Towards World Wide Web 2.0

The following is an email response to a friend at schema.org regarding an initiative called The World Wide Graph (w2g).

Thanks for the link! RSS is a great protocol (which I don't intend on
redesigning or replacing) and the interfaces you linked me to seem to
make great use of RSS -- I do, however, hope to push the web, in its
entirety, towards a slightly different direction. RSS is a great way
of subscribing to content on specific sites, not necessarily as a way
to see what's happening (the derivative) across the entire web. For
instance, I probably wouldn't use RSS if I wanted to see all the
places across the web where a specific "tag" (e.g. a person) was
mentioned. What I imagine with w2g is something much more akin to a
public "facebook tag", where you can register a universal tag (across
the entire web) and be notified when any website on the internet uses
*your* tag, or a tag you follow. Many of the implementation details
are described on https://graph.global.

TL;DR -- That the universal document index (the web) and entity
database should be a public and inherent property of the World Wide
Web, implemented in the style of DNS, and not a centralized system
maintained privately by search engines. W2g is an attempt at creating
such a universal entity database.

Inspiration:
My central goal is to further Paul Otlet, et al's, vision and head
toward an amalgamous World Wide Web (a Universal Knowledge Repository)
freed of arbitrary, discrete "document" boundaries. A world wide web
which isn't experienced through discrete, independent "web pages"
(even if the unit of content an author publishes is still a
"document") but instead the interaction and cooperation of content
within them. More pragmatically, I am suggesting the emergence of a
public layer over the exiting web which enables users (independently,
via applications such as browsers or GNU utilities[1], in a
decentralized fashion; i.e. without a search engine) to make queries
which generate dynamic views/compositions of pertinent content. A
world wide web wherein protected access to an underlying, versioned,
universal indexing system is granted as an inherent property of HTTP
itself.

I think the Internet Archive has the capability of achieving this
reality. And with it could come a new generation of tools for
consuming and republishing this content, and for browsing,
visualizing, and discovering paths through and between knowledge. And
not just knowledge -- our comprehension and understanding of
knowledge[2].

Dissolving the boundaries between discrete documents (via semantic
tagging and publicly accessible indices), and creating a public,
distributed universal entity graph are just the first steps to a much
brighter future. With this foundation in place, we can begin
calculating and dynamically composing paths through knowledge. We can
dynamically generate curricula; responsive sequences of resources for
individuals to learn new things. We can establish previously
unexplored connections between disparate branches of knowledge,
dynamically fact check knowledge against its neighbors, automatically
infer citations, explore epistemological and pedagogical provenance of
knowledge, and even craft self-organizing systems which improve our
classifications and produce new knowledge over time.

Toward this end, I'm working with Ted Nelson, Drew Winget (stanford
digital library), the Internet Archive, and some folks at Google on
planning achievable first steps. The first of which, is exploring how
a *public* universal entity graph will be created using existing
standards, which supports such a future.

From the literature I've read and the experts I've queried, RDF (as a
general concept), schema.org (as an implementation of RDF), and
services like w2g (physical implementations of open/public universal
entity graphs) seem to be one approach towards achieving this first
step.

Without context, this whole mission likely seems a far-fetched fool's
paradise -- that may be so; while there remains a chance, I intend to
devote my entire life towards the mission of progressing *public*
systems for universal knowledge, and I'm confident over my lifetime I
can move the needle, or at least continue contributing towards laying
a foundation which empowers others to do so.

I'm sure you're already overwhelmed by my "essay" response to your
simple question (which hopefully by now I've answered): "What problem
[am I] trying to solve". If you are passionate about these topics and
interested in people's thoughts on the matter, I've began compiling a
compendium (something of a curricula) on Universal Knowledge which
(based on historical precedent) will likely outspan my lifetime --
https://docs.google.com/document/d/1upjXuPM_rVbZcFm3aqdg-T8fHCClxnfMAMVTIaqt-_4/edit#

Beginnings: Originally when I started my phd program, I was working on
an open conversational search engine (similar to DuckDuckGo) called
GnuAsk. It addressed two main contentions with existing search
engines. First, I was annoyed with being limited to a single query as
a medium for interfacing with a knowledge base. At the time, every
query was evaluated independently (unlike markov chains, previously
searches did not inform future searches) and converging on the correct
query was an arduous process. I imagined an alternative engine with
which users conversed using natural language. The engine was able to
disambiguate via conversation and used models to generate accurate
queries based on the progression of the user's dialog. The result of
each conversation was a directed chain of resources (from which
intermediary answers had been obtained), and ultimately an answer to a
specific question. This "list" (DAG) was then compressed into a single
URL called a GNUrl (like a tinyurl) which preserved the provenance
trail of the conversation and could be shared with friends as a single
link. The second reason I was distraught is, search engines are able
to find and compile lists of relevant "documents", but in the average
case, their job stops there. What should happen is, the engine should
direct the user to the exact piece of knowledge they are after, and
other relevant information (they might not have known existed). To
their credit, DDG Zero-click, Google Now, and others are beginning to
address this concern, albeit in a centralized fashion.

Centralized "portals" or services (i.e. google and gnuask) which only
expose a small fraction of their infrastructure, and through
restrictive APIs, are good first steps, but I now realize I was
building features at the wrong level. My goal has since matured to
address how we might be able to enrich HTTP as a protocol and
ecosystem, rather than continue the entrenchment of these monolithic,
centralized services. Someday, I hope the world wide web becomes more
like a public library, where its citizens have access to the
underlying indices and knowledge graphs -- after all, such sentiments
were a great early inspiration for the web. When this day comes, I'll
look forward to the amazing unix-style utilities which filter, map,
and reduce over the internet as a whole, and which empowers everyone's
computer, and all its applications, with ubiquitous search.

Footnotes:
[1] I wrote an essay on this called, "What the Browser is Missing"
(https://michaelkarpeles.com/essays/philosophy/what-the-browser-is-missing.html)
which imparts partial blame to browsers and browser vendors, but I
believe an important part of the equation lies with the HTTP protocol
and how we as a society have grown to (mis-) use it.

[2] Imagine a utility or interface like
https://michaelkarpeles.com/math.html, which is capable of producing a
heatmap overlay to illustrates one's relative understanding / coverage
of different mathematical domains).