Here’s a fascinating look at the history of the research paper abstract from Aileen Fyfe of the University of St. Andrews. I love media stories like these, where necessity generates a new format with unexpected value, which becomes a useful standard, and then totally falls apart as new trends emerge.
In the case of abstracts, it seems the first versions were written by secretaries into Royal Society minute-books as a short-hand for describing the papers that were read aloud during meetings. Because they were such an easy way to understand a paper, editors began using them to make publishing decisions for the Society’s semi-annual Transactions journal. Then someone realized they could be published much faster and cheaper than full articles, and in the 1830s, started publishing just the abstracts in the Society’s monthly Proceedings. A century later during the post-war research boom, the abstract became a vital tool for researchers to keep up with the increasing volume, and specialization, of research.
At a certain point, though, it seems the abstract became a victim of its own success. According to Fyfe:
Third-party summaries might be better-written and more widely-comprehensible, but they did take time and money to produce, usually appearing retrospectively. This was why the Royal Society’s 1948 conference supported the wider use of author-generated abstracts, despite their disadvantages.
‘The present general unsuitability of authors’ summaries for use as abstracts is recognised; nevertheless, if these could be used it would increase the speed of publication and reduce the cost of journals publishing abstracts.’ (Circular letter from Royal Society to UK learned societies, 12 April 1949)
Three-quarters of a century later, not much has changed. Abstracts are mostly author-written today, and, as researchers knew full well by the mid-20th century, highly variable in quality. Some are written as they were originally intended, as executive summaries describing an underlying article’s background, methods, and conclusion. But many take the form of a sort of academic click-bait, advertising the marvels of the underlying paper rather than describing it.
Despite these issues, because abstracts have now been around for going on two centuries, they’ve embedded themselves into the basic structure of research. Today, websites can be penalized by academic aggregators and search engines for not providing them, and many academic sites provide an incentive for authors to write abstracts in more promotional styles by using them for other metadata like the meta description tag, which shows up in search results, and open graph tags, which show up on social media sites. We’ve resisted this trend on PubPub both by not enforcing that authors write an abstract and by providing separate fields specifically for promotional search and social text. That way, publishers, authors, and readers can theoretically have the best of both worlds. But we’re in a very small minority when it comes to academic publishing tools, and our decision causes problems when authors, and particularly aggregators, assume that we use the abstract for metadata fields that are intended for other purposes.
None of this is to say that high-quality summaries aren’t useful in research. They plainly are: good abstracts that provide details about, for example, sample sizes and methodologies are a joy to read. But, like any long-lived structure, it’s worth taking a step back to see how well they match today’s media contexts. As the trend of asking authors to provide “lay summaries,” and even “Twitter abstracts” indicates, we may need to reevaluate the purpose and audience of the abstract, as well as how they interact with today’s very different distribution and discovery mechanisms.