holtzermann17
commented
9 years ago what I've aimed for in the revisions is a clarification that, if each of the SPECS criteria have been met, then we have to attribute serendipity; in itself, it is not subjective. On the other hand, some of the criteria (like value) are subjective, which is why we have emphasised 3rd party judgements. Is this clear?
In the last paragraphs of Section 4 on page 16, the authors discuss how a system may discover or invent something serendipitous, but not recognise that fact, and explicitly call out this exception as not being a "serendipitous system". It seems like this is not an exception to the definition proposed in the previous section, but merely a clarification — the authors have already argued that if a system does not recognise its creation as unexpected and valuable, then it is not being serendipitous. This mirror the standard way of thinking about artificial creative systems, and can probably be made more concise by referencing past work along those lines.