In my opinion, using dictionaries for interpreting claims is fundamentally flawed. When viewed from the highest level, inventing is the creation of something that has never existed before. The thought that someone could adequately describe this brand new thing using old words is fundamentally flawed. This is true in any art, but software is especially laden by this problem.
Software is an odd mixture of ethereal concepts implemented in a physical machine. As these concepts are implemented, descriptors for the concepts are liberally borrowed from any place they can be poached, with the more visually appealing descriptor the better. For example, the term of art “cascading” as used in “cascading style sheets” picks up a visual metaphor of descending water going down a series of steps. In reality, “cascading” refers to the concept that each style sheet can add information to the whole, and that each style sheet can overwrite a setting of a previous one.
This example is merely one use of a label that is created along with an invention, where the label is defined by the invention, not the invention defined by the label.
By its very nature, the invention has never been known by man, so how can a dictionary definition, which was created over decades if not centuries before the invention, ever hope to capture the essence of the invention?
Patent claims are the single most difficult drafting exercise in all legal drafting. This is because of the inadequacy of using old words to describe a new invention.
Because the label is defined by the invention, not the other way around, the only proper way to define the invention is by the written description and the intrinsic evidence of the prosecution history. Dictionary definitions could be consulted, but given very low priority because the natural progression of the language is to not to apply old definitions to new ideas, but to apply the new ideas as new dictionary definitions.