Reconsider our definition of "World Wide Web"

[nigelmegitt]

This is not a troll issue! The W3C has a strong habit of talking about the web as though everyone agrees what it is. It seems to me from e.g. #53 and other issues that what folk think the vision should be is strongly coloured by the beholder's personal viewpoint of what the WWW is. But those viewpoints are not in fact coincident, and the differences take people in different directions.

I searched for definitions of the web, and none of the first page results came back to a w3.org page.

A bunch of definitions from my first page of search results

• Wikipedia
• MDN
• BBC Bitesize
• Britannica
• Techopedia
• University of Oklahoma (very old page by its appearance and content)
• Collins

There is a W3C definition, it turns out, at https://www.w3.org/TR/webarch/ from 2004:

The World Wide Web (WWW, or simply Web) is an information space in which the items of interest, referred to as resources, are identified by global identifiers called Uniform Resource Identifiers (URI).

In those definitions there's a rough consensus that the web is about interlinked resources/webpages available via the internet. But almost all of them omit anything about people! Many of them talk about web browsers rather than user agents. None of them talk about physical objects, or money.

I think we need to consider if our definition remains useful. Then we will have a common starting point that can provide a useful scoping for the vision and strategy.

I don't think there's anything technically wrong with the 2004 definition, but I don't think it's quite right for thinking about our future. My own broader definition of the web would be something roughly like:

• A set of technologies that connect people, information and devices

Getting the breadth and scope right is obviously important, and I might not have it "good enough" yet. But putting people first feels important, and is what drives e.g. the TAG's work on privacy principles, and our collective focus on accessibility and internationalisation for example.

[frivoal]

A set of technologies that connect people, information and devices

That seems to broad, as this definition would seem to also cover internet, the telephone network, the CB radio…

The web is not just any set of technologies that connect people, information and devices. It's a specific set. The challenge is to characterize that set. And while there's certainly more to it than the old definition covers, starting from URLs does seem to make sense to me.

[nigelmegitt]

That's a good point. Possibly it would be useful to scope the technologies to the application layer, rather than focusing on URLs, though URLs are probably a core part of the set of technologies. 2nd iteration:

• A set of application technologies that connect people, information and devices through the interchange of resources identified by Uniform Resource Identifiers (URI).

[swickr]

I concur with @frivoal's concern about the broader phrasing. @nigelmegitt's revision explicitly naming URIs as the defining characteristic could work for me, however I worry that this phrasing may still be too broad. Explicitly including people I do think useful, however.

Another characteristic I would want to incorporate is the notion of a single web; I read the 2004 definition "an information space" as stating the name of a single instance.

The World Wide Web (WWW, or simply Web) is an information space connecting people[, information,] and devices in which the items of interest, referred to as resources, are identified by global identifiers called Uniform Resource Identifiers (URI).

[frivoal]

URI

Nitpicking, but https://url.spec.whatwg.org/ takes the view that

Standardize on the term URL. URI and IRI are just confusing. In practice a single algorithm is used for both so keeping them distinct is not helping anyone. URL also easily wins the search result popularity contest.

Are we in agreement? Neutral? In Disagreement?

[nigelmegitt]

I'm fine with URI. Sometimes they are properly not used for locating resources but for identifying them. Accuracy is important in this use case.

[cwilso]

Note feedback in https://github.com/w3c/AB-memberonly/issues/52#issuecomment-790352207, particularly the following points:

• "web pages", typically consumed through as browser. These are typically served from an HTTP server, but sometimes instead from a filesystem.
• The "web of data" that relies on URLs serving as identifiers and anchors into further data. This is generally RDF-based, very widely used, sometimes incorporated in "web pages", and generally not consumed in browsers but rather by agents such as crawlers or reasoning engines, and can be generated by autonomous devices and monitors as well as people.
• "web applications", typically consumed through a browser, retrieved from the internet
• "web technology applications" (thinks like "miniapps", "browser apps", "W3C widgets", ...) typically built using the same technologies as web pages and applications, but installed locally rather than running on a server.
• "e-books" content that is packaged and distributed (like "web technology applications") in similar ways to web pages. These are often consumed in software that isn't called a web browser, although it does many of the same things with differences primarily due to the different content it is mostly designed for.

[chaals]

Note that the list in the previous was intended to be examples of things in scope, not a complete definition. (Trust me on the intention - it turns out I wrote it :) ).

[darobin]

I definitely agree that it's worth defining what the web is. I'm not entirely convinced by the proposals in this issue, however.

The smaller concern I have is that I'm not sure that throwing "that connect people" into the definition is sufficient to make the definition people-centric, or even to start accounting for the role of people in the web. I agree with the direction of travel, just not that it goes far enough.

My broader concern is with the centrality of /(U|I)R(L|I|N)/ in the definitions. I think that we need to distinguish two things: what the web is for versus how the web happens to be built. I am concerned that we are confusing the best (or perhaps only) way to implement a thing with the thing itself. In most cases, the distinction doesn't matter, but being clear about the difference is important whenever we need to reflect upon potential iterations on our implementation (and more broadly think about architecture).

URLs are very important, there's no question about that. I would contend that the important property that URLs have is (as per the TAG), that "a URI alone is sufficient for an agent to carry out a particular type of interaction." URLs are about doing things, and they provide a universal way of acting on the (web) world that has strong disintermediation properties (which can arguably be improved upon, as discussed here). Not all of it has worked out (auth and payment as currently, for instance, break this URL-sufficiency architectural expectation — 401/402 don't really work) but there is work that might fix those issues (eg. UCAN or ILP). It's conceivable that any implementation of the web's goal will have to include something very much like URLs. But that doesn't imply that every information space that uses URLs is the web. We need to get the causality right.

I have a longer piece that provides more background on this, but I think that if we're going to support constructive conversations about the future, the web needs to be defined in terms of what it is for, what it is trying to do in the world. URLs aren't a mission, they're an architectural consideration. With that in mind, the definition I have been using is:

The Web is the set of digital networked technologies that work to increase user agency.

If you want to increase user agency — understood in a capabilities sense of more things that a person can do and can be — then having a universal and sufficient interface to resources that people can act on is a pretty damn good idea. But URLs aren't why and they aren't enough. URLs won't tell you why you need accessibility or internationalisation. (If you're curious and patient and don't mind reading early drafts, there's a longer discussion of the idea, an early intro to the thinking for a community working on improving the web, and lengthy background notes on the ethics underlying user agency.)

[frivoal]

Taking a step back: what do we want to use this definition for? I suspect that what we're most interested in is getting a sense of scope for W3C, and I am not convinced this necessarily needs a definition of the web. And figuring out the scope doesn't necessarily mean that we have a neat one-sentence or one-paragraph definition of that scope. It means, when faced with a topic, being able to tell whether it belongs or not.

If that's not the goal, what is?

[darobin]

That's a great question @frivoal, and here's why I think 1) it matters and 2) it's not just scope.

I think that Lisa Nakamura gives a great first step:

I teach a course at the University of Michigan called "the Internet is a trash fire," and I don't have to explain to anybody what that means… We put up with this for a long time; we don't seem to know anything different.

I believe that Nakamura is correct. It's not just that the Internet (and by extension the web) is a trash fire but also everyone knows that the Internet is a trash fire. People might not about everything that's part of the trash fire because of how big it is and how many parts of society it touches or because of how they may be insulated from some of its effects, but it's a consensus position.

When you start asking people what they think a not-trash-fire Internet would look like, things are a lot less clear however. In my experience, many people don't know. Quite a few will say that they'd like it to be run by companies that are less bad than the ones we currently have. I understand the sentiment but it's not much to go on ("less bad" doesn't tell you where to go) and I don't think that it's even theoretically possible to get good behaviour under the current conditions, even the best of intentions are almost guaranteed to do more harm than good.

Tech experts don't tend to fare much better. Many seem to have some kind of native envy, but again that really doesn't tell us much. People have peeves and the such but frankly it's very rare that I hear someone talking about the web and think that they have some idea of where we should be going.

The web isn't something that gets finished, which means that we need to proactively develop a sense of direction. I think that we have a lot of the (absolutely critical) incremental/janitorial stuff being worked on, but I believe that the purpose of this document is to help go beyond that.

Personally, I think that's what the definition is for: to provide a north star. Personally, that's why I'm spending as much time on my end trying to figure out what the web is. It's the work of building a vision: setting up the mental tools to figure out what's good and bad with what we have and what we need to change and how. I'm not at all claiming that what I've come up with is great or will get buy-in from anyone else, but I do think that that's the shape of the work.

Perhaps one way to think about it briefly is that the definition should answer the question: WTF is it that we're building here?

[TzviyaSiegman]

It seems to me that we are saying that we need to provide a definition that is both narrow enough to be not a trash fire and broad enough to be evolving. A definition like

The Web is the set of digital networked technologies that work to increase user agency.

is too broad to "not a trash fire". I think that defining the Values will help us get to not a trash fire more than defining the Web will.

[darobin]

I beg to differ: if you understand user agency (which is our term) to map to capabilities (which are about human agency: what people can be and can do in pursuit of lives worth choosing), as seems pretty straightforward, then it's essentially been a leading foundation for not-trash-fire for a lot of people whose job is very directly about human flourishing. (Looking for a short reference, this looks like a fine historical overview.)

I'm not claiming that my sentence is the best way to capture this (it certainly is deliberately hegemonic, which I'd expect to see disputed) but it anchors the web project in a body of work that is very values-aligned with our practices and that has been professionally refined over the years. I think that we might benefit from being more reliant on other people's expertise.

[cwilso]

I'm not sure who the "our" of "our term" is, and likewise, not entirely sure I understand your definition of "user agency", but I have concerns about merely using such a short and ungrounded definition. (I'm not disagreeing with you that "the Web is about increasing user agency", I'm just saying there needs to be more definition for that to be a useful grounding.)

By way of example: of course we will all agree with the statement "the Internet is a trash fire". The problem, though, is that each of has a different, nuanced understanding of what that means - that is to say WHAT, exactly, about the Internet makes it a trash fire. They are very likely NOT the same definition; we all head-nod at the statement, but we have different pictures in our heads about what's on fire.

I would disagree quite strongly that " I don't have to explain to anybody what that means… " if you want to reach any deeper agreement - and get better alignment, to work in a united fashion toward some better lack-of-trash-fire - you DO have to explain what it means, and effectively define a shared view of what's on fire. If, on the other hand, you just want people to nod their heads in agreement that the Internet (or Web, in our case) is a trash fire, you can just make the statement and stop there.

I'm not sure what the description is - and it's certainly not MY description - but I would expect that it would be intended to be a definition that dominates what the W3C does. I expect that's what you meant by hegemonic?

[mnot]

I was talking about this question with @slightlyoff last year (?) and he mentioned this article.

What I take from it (and I think he did as well, but please correct me if I'm wrong) is that the Web is an information space -- but we need to take that much more literally than most people read into it. It's territory; it has features, it has inhabitants, and it's the only existant at such scale. Because URLs connect parts of the space together they have a central role, but it isn't solely defined by them.

What W3C defines is the physics of that space, in terms of what's possible, what's easy, and what's difficult. While the inhabitants have free will and agency, selected (not all) aspects of what they can do can be either enabled or limited by the software that implements W3C specs.

All of that means that answering a question of whether a proposal, technology, spec, etc. is Web-related boils down to how much it applies to / benefits / is intertwined with that space. Sometimes this is an easy call: HTTP, HTML, URL, CSS. Sometimes it's speculative development that if it pans out can be a vital part of the landscape (Semantic Web / Linked Data); other times it's marginally connected to the point of opportunism (Web Services).

[cwilso]

I think we should explore better defining what "World Wide Web" is - but probably not in this round.