Another history of the Web from its architecture

another history of the Web
from its architecture
heterogeneous assemblages and open futures
Alexandre Monnin - Fabien Gandon (Inria, Wimmics)
“Computer scientists have ended up having to face all sorts of unabashedly metaphysical
questions. (…) More recently, they have been taken up anew by network designers wrestling
with the relations among identifiers, names, references, locations, handles, etc.,
on the World Wide Web.”
(Brian Cantwell Smith, On the Origin of Objects).

25th anniversary, yeah!
• Very happy to be here to celebrate the 25th
anniversary of the Web!
• We tend to take the years 1989, 1990 and 1991 to
mark its birth.
• We celebrate the birth of the Web every 5 years
• 1989: 20th anniversary, 2010 : idem, 2011 : idem
• 2014 : 25th anniversary, 2015 : idem, 2016 : idem
• 2019 will mark the 30th anniversary!
 So that eventually we seldom abstain from celebrating the Web but maybe
that’s a good thing and I wouldn’t have it another way.
 Also shows the Web is a project and started as such

architecture of the Web?
a naming system (UDI/URI/URL /URN/URC/IRL/IRI)
http://wimmics.inria.fr
communication / protocol (HTTP)
GET /team HTTP/1.1
Host: wimmics.inria.fr
representation language (HTML, XML, JSON,…)
Fabien is a member of
<a href="http://wimmics.inria.fr">Wimmics</a>
F

architecture of the Web?
• Those elements (defined by Christmas 1990)
where originally designed by Tim Berners-Lee
and still provide the basic building blocks of the Web.
• A special group was formed inside the W3C to ensure that despite the
evolution of the Web, these principles would still hold so that the Web
« doesn’t break » (Henry S. Thompson).
• Yet, these principles also have a history, and quite an interesting one – albeit
rarely told we want to say a few words about it.

disclaimer: Internal history
- more akin to STS than traditional history
- “follow the actors”
- more concerned with ontological question (the
status of the entities that comprise the
architecture for the web) than with « social »
explanations
- neither opposed nor incompatible with
external viewpoints

Obviously!
• Main protocol : “HyperText Transfer Protocol” (HTTP)
• Main language : “HyperText Markup Language” (HTML)
F

Obviously!
« Web » presented at 3rd ACM Hypertext conference
1991 (San Antonio, Texas)
F

Obviously!
• “On 6 August 1991, Tim Berners-Lee posted a summary of the
World Wide Web project on several internet newsgroups, including
alt.hypertext, which was for hypertext enthusiasts. The move
marked the debut of the web as a publicly available service on the
internet” (excerpt from Cern timelines).
• Berners-Lee and Cailliau have both acknowledged influence from
early hypertexts creators.
• Especially Bush and Nelson – despite Bush not having really
produced a hypertext system (cf. Alexandre Serres, “Hyperexte”).
• Berners-Lee met with Nelson and mentioned him in his account of
the early years of the Web (Weaving the Web, 1999).
• etc.
F

the Web never was an hypertext

Nope!
« Web » only demoed at 3rd ACM Hypertext conference
1991 (San Antonio, Texas)
F

Nope!
Systems connected to the Internet at the time where not
hypertexts. What the Web purported to do might thus
have involved something different (image ©CERN) F

Nope!
The Web retained only a few functionalities from
hypertext systems
F

Nope!
The Amaya browser developed by Inria did keep the
“edit” functionality but only locally since it had become
a client-side-only feature. F

Xanadu® did (purport to) support…
• bidirectional links (links should never break!)
• transclusion (Nelson: “A link connects two
things which are different. . A transclusion
connects two things which are the same”).
• copyright, micropayment, and security
• version-management features
• content management features
Xanadu® was never implemented to begin
with… (up until very recently at least - 2014).
The Web = Xanadu® in the wild!

Initial perception
• “WorldWideWeb: Proposal for a
HyperText Project”
T. Berners-Lee, R. Cailliau, 1990
• Hypertext conceptual legacy
“a distributed hypercard”
• Documentary perception
“the universal library”

Yet…
• The use of the word was much broader than reflects in canonical stories.
• First European Conference on Hypertext (INRIA, France, November 1990)
attended by Berners-Lee a: heterogeneity of “hypertexts” and researchers
profiles.
• Helps understand the fact that the Semantic Web was first mentioned as early
as 1994.

enquire
• Bidrectional links?
“The links were bidirectional, in that having added it once, the reverse
link automatically showed up. The relationships applied not, of course,
to the cards, but to the people, things, documents and groups
described by the cards.”
• “He recognized that a system similar to ENQUIRE was
needed, "but accessible to everybody."[4] There was a need
that people be able to create cards independently of others
and to link to other cards without updating the linked card.
This idea is the big difference and the cornerstone to
the World Wide Web.” (Wikipedia)

other systems known for their influence
• System 33: (Steve Putz, Xerox PARC, early 90s) : basis of content
negociation (conneg) a fundamental functionality fo the Web.
• X.500 : the counter example of the “Directory Information Tree (DIT)”
i.e. a hierarchical organization of entries
F

post deployment evolutions: the case of security
• 1990 : HTTP not secured
• 1994 : HTTP + SSL for Netscape Navigator
• 2000 : RFC 2818 “HTTP Over TLS”
HTTPS = HTTP + TSL + adoption
F

even more influences• Usenet : worldwide distributed discussion system.
• UNIX : multi-users & TCP/IP networking protocols
• Free Software Foundation & GNU General Public License (GPL)
• etc.
F

Other competitors of the Web:
Internet-based information systems
• Alex
• Archie
• Gopher
• WAIS
• Prospero
• etc.

compatibility and extensibility
• influence on URL/URI scheme “universal syntax”
• interoperate and finally encompass rivals
• proof of retro-compatibility and flexibility
• becomes keystone
F

the underlying programmatic approach
• RPC Remote Procedure Call (70’s):
• Since 1970, RFC 674, 707, 1058
• NODAL interpreter language for accelerator controls (argument)
network of invocation
• Declarative document languages (mid 80’s):
• SGML and the documentarian culture
• LATEX as a programing approach to documents (CERNDOC)
programming the generation of representations
• Object-Oriented technologies (90’s):
• Object-centric
• Class hierarchies
resource oriented architecture
F

« Agencements? » (assemblages)
« Ce qu’apporte la notion d’agencement (et de manière plus explicite
que ne pourront jamais le faire les notions de dispositif,
d’assemblage ou d’arrangement), c’est une réponse à la question
des sources et des modalités de l’action. (…) Je propose, pour
clarifier une terminologie qui n’est pas exempte d’ambiguïtés,
qu’on réserve la notion d’agencement à la combinaison :
arrangement + action spécifique, et que l’on qualifie l’agencement
pour désigner le type d’action spécifique qui est en jeu. On pourra
ainsi parler d’agencement technique, politique, scientifique, etc.,
chaque agencement donnant forme par les cadrages qu’il organise à
un certain mode d’action collective. »
(Michel Callon)

finding back the hypertext
After the Web took off many hypertexts systems were designed to re-
implement « lost » functionalities:
- Hypertext systems (ex.: HyperG)
- Search engines (esp. Google)
- Web 2.0 platforms (towards a fragmented read-write Web: annotation,
comments, backlinks on CMS and blogging platforms – trackbacks, etc.)
- Languages (ep. XML for transpointing – a feature of Xanadu®, annotations, etc.)
- Wikis (collaborative writing with content and version control) !
- Archives (Internet Archive to keep bits of memories from the Web)

XML
“It might be fairer to say that many of the things Ted Nelson thinks are
important tend to fall outside the scope of things XML is very good for.
Certainly the question of the technically correct way to do persistent
linking, annotations, etc. in XML documents (or the WWW in general,
for that matter) is still open to debate, despite HyTime, TEI, Xlink, etc.
...”
"Ted's place in history is secure because he asked more important
questions than just about anybody. I think he usually offered the wrong
answers, but questions are more important.“
(Tim Bray)

One exemple: Google
• The assemblage produced by Google (which rests on
measures, infrastructure, database tools, business
models, etc.) partially transforms the Web into a
hypertext.
• When Google launches its crawlers which then invoke
resources and dereference representations that are
kept in it index.
• These representations are treated as stables pages
(ignoring content negotiation for example) bounded to
one another through their incoming links.
• Google « networks » the naming system of the Web
thanks to its algorithms (the PageRank and all those
that now support it).

• Once pages are inside an centralized
index, one can safely establish
relations among them by using links
instead of just pointers.
• Of course this is a very partial
stabilization of the living Web in
Google’s index but a very efficient
one performed through the usage of
tools such as MapReduce.
Crédits illustration : Aurélien Bénel

multiple Web experiences
• personalized answer and lost reference
• machine learning and filter bubble
• loosing the universality of the resource / multiplying representations

Beyond « agencements » : l’entre-capture
• Concept borrowed from Isabelle Stengers:
« [le] caractère non symétrique de toute entre-capture, couplage toujours latéral et
partiel, (…) ne fonde aucune « logique » commune : les façons dont l'un compte, du
fait de l'entre-capture, pour l'autre et l'autre pour l'un sont a priori parfaitement
distinctes » (Cosmopolitiques, tome 7, Pour en finir avec la tolérance)
• Strange mereology where the whole (Google + the architecture of the
Web) is smaller that the parts (the architecture of the Web) thanks to
the reductive action of the assemblage (the former only “captures”
bits of the latter, assemblage reduces complexity as much as they add
their own – see Latour’s paper “La société comme possession - la
"preuve par l'orchestre"”).

REST
fixing the a posteriori essence of the Web

standardization and the birth of the W3C
• 1994 W3C to secure Web architecture
• from implementations to standards
through investments of forms (Laurent Thévenot).
• First standards
• between 1994 and 1995 URIs → URLs, URNs, URCs and IRLs
• February 1996 HTTP 1.0
F

« Web resources »
• But controversy on splitting of URIs into URNs and URLs.
• Distinction between documents and things
• What is « on » the Web and what is « outside »
• Largely inconsistent and reactivated by Semantic Web
• Standards revised as soon as 1997 (HTTP 1.1) and 1998 (on URIs - meaning
there was a URL IETF RFC for barely three years).

toward REST
• 1995: Roy Thomas Fielding to shed light on the principles behind the Web
• Main editor of HTTP 1.1 standard (1997) and URI standard (1998)
• REST (for REpresentational State Transfer : Fielding’s thesis 1995 - 2000.
• REST made explicit the core design principles that were discovered in the
Web (realist point of view).
• He also made sure that such principles were to be found in modern Web
implementations (constructivist point of view).
• he actively participated to the rewriting of the standards so as to make them
compliant with the said principles
• principles which had been abstracted and refined from the existing standards and
which were later rewritten due to their (from then on) insufficient RESTfulness!

chicken or egg?
t1) Http – Model for REST
t2) REST – Principles extracted from the model
t3) REST – Refined, becomes a norm
t4) Http – Partial instance of REST
 Fielding strove to make the Web coherent with itself.

“> Logically, REST really had to predate HTTP 1.1 in order for HTTP 1.1 to be so RESTful. No?
No. That is more of a philosophical question than a logical one. HTTP/1.1 is a specific
architecture that, to the extent I succeeded in applying REST-based design, allows people to
deploy RESTful network-based applications in a mostly efficient way, within the constraints
imposed by legacy implementations. The design principles certainly predated HTTP, most of
them were already applied to the HTTP/1.0 family, and I chose which constraints to apply
during the pre-proposal process of HTTP/1.1, yet HTTP/1.1 was finished long before I had the
available time to write down the entire model in a form that other people could understand.
All of my products are developed iteratively, so what you see as a chicken and egg problem is
more like a dinosaur-to-chicken evolution than anything so cut and dried as the conceptual
form pre-existing the form. HTTP as we know it today is just as dependent on the
conceptual notion of REST as the definition of REST is dependent on what I wanted HTTP to
be today.”

Roy T. Fielding/
Tim Berners-Lee
Credits Paul Downey

Abstract resource vs concrete representation
What is a URI naming or designating?
- the Web has neither version control system nor build-in archiving system
(everything disappears of the Web by default)
- content do change over time (diachronic changes)
- content do potentially change every time a URI is queried (URI can identify
service, be the result of many calls akin to mashups, « pages » contain
counters, vary through content negotiation or personalization, etc.
(synchronic changes)
- pointers can no longer be dereferenceable (« links break »)

URI & time
• http://ns.inria.fr/fabien.gandon#me
• http://lemonde.fr
• http://météo.fr/Paris
• http://www.timeanddate.com/worldclock/meetingdetails.html?year=
2015&month=12&day=1&hour=7&min=0&sec=0&p1=181&p2=248&
p3=224
F

Paris Wikipedia Article 2002 has a URI
F

definitly a hierarchy in the core concepts
URI > HTTP > HTMLuniversal scheme content negotiation
F

“7.1.2 Manipulating Shadows. Defining resource such that a URI
identifies a concept rather than a document leaves us with another
question: how does a user access, manipulate, or transfer a concept
such that they can get something useful when a hypertext link is
selected? REST answers that question by defining the things that are
manipulated to be representations of the identified resource, rather
than the resource itself. An origin server maintains a mapping from
resource identifiers to the set of representations corresponding to
each resource. A resource is therefore manipulated by transferring
representations through the generic interface defined by the resource
identifier.”
(Roy T. Fielding & Richard Taylor)

a network of resources
• a network of potential invocation / network of interaction
• indirection, disconnectedness, separation: is at the heart of the Web.
• Deep Web, Dark, Web

There cannot be a map of the Web that
doesn’t fix what is maps through interaction
If a resource is an abstraction, then:
- There are no links between existing pages but URI than can be dereferenced
(or not) to representations
- No central authority guarantees that representations are available making the
Web a decentralized system
- The same locus of computation will give simultaneously give birth to differing
http-representations (accessible content) , more often than not, these
representations will turn out to contain different URI thus drawing potentially
as many graphs in parallel to one another.

a Web of amateurs?
Noted computer scientist Alan Kay once said “the Internet was done so well that
most people think of it as a natural resource like the Pacific Ocean, rather than
something that was man-made. When was the last time a technology with a scale
like that was so error-free? The Web, in comparison, is a joke. The Web was done
by amateurs.”
One Stack Overflow contributor, Stephen Crawley, gives this beautiful reply : “In a
sense he was right. The original (pre-spec) versions of HTML, HTTP and URL were
designed by amateurs (not standards people). And there are aspects of the
respective designs ... and the subsequent (original) specs ... that are (to put it
politely) not as good as they could have been. (…) However, the web has
succeeded magnificently despite these things. And all credit should go to the
people who made it happen. Whether or not they were "amateurs" at the time,
they are definitely not amateurs now.”

Web Web Want
• The “infrastructure” (Susan Leigh Star, see also Jean-François Blanchette) by
definition remains hidden.
• As researchers we have a responsibility to draw attention through our work on
unseen, little known or actively obfuscated (through assemblages)
infrastructures. Be they material or quite abstract – made of standards for
instance, turning them into “matters of care” (Maria Puig de la Bellacasa).
• Potentialities are lying there and as they may help to perform different futures.
• From the knowledge of the architecture of the Web some people, for example,
are working to “re-decentralize the Web”.
 See the SolID (Social Linked Data) initiative.

WebArch between abstraction and infrastructure
• Since the beginnings of the Web HTTP-representations of resources could take
many shapes from:
- PDF files (whose access has to be maintained, they are much less durable than their
paper counterparts!)
- HTML « documents » or mashups,
- Dynamic WebServices,
- Compositions of Webservices, etc.
• There is undoubtedly a tendency towards more abstractions and, at the same
time, more infrastructure (see also “IaaS”, “Infrastructure as a service » = cloud
computing)
• More abstract resources actually translates into more infrastructure (more
infrastructure is needed to actively maintain a Webservice than a PDF file).

Web of Resources & (meta)data:
the way to distinguish resources and links
data
person document
program
metadata
F

back and forth
t-1) Internet – RPC
t0) RPC – Http
t1) Http – Model for REST
t2) REST – Principles extracted from the model
t3) REST – Refined, becomes a norm
t4) Http – Partial instance of REST
t5) Http – RPC
t6) RPC – Internet
Internet of Things – Web of Things
F

a Web of calculus
“We have the potential for every HTML document to be a computer — and
for it to be programmable. Because the thing about a Turing complete
computer is that … anything you can imagine doing, you should be able to
program.” (Tim Berners-Lee, 2015)
This is actually what Ethereum has been doing for a few years now and
branded as genuine Web 3.0.
F

(real containers are much less eco-friendly!)
Allan Sekula and Noël Burch, The Forgotten Space (2010). Still image from documentary film.

From the Web We Want to the Web We can Afford in
the age of the Anthropocene
• Two narrations :
- That of untamed progress towards more technological development (DAO, AI, etc.).
- The Anthropocene, climate change, new climate policy, etc.
 Ever heard of COP 21? If you made it to this venue, I guess you did!
• Not so much incompatible narration as too much compatible
(tech industry is fostering social and climate collapse).

New media theorists (interested in the
social sciences) have already noted the
impact of digital infrastructure on the
planet
(see for instance the work of Jussi Parikka: A
Geology of Media and The Anthrobscene).

Downscaling the Web
How about computer scientists? Attempts are being made to
provide Web access (in particular in Africa) without Internet
connection (see the work of Christophe Guéret)
a) since the Web is not just about connection but also
disconnection, indirection, 404 errors, etc. how can we
imagine a WWA (a Web We can Afford – at least for
some time) without being blinded by the tech salesmen
around us and the commodification of research?
b) Can we go as far as to imagine a Web that would no
longer rests on digitality and its unsustainable
infrastructure?

NOT TAKING OF GREEN COMPUTING,
D*MNIT!!!!!!!!!!

Another history of the Web from its architecture

More Related Content

Similar to Another history of the Web from its architecture

More from Alexandre Monnin

Recently uploaded

Another history of the Web from its architecture