Web of Browsers

Web of browser is an innovation action from the Descent Labex Cominlabs Project. It gathers 2 academic teams GDD, ASAP, and the company APIZEE.

Context and objectives

Browsers are de facto the most widely deployed execution environments in the world. Initially simple HTML readers, they now run complex applications interacting with humans and web services. The recent introduction of WebRTC has further extended the capability of browsers by introducing support for browser-to-browser communication. This turns browsers into a decentralized execution environment where interactions between human and web services are enabled without third parties. 

This is a major evolution for privacy protection: many services can be run without a service provider. This is a major evolution for the right to oblivion: services are powered by the browsers of participants; data and services disappear if no more participants maintain them. Finally, this is also a major evolution for scalability: millions of participants gather an impressive amount of CPU, storage, and human computation resources.

The Web of browsers is a vision where the web is serverless, ephemeral and massively decentralized. Web where pages are hosted by networks of browsers connected through WebRTC.

In this vision, URIs no longer refer to HTML documents or Data, it refers to web sessions; a network of browsers hosting a document. A document can refer to another document using a session URIs allowing navigation. When clicking on a session link, the user’s browser joins the network of browsers, retrieves the document from one participant and can share it.

Each Web session allows real-time updates of document/data as in Google-doc. Unlike Google doc, there is no central server and no limitation on the number of connected browsers hosting a particular document as demonstrated in CRATE [NMM16].

The Web of Browsers is quite different from the current web:

  • This web forgets by default i.e. if no more users participate in the web page sharing, then this page will disappear. We think it is a good thing: remembering a page should require an effort, the loss of the information is the default.
  • This web can be more respectful of privacy: no third party is required to maintain web pages.
  • This web can be hosted on the low-connectivity/disconnected network. Information propagation is based on gossiping i.e. it can be deployed on ad-hoc networks.
  • This web will be real-time.

The objective of the project is to build and experiment with the Web of Browsers. 

We target events such as massive open online courses (MOOC), TV shows, conferences gather larger groups [BPD13]. We think the Web of Browsers should allow users to edit web pages at any time and anywhere, regardless of the number of participants. Even if only a small subset among millions of participants are writing, all participants of the editing session should be able to read and write in real-time whenever they want. In 2013, Coursera gathered 41000 participants for a MOOC entitled “Fundamentals of Online Education: Planning and Application”. However, the course relied on Google tools that only allow a limited number of users to edit simultaneously. The result was a “disaster” according to journalists [ORE13, STR13]. This example clearly demonstrates a useful use-case designed for large groups. [TRA12] reports a similar issue in the context of massively distributed authorship.

Results

We produced SooCrate, an implementation of the web of browsers.

 

We solved 2 important problems for causal broadcast in dynamic networks, that makes finally causal broadcast scalable.

2018

  • [PDF] A. Grall, H. Skaf-Molli, and P. Molli, “SPARQL Query Execution in Networks of Web Browsers,” in THE 17TH INTERNATIONAL SEMANTIC WEB CONFERENCE, WORKSHOP ON DECENTRALIZING THE SEMANTIC WEB, Monterey, United States, 2018.
    [Bibtex] 
    @inproceedings{grall:hal-01805154,
TITLE = {{SPARQL Query Execution in Networks of Web Browsers}},
AUTHOR = {Grall, Arnaud and Skaf-Molli, Hala and Molli, Pascal},
URL = {http://hal.univ-nantes.fr/hal-01805154},
BOOKTITLE = {{THE 17TH INTERNATIONAL SEMANTIC WEB CONFERENCE, WORKSHOP ON DECENTRALIZING THE SEMANTIC WEB}},
ADDRESS = {Monterey, United States},
YEAR = {2018},
MONTH = Oct,
KEYWORDS = {Decentralized Data Management ; Decentralized Applications ; SPARQL Query Processing ; Web Browsers},
PDF = {http://hal.univ-nantes.fr/hal-01805154/file/sparql_query_in_networks_of_browsers_camera_ready.pdf},
HAL_ID = {hal-01805154},
HAL_VERSION = {v2},
}
  • [PDF] A. Grall, H. Skaf-Molli, and P. Molli, “Querying RDF data in Networks of Web Browsers,” in THE 17TH INTERNATIONAL SEMANTIC WEB CONFERENCE, MONTEREY, United States, 2018.
    [Bibtex] 
    @inproceedings{grall:hal-01837114,
TITLE = {{Querying RDF data in Networks of Web Browsers}},
AUTHOR = {Grall, Arnaud and Skaf-Molli, Hala and Molli, Pascal},
URL = {http://hal.univ-nantes.fr/hal-01837114},
BOOKTITLE = {{THE 17TH INTERNATIONAL SEMANTIC WEB CONFERENCE}},
ADDRESS = {MONTEREY, United States},
YEAR = {2018},
MONTH = Oct,
PDF = {http://hal.univ-nantes.fr/hal-01837114/file/querying_rdf_data_in_networks_of_browsers_camera_ready.pdf},
HAL_ID = {hal-01837114},
HAL_VERSION = {v1},
}
  • [PDF] B. Nédelec, P. Molli, and A. Mostefaoui, “Causal Broadcast: How to Forget?,” in The 22nd International Conference on Principles of Distributed Systems (OPODIS), Hong Kong, China, 2018.
    [Bibtex] 
    @inproceedings{nedelec:hal-01923830,
TITLE = {{Causal Broadcast: How to Forget?}},
AUTHOR = {N{\'e}delec, Brice and Molli, Pascal and Mostefaoui, Achour},
URL = {https://hal.archives-ouvertes.fr/hal-01923830},
BOOKTITLE = {{The 22nd International Conference on Principles of Distributed Systems (OPODIS)}},
ADDRESS = {Hong Kong, China},
HAL_LOCAL_REFERENCE = {ACTI},
YEAR = {2018},
MONTH = Dec,
KEYWORDS = {Causal broadcast ; complexity trade-off ; large and dynamic systems},
PDF = {https://hal.archives-ouvertes.fr/hal-01923830/file/paper.pdf},
HAL_ID = {hal-01923830},
HAL_VERSION = {v2},
}
  • [PDF] B. Nédelec, P. Molli, and A. Mostefaoui, “Breaking the Scalability Barrier of Causal Broadcast for Large and Dynamic Systems,” in 37th IEEE International Symposium on Reliable Distributed Systems (SRDS), Salvador de Bahia, Brazil, 2018.
    [Bibtex] 
    @inproceedings{nedelec:hal-01778901,
TITLE = {{Breaking the Scalability Barrier of Causal Broadcast for Large and Dynamic Systems}},
AUTHOR = {N{\'e}delec, Brice and Molli, Pascal and Mostefaoui, Achour},
URL = {https://hal.inria.fr/hal-01778901},
BOOKTITLE = {{37th IEEE International Symposium on Reliable Distributed Systems (SRDS)}},
ADDRESS = {Salvador de Bahia, Brazil},
HAL_LOCAL_REFERENCE = {ACTI},
YEAR = {2018},
MONTH = Oct,
KEYWORDS = {causal broadcast ; large and dynamic distributed systems},
PDF = {https://hal.inria.fr/hal-01778901/file/paper.pdf},
HAL_ID = {hal-01778901},
HAL_VERSION = {v2},
}

2017

  • [PDF] A. Grall, P. Folz, G. Montoya, H. Molli, P. Molli, M. V. Sande, and R. Verborgh, “Ladda: SPARQL Queries in the Fog of Browsers,” in 14th ESWC 2017, Portoroz, Slovenia, 2017.
    [Bibtex] 
    @inproceedings{grall:hal-01585146,
TITLE = {{Ladda: SPARQL Queries in the Fog of Browsers}},
AUTHOR = {Grall, Arnaud and Folz, Pauline and Montoya, Gabriela and Molli, Hala and Molli, Pascal and Sande, Miel Vander and Verborgh, Ruben},
URL = {https://hal.archives-ouvertes.fr/hal-01585146},
BOOKTITLE = {{14th ESWC 2017}},
ADDRESS = {Portoroz, Slovenia},
YEAR = {2017},
MONTH = May,
KEYWORDS = {Delegation ; Linked Data ; Triple Pattern Fragments ; RDF ; SPARQL},
PDF = {https://hal.archives-ouvertes.fr/hal-01585146/file/demo%20-%20Ladda%20SPARQL%20Queries%20in%20the%20Fog%20of%20Browsers.pdf},
HAL_ID = {hal-01585146},
HAL_VERSION = {v1},
}

References

[NMM16] Brice Nédelec, Pascal Molli, Achour Mostéfaoui, CRATE: Writing Stories Together with our Browsers. Demo 25th World Wide Web Conference 2016. CRATE is a real-time distributed and decentralized CollaboRATive Editor running in web browsers. https://github.com/Chat-Wane/CRATE

{BPD13] L. B. Breslow, D. E. Pritchard, J. DeBoer, G. S. Stump, A. D. Ho, and D. T. Seaton. Studying learning in the worldwide classroom: Research into edx’s first mooc. Research & Practice in Assessment, 8:13–25, 2013.

[ORE13] W. Oremus. Online class on how to teach online classes goes laughably awry. Slate US, February 2013.

[STR13] V. Strauss. How online class about online learning failed miserably. Washington Post, February 2013.

[TRA12] B. Tomlinson, J. Ross, P. Andre, E. Baumer, D. Patterson, J. Corneli, M. Mahaux, S. Nobarany, M. Lazzari, B. Penzenstadler, A. Torrance, D. Callele, G. Olson, S. Silberman, M. Stunder, F. R. Palamedi, A. A. Salah, E. Morrill, X. Franch, F. F. Mueller, J. J. Kaye, R. W. Black, M. L. Cohn, P. C. Shih, J. Brewer, N. Goyal, P. Nakki, J. Huang, N. Baghaei, and C. Saper. Massively distributed authorship of academic papers. In CHI ’12 Extended Abstracts on Human Factors in Computing Systems, CHI EA ’12, pages 11–20, New York, NY, USA, 2012. ACM.

[FMP+16] Davide Frey, Achour Mostefaoui, Matthieu Perrin, Pierre-Louis Roman, and François Taïani. Speed for the elite, consistency for the masses: differentiating eventual consistency in large-scale distributed systems, Proceedings of the 35th Symposium on Reliable Distributed Systems (SRDS 2016), Budapest, Hungary, September 2016 (10p.)

[NTD15] Brice Nédelec, Julian Tanke, Davide Frey, Pascal Molli, Achour Mostefaoui. Spray: an Adaptive Random Peer Sampling Protocol. [Technical Report] LINA-University of Nantes; INRIA Rennes – Bretagne Atlantique. 2015. <hal-01203363>