Essentially the same question was asked in a letter by a reader in the current issue of Physics Today. Thomas E. Phipps Jr. writes
Recently I had wanted to consult a one-page comment that had appeared in the American Journal of Physics 18 years ago. I could have gone to my local university's physics department library and copied the page for 10 cents. However, being 82 and lazy, I preferred to go online to the AIP website, where I discovered that the page I wanted was available for downloading at a price of $19. Oddly enough, I paid this. [...] But I wonder how such a pricing policy squares with some of the declaratory words emanating from AIP. [...] Simply put, what is not-for-profit about charging $19 for a one-page download of 18-year-old material?
He refers to the mission statement of the American Institute of Physics (AIP), the publishers of both Physics Today and the American Journal of Physics, which says that AIP "serves physics and related fields [...] by serving [...] individual scientists, educators, students, R&D leaders, and the general public with programs, services and publications - information that matters."
Physics Today invited Fred Dylla, executive director and CEO of the AIP, to respond to Thomas Phipps, and here is part of his answer, which gives some insight in the prices involved in publishing papers in scientific journals:
It does seem inappropriate to pay $19 for a one-page download of an 18-year-old article. But one has to dig below the surface to understand the economics of scientific journal publishing as a context for the pricing of such journal products by nonprofit publishers. [...] Producing a high-quality, peer-reviewed archival journal such as AJP involves significant costs, including those for a reliable online platform that has made AJP and other member-society journals available to a much wider audience than did the former print-only subscriptions. AIP has also made major investments to digitize and make available electronically journal issues that were published in print long before the industry made the transition to digital. Those real costs are recovered, by and large, through institutional subscriptions paid by libraries and research institutions. The cost of producing one typical article is between $1500 and $3000. Considering the average journal subscriber base, a $20 price for a nonsubscriber to download an article is not out of line.
He further emphasises that AIP's journal server provides free access to online search and abstracts, that its prices are lower than those of commercial publishers, and that AIP invests its return in outreach services subsidised membership subscriptions.
Fred Dylla has an important point: producing scientific journals, making accessible back issues, and running index database and repository servers creates costs that are easy to overlook. Moreover, it's definitely true that access to scientific papers is much easier now than just 15 years ago: Back then, the only way to read a paper would have been a visit at the next university library that has a subscription - compared to that effort, spending $20 may be not that expensive. And the prize of a single paper download probably has to be balanced with the typical total number of downloads a library "buys" by its subscription fee.
But I there are two points that don't not quite convince me about this argument:
First, the "one size fits all" prizing is a bit odd: While $20 might be an understandable prize for a 100+ page review article that compares to a book, it's not straight-forward to see why a one-page short note should cost the same. Actually, this may be due to technical reasons and the capacity of current online payment systems, and could change in the future.
But second, the price tag at online papers is relevant only for potential readers without institutional affiliation providing access. There could be a substantial group of highly trained and qualified people outside universities willing to pay for the download of papers, because they want to read first-hand about scientific developments. Could it not even be profitable for publishers to grant access at a lower prize, a prize that invites to pay instead of saying "we don't want our papers being bought"?
I do not know if there are any market research studies related to this point, and if lowering the prize could indeed be more than compensated for by increased sales. I know that I would actually have already bought quite a few papers if the prize was lower. What about you? Did you ever consider buying a paper, but were discouraged because the prize tag said $20, instead of, say, $5.00 ?