The fate of Mendeley and Connotea should serve as reminders of the dangers of relying on free services. The users’ needs and desires come secondary to the real customers — in Google’s case, the advertisers; in Facebook’s case, those purchasing data and access to users. When a company has no business model, then the terms of service they provide can turn on a dime when one comes along. Think of the recent Instagram furor, or all the companies built around providing services for using Twitter, and how doomed they are now that Twitter is shutting down the very openness that made them what they are today.

It’s far better to pay for a service that’s critically important to you than to rely on one where you are not the sole focus of the company providing that service. If you’re the paying customer, then the company’s health is based on keeping you happy. Mendeley and Connotea are showing us that a free service can at any point be shut down or suffer drastic changes with little regard for the user’s best interests. If you’re an altmetrics startup that was planning on building a business around Mendeley or Connotea bookmarking as a key measure of researcher interest, what do you do now?

These economic realities stand in stark contrast to the notion of the “decoupled journal,” a vision of the future where each aspect of academic publishing is separated out and purchased by the author in an a la carte fashion. So far these services are not providing much evidence that they are viable standalone businesses. Aside from the factors discussed by Takats, many are competing with “free” and have that hurdle to overcome. Why would a researcher pay Rubriq $700 for a round of peer review when Peerage of Science, or pretty much any journal on earth (other than F1000 Research) will provide one for free?

Rather than breaking things down into individual components, perhaps we’re instead looking at a different sort of decoupling — a separation instead between services aimed at separate customer bases: authors and readers.

For any selective journal with a high rejection rate, balancing a budget in an author-pays OA economy is problematic. The three options available to improve one’s financial situation are to raise author fees, to lower one’s standards and publish more, lower quality articles, or to cut costs. And since the author will have replaced the reader as the primary customer for the journal, reader-centric functionality will be a prime target for cost savings. This comes at a time where we’re just on the verge of a revolution in new tools, in semantic technologies, API’s, alt-metrics and text-mining.

These approaches offer great promise, but also require significant investment in development, implementation, and maintenance. The only path to growth for these technologies may indeed require a decoupling, with the needs of authors paid through article processing charges (APCs) and the needs of readers (and others looking to reuse the articles) through site licenses to suites of tools.

With the acquisition of Mendeley and so many other reader service startups, the large commercial publishing houses may have already started down this path. Many see a lateral move taking place, with revenue streams merely shifting directions within the same company, from journal subscriptions to a combination of author fees and service fees.