute a patent infringement.

There is no trace of extensive scientific work in Claim 1. Every professional who comes up with an algorithm for music compression has a high risk to infringe Claim 1.

Why?

How can it be that a research institute which has put a lot of real effort into the development of MP3 does not patent that real effort, but just trivialities?

One answer reads: for economic reasons. The broader the patent claim, the more money can be made out of it. If you can ask for license fees not only from real plagiarists but also from all those who accidentially came up with the same idea, you make more profit.

Another answer lies in the nature of software development. There is no need for “protection” of innovative software by a patent. Copyright and the possibility to separate the source code from the executable binaries provide perfect means to hide your know-how against competitors. When a software developer wants to make profit from a new compression algorithm, he launches a software library which implements the algorithm – as fast as possible. A software developer must hurry up to remain ahead of his competitors. When he waits some years for a patent application, his corner on the marked will be already filled by someone else.

On the software market, patents are not appropriate to protect individual innovations. Instead, they serve as weapons in a long-time strategy to prevent competition. From this point of view it makes sense for the patent applicant to file patent claims as broad and trivial as possible. That way he maximises the chance that an innovative competitor will unwittingly infringe these claims.

Non-Trivial Software Patents Are Harmful, Too!

So even the praised MP3 patents are trivial, regardless of the potentially big efforts behind the development of the algorithm. But let's assume for the moment that the claims indeed adverted to the complex parts. Wouldn't they state an incentive to innovation and thus foster economy?

Even in this case the answer was a clear “no”.

MP3 is far from being the best compression algorithm for music. Ogg/Vorbis for instance is superior in every respect and it is not encumbered by patents. Nevertheless MP3 is used much more than Ogg/Vorbis. Why so?

In the software market, interoperability plays an even more essential role than it does in other branches of trade. For instance in automotives it is okay when a spare part for a Ford does not fit in a car by GM. On the other hand there is no chance for a multimedia software which can play Ogg/Vorbis but not MP3 to be profitable.

The MP3 patents give Thomson Multimedia and Fraunhofer-Gesellschaft full control over the world-wide de-facto standard for music compression. At the moment these patent-owners are generously forbearing the existence of MP3 playback software which is available for free. In 2002 there were some plans to change this. It did no happen back then, but it can happen any time. An entire sector of industry is dependent on the good will of the owners of specific patents.

Concerning the recording of MP3 files, the patent owners are taking drastic measures in demanding license fees for any software which can compress MP3. This led to the retraction of almost all free and Open-Source MP3 encoders. As the only exception, LAME has resisted the legal pressure so far.

In addition the patent owners say that their patents cover the full spectrum of lossy audio compression, including Ogg/Vorbis. Whether this is true has yet to be carried out in front of a court.

Society Loses

To summarise:

• Even if there was extensive research and development, the patents are trivial.
• Although there are better alternatives which do not infringe the patents, the need for interoperability gives the owners of the patents control over the entire market.
• The patent owners selectively exclude specific competitors – in particular: free and Open-Source software – from the market.

A company exploiting holes in the law to make profit at the expense of others cannot be blamed. The problem lies in the system and can only be solved in the system.

It might be that patents are profitable for their owners. But they are, for certain, causing economic harm to the software industry and to society.