Open Scholarly Publishing

MAY 16, 2016

This post outlines how an open model of scholarly publishing would work and how it would be funded.

Our current system for scholarly publishing is highly inefficient and poorly suited to the Internet age. An open model would cure many of the current ills as well as offering greater flexibility and greater potential for innovation. However, exactly how an open model would work, and how it would be funded has not always been clear.

The following sets out a simple model for scholarly publishing built around open information both in scholarly content and in the mechanisms of scholarly discovery. As well as the outline of of the overall dynamics – e.g. who publishs where and how – it also describes a sustainable and efficient funding mechanism which is key to demonstrating that an alternative, open, system is truly viable.

Summary of the Open Model

These are the three main components of the open model:

  • Funding Pool provided by libraries or directly by government.
  • Hosting Platform(s) for papers, books and other scholarly artefacts. Does almost no filtering. Funding provided from the Funding Pool.
  • Filtering and Selection Providers. Largely separate from the Hosting Platform(s). A replacement for traditional journal filtering based on: NotAJournals, automated systems (e.g. search engines), and aggregated individual reviews. Paid for from Funding Pool (and/or user fees).

Platform for hosting and storing papers

All papers are uploaded to one or more open hosting platforms (repositories) from which free and open access is allowed. Each artefact receives an unique identifier.

The platform does almost no selection or filtering - anyone can post modulo some basic checks and/or recommendation by peers already on the system.

One would probably support multiple (e.g. 3-5) platforms to provide for competition. Each platform is paid from the Funding Pool based on:

  • Number of papers they host (perhaps adjusted for size and type). This amount is adjusted annually (downwards) based on actual costs
  • Quarterly user feedback survey of their quality and performance
  • Performance in cost against other platforms - lower cost means a bonus (a form of yardstick competition).

All platforms must be open both in terms of being open services (powered by open software etc) and in allowing free and open access to their content.

Current example:

Filtering and Selection

Permitting and stimulating innovation in filtering is one of the main benefits of this new model. We therefore emphasize that whilst we set out three specific methods the open model can use there may be more.

The three main approaches are:

  • NotAJournals. A list of recommended articles produced by a group of scholars or other reviewers.
  • Automated. Any kind of platform offering automated or semi-automated filtering and selection.
  • Aggregation of individual user reviews. Will ignore this here – it is discussed in more detail in the original paper and is a more radical change.

NotAJournals produce regular list of selected papers perhaps with short reviews and/or recommendations. The list is open and may be redistributed. Similar to current Journals but without any exclusivity on content – a paper may be listed in many NotAJournal. In addition, their list is open and freely accessible to anyone.

NotAJournal are funded by a reumenaration rights model. That is a central pot of funding is created and payments to NotAJournals are made roughly proportionally to the number of subscribers to the NotAJournal.

Will probably want to adjust pure “usage” metrics to allow for “value” differentials and, for example, the size of community to make sure smaller communities are well served. In addition, one may want to engage in some redistribution and pay less to “top” NotAJournals and less to smaller ones to counteract the tendency for concentration in this kind of market (the tendency to end up with a few “star” NotAJournals).

Issues around reuse (e.g. NotAJournal B copying NotAJournal A would be dealt by agreement with some an adjudication process for disputes).

Automated services would follow a similar model being paid (roughly) in proportion to use from the funding pool. Importantly, to receive any payments the service must be open as per the Open Software Services Definition – that is, its software, algorithms etc must be open along with its data (at least all non-personal data).

Finally, we emphasize that:

  • There is no requirement that a NotAJournal or a filtering servce be non-profit. For-profit entities are very welcome.
  • A NotAJournal or filtering service can choose to be “closed” but they must then opt-out of receiving funding from the funding pool and should receive no public support of any kind. The would therefore need to support themselves entirely from other sources, for example user fees (though these must be paid out of scholars own pockets and not funded institutionally).

Current example: there are no perfect examples today that I know of, but the editorial boards of community oriented open-access journals would be very similar to the editorial board for a NotAJournal.


All payments both for platforms and should be public and openly shared in the community. Transparency is important in promoting trust, efficiency and accountability and fits naturally with the overall open approach.


In 2009 I wrote a paper about scholarly publishing, whose main message was that scholarly publishing serves two goals:

  • Hosting and Distribution — helping people to read. Providing a distribution mechanism for scholars to share information, for example by putting their paper online and providing readers with convenient access (e.g. downloads!)
  • Selection and filtering — helping people decide what to read. Enabling readers to filter and select from the vast range of materials to identify the set of things they actually do read.

These goals are distinct and they need not be pursued together – in fact, it is usually inefficient to do so. However, today we do pursue them together using a single mechanism in the form of the scholarly journal or publisher. Combined with the exclusivity of copyright, and the assignment of that copyright to publishers, this is causing serious harm to the efficiency of the scholarly enterprise.

The lock-in and platform power that are the cause of these ill-effects themselves arise largely from the second activity: the selection and filtering function. Put crudely: everyone has to buy Nature because it is a top journal, it is a top journal because everyone wants to publish in it and everyone wants to publish in it beause everyone reads it etc. This is the reinforcing cycle of journal lock-in which creates pricing power and stands in the way of change. Moreover, not only is the selection and filtering function the source of the platform power, but much of the harm and inefficiency in the current system comes from the fact that journals and monographs are such an inefficient way to do selection and filtering.

The 2009 paper set out the design for a new approach and we follow that here to a great extent. However, the original proposal was missing any details of financing – who would be paid and how.

Performing hosting and doing selection and filtering may not be a cheap business. We need to know the means and mechanism that would pay for them. The means for doing this we already have in the form of the existing funding to libraries and research institutions. However, we also need a mechanism for distributing those monies – after all, it those monies today that pay for the current highly inefficient system. The open model offers a new solution based on “remuneration rights” whereby these collective funds are used to fund open distribution and filtering in a manner that is transparent, efficient and attunded to actual usage and demand.

A Note on Openness

Open here means “open information” not “open participation”. It is about opening up information not changing the structure of participation in scholarship per se – though opening up information will also allow for more flexilibity and innovation in important areas.

The openness of information is in two areas. First, and most familiar, it is in the form of “open access”: all research content would be open, free for anyone to use, share and build on (see the Open Definition for details).1 Second, and less familiar, it is about openness in the selection and filtering of that information – for example, the list of papers in a journal is itself information that can be open, so too, the software and algorithms of a scholarly discovery engine is information that can be open.

We see openness as important in both areas: both in scholarly content and the mechanisms of discovery.


This post arose out of the excellent discussion after a recent presentation at the University of Cambridge on the topic “Making an Open Information Age” (based on my upcoming book). My thanks especially to Ruper Gatti, Peter Murray-Rust and Jenny Molloy.


Please post questions and comment in the Open Knowlege forum here:

The above necessarily omits many details and questions and comments are very welcome.

  1. I emphasize that traditional requirements for attribution and credit would be preserved – indeed, enhanced. ↩︎