Is an editorial in the New England Journal of Medicine an editorial, or is it an article?
What kind of a question is that? It says so at the top: it is an editorial. In the classification scheme used by many citation databases, it carries the type "editorial" – a type that is filtered out in many analytics use cases. But editorials in the NEJM are opinion pieces, often heavily cited, that are rightly seen as a critical part of scientific discourse. One would wish that more journal editors guided their fields with strong editorials.
So why would we treat such content differently from an article? The answer comes from my number one lesson in content modelling, which I have been repeating for twenty-five years: model the content, not the words.
What "ed" really contains
If we look at the records classified with type "editorial" in a large citation database, we see very few actual editorials. We see prefaces. We see the front matter of books. We see preliminary pages. All of these have been classified as "editorial" over the years, because it was the closest available type. Just as words in natural language change meaning over time, the meaning of "editorial" in the database has changed. The label stayed the same, but the population changed. Today, "editorial" no longer reliably means editorial.
The question is not whether something is called an editorial. The question is: what is it, functionally, in the data? Seen this way, there is justification for classifying an influential NEJM opinion piece as an "article". It no longer fits the category it sits in.
Conference prefaces as a gold mine
Dismissing prefaces as things that can be filtered out is equally wrong. A few years ago I determined that prefaces of conference proceedings are extremely valuable content: they contain the venue, the organizing committee, sponsoring bodies, keynote speakers, and often a short history of the conference series. From prefaces alone, we could have built the best conference database in the world.
But prefaces do not have abstracts. Therefore they belong to a set of item types that are largely empty in a citation database – and for that reason are routinely filtered out. But why are they empty? Nothing stops us from generating a useful summary of the preface and calling that the abstract. Nobody said the abstract has to be the portion of the text marked "abstract" by the author. We used to create abstracts ourselves. We stopped for cost-saving reasons. It has only become easier.
Systematic reviews
Systematic reviews are perhaps our most valuable content type. It has been known to happen that a systematic review is classified as a "short communication", simply because the production turnaround time of a short communication is faster, allowing the review to be rushed into the next issue. If systematic reviews are important to us, we need them regardless of what type the publisher assigns. Their label cannot be trusted, and therefore it cannot be the basis for our decisions.
The practical lesson
Document types have become disconnected from any business outcome we would want to achieve. The last thing we should do is take decisions about the eligibility of content based on an ill-fitting one-word label.
Is an editorial an editorial? No – what matters is not what it is called, but what it is.