Conclusions

We have given evidence that software inventions can be classified in a Software Invention Cube that distinguishes the phases of the software engineering life cycle, aspects that play a role in inventions, and IPR views on them. The 100 combinations generated by this cube have widely different characteristics regarding possibilities for protection.

We are now in the position to answer the four questions that we raised in Section 2.2:

• What is the ``state of the art'' in software?
• What is a person ``skilled in the art'' of software?
• What is a software invention?
• How can a software invention be compared with the state of the art?

As our analysis shows, these questions are very hard to answer in general since they strongly depend on specific circumstances. However, in a structured context, like the SWIC, specific answers are possible as has been shown by our examples.

The SWIC can be helpful when used systematically for writing patent applications, for organizing prior art databases which are in fact ``existing invention databases'', for reverse engineering ``systems'' into constellations of embodied inventions and into families of described inventions. Much of the old heritage will have to be restructured in terms of described inventions for which patents cannot be filed anymore. Looking at existing practice this poses a significant modularization problem that can be solved (at least in principle) by means of a systematic matching with the SWIC.

Most studies related to software patents focus on only a small part of the SWIC. In particular, Engineering phase = Implementation and Technical aspect = deliverables. The SWIC makes clear that significantly more cases have to be considered.

An intriguing self-referential question is whether the software engineering life cycle itself is a software invention. This question can also be asked for the Software Invention Cube. Both are scientific results rather than inventions, as a consequence both are not patentable.

Our main conclusions are the following:

• To the best of our knowledge the Software Invention Cube is a new perspective on the subject of IPR on software.

• Software inventions do exist.

• No meaningful distinction can be made between software inventions and software discoveries.

• Although copyright on software plays a prominent role in the debate on software patents, we have shown that only in very few cases copyright is an alternative for patents to protect software inventions.

• Protection systems for software should be based on clear principles of knowledge organization. The Software Invention Cube is a proposal for such a principle.

• In a first approximation the state of the art in software is contained in the body of knowledge as documented, for instance, in SWEBOK [10]. However, important links exist to other areas such as psychology, sociology, business administration, economics, manufacturing, and electronics. Due to the wide applicability of software and software inventions, this may amount to links to most areas of knowledge.

• Due to the vast amount of knowledge that constitutes the state of the art in software, the concept of someone ``skilled in the art'' is useless. More specifically, it is useless unless this notion is specified in further detail. SWIC provides a possible decomposition of ``the art'' into manageable subareas where persons ``skilled in (that part of) the art'' can be identified.

The desirability of the protection of software inventions has technical, legal, economic, and even moral aspects that we have explicitly not addressed in the current paper.