PKP 2011 Hackfest - Liquid Publishing Project
What is this?
Liquid publishing stands for making scholarly communication more efficient through collaborative authoring, ongoing open review, iterative editing and easy progression from one publication format to another (e.g. draft, article and monograph).
While implementing a basic Wiki-style collaborative authoring model, liquid publishing needs to respond to the specific needs of scholarly researchers, both, in their roles as authors and readers:
- citability (traditional citation formats must be able to point to "liquid" publications and vice versa)
- as much integration as possible with other research tools and publishing formats like reference managers, mind maps, e-Publishing software, research databases, etc.
- quality assurance through ongoing peer review
- a versioning model that allows for intermediate versions and fully peer-reviews "stable" versions
- accessibility, e.g. indexing, harvesting, etc.
- archiving and long-term availability
- reputation management, e.g. usage statistics, impact factor metrics, author attribution system, etc.
We'd like to base this off of existing software, e.g. MediaWiki, and implement only a few extra modules required for scientific publishing.
What might be done in this project?
- specification of plug-ins
- usability and graphics design
- implementation / testing
Systems could be developed in which anyone who cited a work in their article that was subsequently updated would see the citation updated so that it indicated the version cited, as well as the updated version and the differences highlighted.
Similar features are already implemented in Wikipedia. We could reuse the "subscribe to an article" feature which gets you informed via email whenever that article changes and allows you to look at the differences since the last time you visited.
See  for another example.
We also have to decide whether we should use some kind of pagination or paragraph numbering? How do we deal with changed paragraph numbers/pagination after an update?
Open Peer Review
A major version of a publication should only be released after considerable quality assurance has been done so that the scholarly standards of the research field in question can be maintained in a liquid publishing model.
We could either implement a traditional "closed" peer review process for that or even some controlled crowd sourcing process, i.e. an "open" peer review model.
This is somehow similar to . They use a classical review committee and controlled updates only.
A possible peer review process could look like this:
- Configure your Wikipedia to make the main article page only changeable by the author and the discussion page changeable by the public. Stock Wikipedia has a "this is a minor update" checkbox. If you check that then people subscribed to the article will not be notified of the change. We could rename that checkbox to something like "this update has not been reviewed". Then change Wikipedia to set that box by default and disallow authors to unset it.
- Proposed changes to the article as well as unstructured thoughts, ideas, new research data, etc. should first be added to the discussion page of an article where they can be "crowd reviewed" for some time.
- If the author finds some time to structure the new material into some well formulated text then (s)he can update the main article page with the new results. Authors who cite the article will 'not' be notified of such an unreviewed change. Some message on the main article page should show up after such an update to alert readers that this is no longer a peer reviewed version and point to the last reviewed version for actual reading and citing.
- Whenever the author thinks that (s)he has reached a version of the article that is worth the valuable attention of their fellow peers up the citation chain then (s)he'll request a review, eg. via a simple "please review my article" button.
- A group of reviewers of the article will then be alerted. Reviewers can request changes, completely reject the update, etc. as in any peer review process.
- As soon as, say, three reviewers have voted favorably for the change, then Wikipedia's stock change alert will automatically be sent out to notify upstream authors that a new peer reviewed version of the article has been made available by the author.
Upstream authors will receive a link to "show me all differences since the last reviewed version" which is trivial to implement in Wikipedia. They can then check whether the change is relevant to their own research, update their citations if required, etc.
Upstream publications that are in the same system can be checked automatically for affected citations. Other online publications can be checked for affected citations via a simple Custom Google Search. And so on...
See also the related Open Reviews project.
This goes even further than open reviews. While reviews lead to annotations leaving the original text as is, collaborative authoring could mean that an authorized group or even grand public is allowed to edit a publication.
Collaborative publications must provide a means to partition contributions into citable entities (monograph, article, page, paragraph) and identify the contributions of all authors to these entities, e.g. similarly to the "blame" function in modern version control systems.
On this basis the meta-data of "liquid" publications must be prepared in a way that can be distributed through existing protocols to ensure visibility and accessibility:
- OAI-PMH (at least DC)
- classic citation styles (APA, Chicago, MLA, Vancouver, ...) as well as structured citation styles (BibTeX, EndNote, ...)
- integration with redirection services (DOI, URN, PURL, Handle, ...)
Moving between publication formats
Several paragraphs can be linked into an article. One Wikipedia page represents one article.
Several pages can be bundled into a monograph.
Earlier versions of publications must remain available even after such bundling has taken place (to keep earlier citations of the work valid and for vanity purposes.
The Wikipedia sister project "Wikibooks" has a good solution already to transform a loose collection of thematically related articles into a sequential monograph.
It is possible at any time to transform a "wikibook" into an actual eBook and download it, e.g. in PDF format.