A film by AtomGraph

The web is still at 2.0.

The web was designed to link any piece of data to any other. Most data is still locked inside applications, where no link can reach it. We build software that puts it back on the web, for people and agents alike.

Enter
  1. 1968Objects
  2. 1970Relational databases
  3. 1974SQL
  4. 1979Model–View–Controller

The web runs on software that is older than the web itself.

Each was designed for programs on one computer. None was designed to link data between computers.

1989

Then came the web.

The first browser, in 1990, could edit pages as well as read them.

  1. 1996CSS
    style
  2. 1999RDF · XSLT
    data · views
  3. 2006Linked Data
    links
  4. 2008SPARQL
    questions

The web got tools of its own — one for every part of a page.

Each is an open standard, designed to work across the web.

  1. 2004Web 2.0

Web 2.0 was built with the tools from before the web.

When websites learned to take our input, they were built from the objects, databases and controllers of the 1970s. Each website kept its data in its own database. You can see the data on the page, but it has no address of its own, so nothing can link to it.

Still 2.0.

The real problem

The problem was never the web. It is the way we build software.

Every application is a new program that hides its data inside instructions no one else can read. In 2017, The Atlantic warned of a coming software apocalypse: code too complex for anyone to understand. Now machines write that code faster than anyone can read it.

  1. 2001The Semantic Web
  2. NowThe agents

In 2001, the Semantic Web promised software agents that would act on our behalf. Now they are here.

And they find walls: pages made for human eyes, and data locked behind private interfaces. So translators are being written again, one for every application.

The other way

Nobody writes a new browser for every website.

One generic program reads them all. We build the same for applications: one generic platform, where every application is data, on the web itself. People and agents use the same addresses, with no adapter in between.

An agent is a user agent.

Next

Applications described, not coded.

When you ask an agent for an application, it doesn't have to write code. It describes the application as data, in a few dozen statements, and the platform runs it. Those statements combine with everything else on the web.

The application is data.

Company

AtomGraph is the company. LinkedDataHub is the platform.

LinkedDataHub is our generic platform: open source, under the Apache 2.0 licence, built on the web's own standards.

The full argument is in First Principles of the Web, free to read online.

Selected work and partners —
NXP SemiconductorsThe Royal LibraryOntopicSataliaDydra

Talk to usExplore LinkedDataHub

Still 2.0 · a film by AtomGraph

Sources

Every claim on this page, and where it comes from.

1968–1979
SIMULA 67 (1968), Codd (1970), SEQUEL (1974), Reenskaug MVC (1979)
1989
Berners-Lee, Information Management: A Proposal, CERN
1996–2008
W3C: CSS 1996, RDF 1999, XSLT 1999, Linked Data 2006, SPARQL 2008
2004
“Web 2.0” named at the O’Reilly Media Web 2.0 Conference, San Francisco, October 2004
2001
Berners-Lee, Hendler & Lassila, “The Semantic Web”, Scientific American, May 2001
2017
James Somers, “The Coming Software Apocalypse”, The Atlantic, 27 September 2017
Generic software
First Principles of the Web, Ch 21: “the browser. One client for every website. Nobody writes a per-site browser.”
“Still 2.0”
First Principles of the Web, Ch 11: a falsifiable prediction. It is refuted by “a fused-architecture deployment whose state is machine-consumable at web scale without a compensating adapter layer.”
Film
Archive footage credited in frame; era texture simulated and declared; narration is a synthetic voice.