Image: By Malvika Sharan @MalvikaSharan Twitter https://twitter.com/MalvikaSharan/status/1043084660522270722
A rare breed of open Web researcher, testing assumptions about publishing and academic freedom by creating the demonstrative software ‘dokieli‘. A browser based, decentralized publishing software, designed on the principles of—empowerment, individual autonomy, decentralized and interoperable applications, universal access, and a social Web. And why the experiment “This ‘Paper’ is a Demo“, borne from a healthy dose of stubbornness, came about. The current line of thinking at a high-level is captured as part of the ‘Linked Research‘ initiative.
We are at the FORCE11 annual conference taking place at McGill University in Montreal and I have the opportunity to interview Sarven after a brief presentation he has just finished, titled “Social Scholarly Web“. I say a brief presentation as the conference is a jam-packed few days, with three hundred and fifty guests from backgrounds across academe—librarians, technologists, and scholarly communications industry—rushing around between the three simultaneous presentation tracks. In his presentation, Sarven pointed to a definition of the Decentralized Web (DWeb) from the 2017 paper “Systematizing Decentralization and Privacy” (Troncoso et al. 2017), which synthesizes a history of the field. And among other things a series of brief demos of dokieli, the experimental publishing software Sarven has built, are hurried through, of course all are self-hosted on the open Web, the trademark of this auto-icon of the DWeb.
I start off by asking Sarven, “What’s your elevator pitch for dokieli?” He replies very succinctly, “it’s a publishing tool using the read / write functionality of the Web.” I follow with the question, “What is the ‘problem space’ in Open Science that you are dealing with?” Sarven points to the paper he has already referenced in his presentation “Forces and Functions in Scientific Communication” (Roosendaal and Geurts 1997), where the authors state
the linear information chain is being fundamentally transformed into an interactive communication network… A starting point is that our current policies and practices in science and communication are not ideal for an optimal exchange and refinement of our knowledge
(Roosendaal and Geurts 1997)
We agree to start at the beginning of Sarven’s research journey and how dokieli came about. Sarven explains, “I was working at StatusNet back in 2008-09, in UX / UI, it’s this experience on federated social micro-blogging that gave me a foundation in Web architecture and informed my enquiry into what is currently known as ‘Social Web’.” At this time there was a push for autonomy and data portability which went under the name of federated media, this was in response to the growth of the enclosure of the information commons by social media monopolies such as Facebook and Twitter. This trend of enclosures also went against the long-standing vision of the Web as a many-to-many-media borne out of the 1960s future casting as exemplified by fellow researchers following the dictum of “show, don’t tell“ acting as public experimenters, with The Mother of All Demos (Engelbart, 1968), or the self-published book such as Computer Lib/Dream Machines (Nelson, 1974) by Ted Nelson, who coined the term hypertext.
Sarven says, “I returned to academia working on linked statistical dataspaces and analysis, and in 2015 took up a PhD at Bonn University, and then I joined the Solid project at MIT Computer Science and Artificial Intelligence Laboratory (CSAIL)’s Decentralized Information Group (DIG) as a visiting researcher for a year.” The Solid project is part of the ongoing vision of the Web and is headed by Tim Berners-Lee. Solid stands for Social Linked Data and incorporates a particular set of Web-centric open protocols and standards that is being developed and is described as re-decentralizing the Web (Tim Berners-Lee, 2016). “I’ve been self-publishing since my Masters” says Sarven, “when I saw the Semantic Web (SW) community members were not using their own Web tools, and available standards, and instead submitting PDFs as papers, I decided to extend my own Web self-publishing to prove that one could achieve the same quality, and more, using the Web stack. I took an HTML document and applied ACM paper typesetting styling, simply using CSS, and enriching the document with RDFa. This is where the tongue-in-cheek meta-paper “This ‘Paper’ is a Demo” came from. The paper was an ongoing experiment, I performed the paper, I read the paper out loud and recorded and embedded the audio, and did the same with video—graphing it, enabling executable code in the browser, included different stylesheets for print-views (like ACM, LNCS), and more. dokieli was born out of this hackish demonstration, as a provocation, I wanted to dismiss what was informally raised as hypothetically impossible in the SW research bubble. As far as I was concerned, this was trivial and common stuff on the Web for a decade or more.”
dokieli is described on its website as “a client side editor for decentralized article publishing, annotations and social interactions”. The term client side editor means it runs in the Web browser. The three decentralized functions—publishing, annotation, and social interactions—are handled by various protocols and standards, as well as decentralized architectures. For instance, users can optionally participate by using their own “WebID” (personal identifier with a structured description) that is stored and served from a location which they prefer. Similarly, users can author or annotate articles that are self-hosted or have write-access to where they are stored. Critically, all content and data created by an individual, no matter what the application is they use to create it, is stored in their own personal datastore. This can be something the user hosts themselves (for the more tech-savvy) or which they choose to be hosted by a person or organisation they trust—whether something they pay for, or a university or other institution. This also includes the social interactions, like notifications about activities (sharing, liking, reviewing, replying), so that they can be used by other applications. The point of dokieli is to serve as one such application which enables and encourages people to have control over the data they produce, rather than handing it over to some third-party (as is the case with most current mainstream social media, or academic publishing services). dokieli serves as a reference implementation, or demonstration of what is possible for interoperable, Web standard compliant applications and services. Sarven hopes that other developers can follow its example and build similarly user-empowering Web applications. Sarven said about dokieli, “dokieli is being made through experiments in an academic context, but is a general purpose tool and can be considered domain agnostic. It is also worth mentioning that W3C made such a read-write publishing tool as a web browser called Amaya, that used web standards of the time, it was developed between 1996-2012 and serves as an inspiration for dokieli.”
I ask Sarven about dokieli’s relationship to the Solid project. “I was welcomed to the Solid research group by Tim Berners-Lee and dokieli came about from my previous Web experiments. There was a good fit between the two projects as I was using what was being developed in Solid, its tooling, technology, and infrastructure. I was trying out advanced approaches, to figure out what’s not standardized, and to see how far we can take tooling that is purely based on open Web standards. dokieli can interact with Solid-compliant servers.” During that year, Sarven worked on Linked Data Notifications (LDN), a W3C recommendation of the Social Web Working Group, which was implemented and exemplified as part of dokieli and Solid, as well as thoroughly tested and written up as a peer-reviewed (and award-winning) academic paper “Linked Data Notifications: A Resource-Centric Communication Protocol” (Capadisli 2016) for the Extended Semantic Web Conference. dokieli and LDN also work well alongside other W3C recommendations like Web Annotation Protocol, ActivityPub, and the Linked Data Platform.
The question came up of how dokieli is run, maintained, and takes user feedback. Sarven replied, “people either do or don’t buy into the principles—interoperability, using the Web layer—open source is taken as a given now; what is important for decentralization is interoperability, if we want to enable technology to serve scholarly communications.” Sarven actively solicits input from various communities, including decentralized Social Web hackers and advocates, Semantic Web enthusiasts, and Open Science practitioners. The source code on GitHub is connected to a Gitter chat, where potential users or interested parties can give feedback, ask for help, or make feature requests.
“Questions can be addressed and ideas applied by using the dokieli tool as a location for experimentation, for example addressing social implications of annotation and UX as in a conversation. The dependencies around dokieli still need to be fully developed to come along with the ideas, and this is a bi-directional process, with many more rounds of tests to go. But it’s possible to go quite far by using the Linked Data (LD) design principles and related Web standards.”
After talking in-depth about dokieli I wanted to open up the focus and ask some questions about the wider Decentralized Web movement and the different technologies involved, as well as the value to scholarly communications of the Decentralized Web. The fast moving FORCE11 conference was starting to puncture our interview bubble as people emerged from the round of presentations just finishing, excited and networking away, as they buzzed to their next session. Time was short as people wanted to catch up with Sarven and also talk DWeb, which is a very good side of the FORCE11 conference, which encourages a leveling of hierarchies itself. The question was about Sarven’s view of other technologies in the DWeb / P2P space—Maidsafe, Blockchain, DAT—Sarven had one quick reply, “I think there is a healthy net gain with different approaches and experiments towards a common DWeb vision. However, we should be careful about different forms of social agreements that surround each design choice.”
My final question was about DWeb and Open Science. “How to improve the making of the digital infrastructure and tools of science knowledge and communications?” Which, on so many levels, needs vastly improving to come anywhere near addressing Roosendaal and Geurts 1997 observation, or what you could also say is an understatement, “current policies and practices in science and communication are not ideal for an optimal exchange and refinement of our knowledge.” (Roosendaal and Geurts 1997) “Bad incentives,” is Sarven’s reply, “currently what is dominant is: how to bring value and monetize usage; development is driven by ‘the next big thing’; and the rush for control of data, designing vendor lock-ins, being the measure of success. There are complex decisions for decentralization of systems, what should be done instead is to look again at our assumptions, and what are the benefits for society and the individual. What are the constraints of “open” approaches? Are we designing socially and technically inclusive systems? Who ultimately controls your online identities and data? What are the real costs of participation? How are we dealing with social expectations and institutional pressures?”
Capadisli, Sarven. dokieli. Accessed 27 March 2018. https://dokie.li/.
Capadisli, Sarven, Sören Auer, and Reinhard Ried. ‘This “Paper” Is a Demo’, 24 February 2015. http://csarven.ca/this-paper-is-a-demo. Appeared In: ESWC-D-2015
‘FORCE11’. Accessed 21 October 2018. https://www.force11.org/.
Capadisli, Sarven. ‘Social Scholarly Web’, 12 October 2018. http://csarven.ca/presentations/social-scholarly-web/.
Troncoso, Carmela, Marios Isaakidis, George Danezis, and Harry Halpin. ‘Systematizing Decentralization and Privacy: Lessons from 15 Years of Research and Deployments’. ArXiv:1704.08065 [Cs], 26 April 2017. http://arxiv.org/abs/1704.08065.
UCL. ‘Auto-Icon’. Bentham Project, 17 May 2018. https://www.ucl.ac.uk/bentham-project/who-was-jeremy-bentham/auto-icon.
Capadisli, Sarven. ‘#FORCE2018 Presentation on Social Scholarly Web: http://Csarven.ca/Presentations/Social-Scholarly-Web/ … #LinkedResearch’. Tweet. @csarven (blog), 12 October 2018. https://twitter.com/csarven/status/1050821223565922305.
Forces and functions in scientific communication, 1997, Roosendaal and Geurts ‘Forces and Functions in Scientific Communication’. Accessed 16 October 2018. http://www.physik.uni-oldenburg.de/conferences/crisp97/roosendaal.html.
StatusNet. Wikipedia, Accessed 22 October 2018. https://en.wikipedia.org/wiki/StatusNet.
Engelbart, Douglas. “The Mother of All Demos.” Fall Joint Computer Conference, 1968. https://www.youtube.com/watch?v=yJDv-zdhzMY.
Nelson, Ted. ‘Computer Lib/Dream Machines – Wikiwand’, 1974. https://www.wikiwand.com/en/Computer_Lib/Dream_Machines.
‘Hypertext’. Accessed 22 October 2018. https://www.wikiwand.com/en/Hypertext.
‘MIT CSAIL’. Accessed 18 October 2018. https://www.csail.mit.edu/about.
‘Decentralized Information Group’. Accessed 25 October 2018. http://dig.csail.mit.edu/.
Keynote Address: Tim Berners-Lee – Re-Decentralizing the Web – Some Strategic Questions, 2016. http://archive.org/details/DWebSummit2016_Keynote_Tim_Berners_Lee.
‘WebID’. Wikipedia, 14 October 2018. https://en.wikipedia.org/wiki/WebID.
‘Amaya’. Accessed 15 October 2018. https://www.w3.org/Amaya/
‘Solid’. Accessed 22 October 2018. https://solid.mit.edu/.
‘Linked Data Notifications’. Accessed 25 October 2018. https://www.w3.org/TR/ldn/.
‘Socialwg – W3C Wiki’. Accessed 25 October 2018. https://www.w3.org/Social/WG.
Capadisli, Sarven, Amy Guy, Christoph Lange, Sören Auer, Andrei Sambra, and Tim Berners-Lee. ‘Linked Data Notifications: A Resource-Centric Communication Protocol’, 1 May 2016. http://csarven.ca/linked-data-notifications.
‘14th ESWC 2017’. 14th ESWC 2017. Accessed 25 October 2018. http://2017.eswc-conferences.org/.
‘Web Annotation Protocol’. Accessed 25 October 2018. https://www.w3.org/TR/annotation-protocol/.
‘ActivityPub’ Accessed 25 October 2018. https://www.w3.org/TR/activitypub/ .
‘Linked Data Platform 1.0’. Accessed 25 October 2018. https://www.w3.org/TR/ldp/.
dokieli Is a Clientside Editor for Decentralized Article Publishing, Annotations and Social Interactions: Linkeddata/Dokieli. https://github.com/linkeddata/dokieli.
Herbert @firstname.lastname@example.org. ‘Very Much Agreed. Couldn’t Care Less Whether Infrastructure Is Operated Using Open Source, Commercial, Whatever Software. It’s about Whether the Infrastructure Components Provide Open, Interoperable, Web-Centric Interfaces.Https://Twitter.Com/Csarven/Status/1050487942265884672 …’. Tweet.@hvdsomp (blog), 11 October 2018. https://twitter.com/hvdsomp/status/1050491565552984064.