-
Notifications
You must be signed in to change notification settings - Fork 15
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Who are the standards voluntary for? #177
Comments
I have always understood this to mean that they are voluntary for the (potential) implementers of the technology being specified. Anybody is free to implement a specification as written, or to create a product which ignores some (or many) normative requirements of the specification, or to not implement it at all. Unsurprisingly, deviating from a broadly agreed specification is likely to cause interoperability problems, as the specification is meant to facilitate and enable interoperability, but W3C does not compel implementers to support or conform to any particular specification. We do not ensure that users of the technology (I am using the term broadly, including implementers of other technologies that have a dependency on the one we are specifying) have a guaranteed right to use products as they wish would exist. They need to use products as they are, which may or may not match their desires, and may or may not match specifications, may or may not match input they gave while into the spec writing process. They depend on the technical and business decisions of those who make products available, not on W3C's (non existant) ability to compel companies to ship any product any particular way. To me, that's the nature of voluntary: W3C attempts to achieve desirable outcomes by attempting to facilitate the emergence of a common understanding of what is best for everyone, taking into account wide input not limited to implementers, with specific care towards wide and horizontal review to make sure nobody's needs are accidentally ignored. But ultimately, W3C does not have the power to compel anyone to abide by what W3C says. So technology providers should listen to broad input, but can ship whatever they want. Overtime, the consensus building process and the resulting specifications are expected to influence implementations towards better behavior, and at the same time, the reality of what implementations already do or are willing to do influences specifications to stay close to reality. Users of technology may wish that the products they use or depend on were different, and to the extent these products do conform to W3C specifications, they may feel that a specification that does not correspond to their desires is depriving them from the choice of using a product that would work differently. It is possible to consider a different model, where specifications would be binding and require implementers to do what is written, not out of consensus and self-interest, but because of legal requirements. That is the case in many other industries. However, that would require a radically different way of working. First, it's not clear where we'd derive the power to compel technology providers to implement certain things. We're not in a situation where companies need a government-granted exclusive right in order to operate, which could be made subordinate to respecting the spec. Nor are the technologies we develop (primarily) purchased in a regulated market where only compliant implementations may be procured. Maybe we could ensure that spec are patent encumbered, and Members could build some shared patent portfolio to go after anyone who doesn't comply, but that seems rather far from the way things are done now. Even if we somehow got there, the ability to compel implementers would call for totally different process. I cannot imagine that consensus-based chartering or group decision would work in such a world, nor Formal Objections as they exist today and how they're resolved, nor, to be honest, the patent policy. Different industries have have legally binding standards, and the processes and cultures are nothing like what we know in the world of W3C/IETF/WHATWG/Unicode… If the decisions were binding, there would probably be immense pressure to limit what we do specify and who can have a word. Efforts like WPT would likely not work: instead of testing as much as we can to foster interoperability, we'd probably test as little as we can get away with to demonstrate compliance. Same for specs: if they were binding, there would be a lot of pressure not to specify anything that's not strictly necessary. It's in that sense that I think this belongs in the vision: the fact that our specifications do not force their implementers to be compliant, and that we rely on consensus and self-interest instead, is core to how all our processes. So we should be upfront about it. |
I think @darobin is making a point that @frivoal is missing a bit in the reply. (This debate is familiar to people who participate in Acceessibility Guidelines work at W3C) There is a current discussion in the HTTP WG at IETF about the various URL standards and who has to follow what that is illustrative: https://lists.w3.org/Archives/Public/ietf-http-wg/2024JulSep/0281.html Following @darobin's suggestion that we begin with the principle that we cannot force Google or Apple to implement a standard, I think that is a good starting point. W3C standards are adopted or ignored (or more usually a mix) based on choices made by implementors, guided by market and other pressures. W3C does not enforce conformance to WCAG, but many countries do in various ways. We don't enforce conformance to a security policy on implementors, but we deliberately put them into technology specifications that the market can effectively enforce. We explicitly design around patent law in various ways that influence implementation. I think it's helpful to recognise that there is something more than a purely voluntary system here, although I don't have a snappy way of framing it in text right now. |
Maybe I am missing something, but my reaction was driven by the comment about border-radius, which to me, seemed off topic and confused / confusing. The fact that people who use a technology standardized at W3C (CSS) and implemented by browsers can chose not to use it, to me, is orthogonal with "voluntary standards". I mean, I guess, in a the sense that it is not mandatory for every author of every website in the world to use border radius, you could consider it voluntary to use it or not, but I do not find that definition useful. It's merely an optional feature of a programing language. You don't have to use it. I also don't have to send SMS messages (and if I do, I don't have to use accented characters), nor do I have to watch TV (and if I do, I don't have to turn on the subtitles). But my freedom not to use a product or certain of its features doesn't telecom or broadcast standards aren't voluntary standards. It may be silly to ship a browser where borde-radius doesn't work or works differently from what the spec calls for, but it's not forbidden. The same isn't true for 5G antennas or car braking lights. Following the standard there is compulsory, not voluntary. "we cannot force Google or Apple to implement a standard" is true, but this has nothing to do with Apple and Google specifically. It's not about them being big tech who can do what they want, while smaller entities have play under different "involuntary" rules, which is occasionally what's implied when questioning whether our standards are really voluntary. It has everything to do with the fact that implementing/following (or not) a specification is what's voluntary for the implementer of that specification. Whoever implements a browser gets to decide whether to voluntarily do what specs say about things that should be implemented in a browser. The fact that there's not that many browser implementer is unfortunate, but not doesn't change that. When we're talking about ActivityPub, or TTML, or RDFa, or other things which are meant to be implemented by not-a-browser, then what Apple and Google think is of no more relevance than anybody else, and these standards are voluntary to follow (or not) by implementers of those technologies. |
I think there are a number of things here that are incorrect. Like XML before it, RDF, RDFa, TTML, were all designed with the idea that they would be implemented in browsers, as well as other user agents. When a small group of the user agents known as search engines started implementing schema.org in a way that determined which resources people would be likely to discover, their crucial position in the infrastructure meant that choices about whether to implement them were no longer really voluntary for the developers of web content who seek to actually make a living from their work. I believe @darobin's not too subtle implicit point is precisely that "what Apple and Google think" about which specifications they like IS of far more relevance than almost anybody else (there are other stakeholders who have a similar strength, but unlike e.g. web developers, or people who make videos, they are few in number - readily countable in practice). Indeed, that while some stakeholders can genuinely choose what to do, those choices effectively constrain millions of others, and thus describing the standards as "voluntary" is quite likely disingenuous or plain misleading. You can make cars without braking lights, or with green ones. But choices made by a small set of actors (primarily governmental) mean that you usually can't sell many of them. This seems to be a good parallel to our situation, and yet you describe those cases as compulsory which is a clear "antonym" to "voluntary". As I noted above WCAG, which gets tied in decade-long knots over questions of browser implementation, is widely mandated by government-backed obligations, thus even more obviously matching your examples of "compulsory" standards. (For clarity: having worked and argued since the 90s to make WCAG primarily a technical standard I am not suggesting W3C shouldn't produce it - I think it is one of our great success stories despite being a very long way from perfect). |
TL;DR: to answer the original question: W3C standards are voluntary for the (would be) implementers of said standards. We operate in a market without de jure regulation. Implementers ship (or not) whatever it is they want. That would be true if W3C didn't exist. It remains true after W3C exist, regardless of what we write—or don't—in our specs. This makes the standards voluntary, as opposed to mandatory/compulsory. The fact that W3C does exist and does produce specs enables implementers to coordinate with each other and with other stakeholders to achieve greater product interoperability, and it draws in a much broader range of input (from other members, users, horizontal groups, wide review …), which makes the specs reflect many different viewpoints and be more balanced and thought out than anyone could do in isolation. That makes the standards useful. They're still not mandatory, and implementers can still do what they want with their own products. Some product design decisions may be less useful than others, and choosing to release something which doesn't work with the rest of the ecosystem may not be wise, but it's allowed. W3C has no power to disallow it, and has processes set up acknowledging this absence of power to compel anyone to implement anything.
Of course reality is constraining. Millions of people who would like to use certain products have to use what exists, not what they wish existed instead. Gaining a collective ability to demand that certain actors provide what you want may be useful, but it would not be voluntary. Quite the contrary.
But that's very different indeed. You actually cannot market cars without braking lights. It's not just that you won't sell very many of them because they're less useful or because some influential stakeholder dislikes that. You just cannot do it at all because it is a regulated market where entities with de-jure power have made it forbidden to market cars unless they meet certain standards. Cars without braking lights, or without a catalytic converter, are not just less useful; they're not allowed. Toys with lead paint are not just a questionable product design which few people favor; they're not allowed. In the Web's case, W3C does not have to power to make it illegal to ship a browser without border-radius (or without https, or with the wrong kind of outline algorithm, or with insufficiently isolated cookies…), nor to make it illegal to ship a would-be mastodon client which uses YAML instead of JSON-LD as its ActivityStream data format. While governments could do that, they have not chosen to. Building and selling non compliant products may be silly, but it's allowed. And sometimes, it's not even silly. Against the W3C's will, browser engines shipped HTML 5 engines, not XHTML2. They could do so because following standards—or not—is voluntary. And it turned out not to be silly. W3C publishes EPUB as a spec for ebooks. That doesn't prevent Kindle from running on Mobi and AZW instead. Book publishers have an interest in publishing books as EPUB (which is standard), or Mobi or AZW (which aren't). Publishing as epub0 or WebBook is also allowed, though it may not gain you much market share. Following W3C standards for EBooks is voluntary, and you can chose to do something else. Choosing poorly may not get you very far, but you can. Even on the user side, if you want to consider a web site author as an implementer (of sites, not of browsers), anybody is free to make a website which uses XHTML2 instead of HTML, or VBScript instead of JavaScript… It may not be useful, since the rest of the ecosystem won't know what to do with such a thing, but it's allowed. WCAG is an exception indeed, as some governments have made it compulsory to follow in some circumstances. I would not call WCAG a voluntary standard, at least not in all circumstances. But that's atypical for W3C work, and fundamentally W3C is not structured to produce things which are legally binding. |
Yes, it is illustrative of IETF standards being voluntary too. If some pieces of software don't follow RFCs, this may cause some interoperability challenges and make some unhappy people reach out to mailing lists. Some spec violations may just be bugs, and some may be (consequences of) attempts at evolving the technology. Maybe implementations will get fixed. Maybe the spec will evolve. Maybe there will be a failure to reach consensus. In any case, nobody's product is getting pulled out of the market by authorities due to non compliance, nobody's license to operate is being revoked, nobody is being fined, nobody is going to jail. |
I think that @darobin is bringing up a point that is mostly about the subtext of 'voluntary' as opposed to the actual intended meaning. If that assumption is wrong, stop me; but I don't think that it tracks in this case.
Sure, and the W3C has no forcing function to affect a company's free will. From the perspective of the Vision for how the W3C works, all the standards work is voluntary. You can argue (I think you are?) that there exist other forces which mean that the work companies take on as a result of things becoming W3C standard are not always voluntary. But, c'est la vie! Just because we say our consensus standards are voluntary doesn't mean that there aren't other forces that affect what companies bring to us. To take your argument to it's furthest conclusion, it is not voluntary for a publisher to work on attribution measurement because marketers want it to exist before providing advertising spend. That would be a bit far-fetched, and while a bit of a straw-man, I think it illustrates why the word voluntary still holds properly in this case. The publisher can do that work of their own free will, even though it is influenced by outside forces. Yes, some companies have more power than others, but that is not a W3C bug or feature, we can't solve capitalism in the vision document, and IMHO contributor's work is ultimately voluntary.
We have come across the idea of 'defining the terms' frequently in the vision document conversations, and I will state the same thing I said during the breakout session at TPAC - as an engineering first organization, we're bad at taking a step back and letting the Vision be qualitative and interpretive. I would urge us to do that here. Yes, interpretation can lead to disagreements, but this document is meant to be a guide, not a process doc. |
@mgendler said
That's a core question. As I understand it (see #126) this document is intended to guide fairly important decisions e.g. by stating what is, and what is not, relevant for us to do. I suspect there is an underlying philosophical issue here. But perhaps we can deal with this by not saying "voluntary", before we manage to "fix capitalsim". I am not a fan of "defining terms" as much as finding an effective way to say what we mean and understand. But unfortunately, I don't have a snappy phrase to offer here. |
Can someone brief me on the AB's fixation on "fixing capitalism"? What does this have to do with anything? Is there some kind of Milton Friedman reading club that I should know about? I'm quite comfortable with qualitative and interpretive, I pretty much live in that neighbourhood. But I think that that makes it okay to mislead people? The situation is quite simple:
I'm not saying that's wrong (though, you know, some people might think it looks an awful lot like "fixing capitalism"), but it's hard to see how it's voluntary even if I tap into French Philosophy for qualitative interpretation. It's pretty clear that there are two different categories of standards at play here. Which, again, is fine. We just shouldn't lie about it? We largely abandoned the de jure/de facto distinction because it didn't prove itself to be very relevant. We also gave up on telling people that we didn't do standards, only at best some recommendations. We have a clear enforcement mechanism: coordinating implementers. Maybe the simplest fix is to remove "voluntary" since it's not a truthful depiction of what the W3C does (or envisions itself doing)? |
I'd argue it's a personal fixation more than an AB fixation, but that's just me :) I'd be happy to join any reading club with the gathered company, though maybe we can find something better than Milton.
I'd disagree with the simple distinction you're trying to express. I think it requires a little more nuance on the second scenario. The W3C can (and does or plans to) write standard rules, generally speaking, those are informed greatly by work from the browser vendors who would need to implement those standard rules if they are to truly be standards in practice. When the browser vendors implement those standards, they become rules of the road for anyone functioning on that browser. (I feel like I'm deeply oversimplifying here, but I'd like to not write a full thesis this early in the morning.) If I'm understanding you correctly, the fact that they become rules of the road means that they are no longer voluntary for all participants on the web. While I agree that's the case, the actual implementation of the standard, which falls on the browsers, is voluntary. And as far as the standards process works, that is what means for a standard to be voluntary. We know this because different browsers implement different things all of the time. Yes, in the vast majority of cases, when things reach standards they are coordinated implementors, but that is not always the case.
Sorry for taking this out of order, my brain is waking up and I scrolled down before I scrolled up.
Agreed, and then disagreed. It's not in my capabilities to do the later, but I don't think that precludes me from making statements within that context.
I think here-in lies the rub. I'd be happy to take a gander at new language (though in order to get the vision to statement, I think our intent is to limit new bikeshedding and just try to fix the old bikesheds), but it is hard to envision a change to our language based on the comments when I agree with the current phrasing. |
I think you're making it more complicated than it is or needs to be. If we follow your reasoning, the prohibition on theft is voluntary because the police or judicial systems decide whether and when to implement it. They actually do, but it is still considered mandatory (as opposed to many unenforced laws) because the enforcement is effective and credible. And that's the test: whether a rule is voluntary and what the rule is entirely defined by its actual enforcement and the beliefs of the people upon whom it is enforced. This cuts both ways:
The question is therefore: are there standards made by W3C such that their rules are effectively and credibly enforced on a significant constituency? And the answer is: yes, absolutely. The Process is designed in a way that supports it: implementers have a core say, you're not supposed to agree to an unimplemented standard. We constantly (you yourself @mgendler!) talk about it in those terms. We say things like "sites will be prevented from tracking" or "browsers will enforce strict partitioning." Those sentences make no sense if the standards are voluntary. (And browsers aren't the only enforcement mechanism. We also have standards about standards that are enforced on standards makers by horizontal review.) I'm not saying that that's necessarily a bad thing. As the editor of some of the relevant documents, I think that those non-voluntary standards are worthy contributions to the world. But we shouldn't misrepresent the nature of our work to the world. |
Who would I be if I wasn't? :)
But the actual enforcement in this case doesn't live with the W3C. I recognize what you're saying entirely, and yes, I agree that I talk about the actual mechanisms of power balances and how they affect the web ecosystem plenty and think they're incredibly important. I think I'm uncomfortable with the idea that people would participate in W3C work non-voluntarily. What we choose to work on is voluntary, and I think that's the sentiment I see being shared here. I don't think "by building consensus on voluntary global standards" implies that there is no enforcement mechanism on the standard anywhere in the world - after all, what kind of global standards would they be if everyone ignored them? I think we're assigning two distinct meanings to this phrase (close, but distinct) and there's, as with so many of the past issues on the Vision, no way to satisfy both of our meanings without adding about 4 paragraphs of context which is not in the interest of the document itself. |
This is essentially about W3C's role as regulator (as in @mnot's blog post). I'd be in favour of adding to the Operational Principles, as suggested in #192 (comment). Given one of the purposes of the Vision is to "be opinionated enough to provide a framework for making decisions, particularly on controversial issues", I think more work is needed to identify what those principles really are. |
Well, you wouldn't be me either, for one.
Ah, so you're saying that if the legislative branch doesn't do its own enforcement then the law is voluntary? You know that's not true! It's also not correct in the specific case:
I really don't think that this is what the text is intended to say — if it is, we're even farther from consensus on this. I think that the use of "voluntary standards" is often confusing in how it's used, but I don't think that people ever mean "standards that one voluntarily participates in setting"; rather, people mean "standards that are adopted voluntarily." Some W3C standards are adopted voluntarily by those who deploy them; some are not. |
W3C doesn't do any enforcement. Period. Its recommendations are voluntary for implementers to adopt. As XHTML shows, they're very free to ignore those recommendations. Once they're implemented, they constrain the behaviour of other parties -- authors, publishers, users, etc. That's not voluntary at all, unless you consider walking away from the Web a reasonable option for them (and they do have other options). However, it's probably not a good idea to talk about those constraints in terms of 'voluntary', because that word has fairly deep roots in various legal authorities on standards, and it doesn't apply to non-implementers. Generally these concerns are talked about as 'policy issues' which is pretty poor I know. |
Happy to be corrected on this, but I'd add that while there is interest in using or building on existing W3C standards (as well as those from other bodies), they remain voluntary for W3C specifications in development. If W3C standards were enforced within the W3C, we'd see a different wiring of W3C standards and a fundamentally different ecosystem. |
I think you're focusing on the word "enforcement" as used in regulatory settings, i.e. as in "enforcement action," which is a specific meaning (and I think primarily one in common law countries). From an institutional analysis, enforcement is what makes the rules apply effectively — what Ostrom calls the "OR ELSE" clause. And, well, "or else it doesn't work" is very strong enforcement (low monitoring cost/high rate of success). I don't have an opinion as to whether we should give precedence to regulatory vocabulary or to institutional vocabulary, in fact I don't care at all. What I do care is that we:
That is why I framed the issue in terms of "who is this voluntary for?" If we say it's voluntary, we have to be clear that it is voluntary for browser vendors. If, however, we claim that we make standards for a wider community, then we cannot claim that they are voluntary for that community. It's possible to update the Vision to clarify that it only really applies to our implementers class. I don't think that that's the best outcome, though, but I can live with it if it is clearly stated. |
This is an attempt to a minimum addition in the same "voice" as our other principles to satisfy what appears to be a rough "can live with" consensus overlap in opinions in issue #177, per the original issue author's comment: > It's possible to update the Vision to clarify that it only really applies to our implementers class. I don't think that that's the best outcome, though, but I can live with it if it is clearly stated. Edit suggestions welcome by the chair, or other suggestions to simplify or shorten without losing essential aspects or changing the "voice". Wordsmithing and other aesthetic editorial suggestions are not invited at this time and are likely to be rejected. This PR is not a next-Note blocker, however is made in a good faith attempt toward at least a partial (if not complete) resolution of a recent issue with commentary from many participants.
Right. W3C is very much a "voluntary consensus standards setting" organization in the sense used by the US government in particular. See for example: https://obamawhitehouse.archives.gov/omb/fedreg_a119rev/ Many of these documents refer back to (Carl was a long-time W3C participant and AC/AB member, many discussions of the US-centered voluntary standards systems reference it) A recent book that apparently goes into some depth on the history of "voluntary" standards ... looks relevant too. Whether or not the US voluntary standards system is appropriate for the rest of the world or for the future, (see for example https://www.eastwestcenter.org/sites/default/files/private/ps066_0.pdf) the term is well-understood by the people founding W3C, and I'd be wary of the Vision trying to re-define it. Referencing one or more of these resources might be helpful to readers however. |
See also this post for a discussion of the in-force legal authorities on voluntary standards. |
Inspired by Tantek Çelik's PR w3c#194 Attempts to capture the core definition from Mark Nottingham's essay https://www.mnot.net/blog/2024/03/13/voluntary Addresses issue w3c#177
Inspired by Tantek's idea of adding "Voluntary to Implement" as an operational principle, this is my suggestion and attempt to capture the gist of @mnot's post:
I considered expanding it a bit more, e.g. to clarify why this is a good thing (allows failure, reduces pressure for control over the standard in favor of pressure to get meaningful consensus, etc.), but figured I'd start minimalist. Suggestions for improvement / rewrites welcome. |
@fantasai I think it's a worthy attempt, but the problem I see with that approach is that it sweeps the actual decision under the rug. I think it's always easier to work from an example so let's compare two imaginary specs:
When we say "voluntary to implement" we are hiding the fact that there was a decision, made in W3C with the goal of it being enforced by user agents, that the exact same person who was an author with voluntary rights in one case is now an attacker with no rights in the other. That's not an innocuous decision, and it certainly isn't a voluntary one for authors who are disenfranchised between the two. Now, there are plenty of good reasons why we should sometimes treat some people as attackers instead of authors. But are we sure that there are no bad reasons to do so? Are there no grey areas where we might have to have hard conversations? Shouldn't we be principled when we decide to make compliance non-voluntary? |
I admit to not being smart enough to follow the abstract philosophical musings here. But I'm intrigued by the Rick Rolling Prevention scenario : Browser Product Managers at various companies building products on Chromium / WebKit / Gecko got lots of complaints from users about being rick-rolled and requests for a feature to block surreptitious links to Rick Astley's song . The PMs got together at W3C and developed an API to block them. Now whoever owns the rights to Never Gonna Give You Up is making a stink because Evil Big Tech is conspiring to destroy their business model of tricking people to play the thing on YouTube and collecting $.001 or whatever each time. 😉 Where is the grey area here? What hard conversations do we need to have about whether the Rick Rolling Prevention API is "voluntary" or not on whom? Or maybe we could talk about a real example currently before the AC: The proposal to deprecate the Vibration API because 2 of the 3 main browser engines stopped supporting it because of privacy concerns. How does that real situation shed light or not on the question of "voluntary" standards? In my perspective, it indicates a "decision, made in W3C with the goal of it being enforced by user agents" is not a useful way of framing the issue. W3C didn't make a "decision" with the expectation it would be "enforced" by browsers. Rather there was consensus at one time across a group of users/website developers that a Vibration API would meet a set of user needs, browser engines could implement it, and users would appreciate the feature. Reality and/or perception changed, and that consensus collapsed ... so the question is whether to deprecate the Recommendation immediately or fix the API ASAP. "Enforcement" just doesn't seem like a useful framing here. The API was always "voluntary" to implement and use, once was widely implemented and used, and when issues arose it was less widely implemented and used. W3C is the venue where users and implementers get together to find consensus ways to solve problems, and ratify resulting interoperable, accessible/internationalizable/secure/privacy-preserving solutions as Recommendations. When the real-world interoperability and privacy criteria are no longer met, it deprecates the Recommendation. What is the problem with calling that "voluntary", and what is the benefit of twisting our heads into knots about what else to call it or what corner cases that doesn't cover? |
The problem with calling it voluntary is that a very small group make a decision that impacts a very very large group, and determines what they can choose or not. I think we should recognise, as part of our vision of the organisation, that we effectively provide a discussion forum where decisions can be made that have a concrete impact on the choices available to the web "as a whole". Part of the point of the multi-stakeholder model we embrace is to ensure that various people with different perspectives can come to the table, and say their piece, so those in a position to collectively enforce a decision hear it, even if they then decide to do something that doesn't satisfy those people's goals. (Examples of this happen all the time - it is IMHO a key piece of what we do at W3C, and a key reason why many members are here). It seems reasonable to remove the word "voluntary", because it is somewhere between dsiingenuous and outright misleading for anyone who isn't broadly familiar with how standards are made, implemented, and applied, and because removing it doesn't seem to impact our vision of the Web we are building. I think how to describe the way that standards are made and implemented and the impact they have on the web is a separate part of the puzzle, and is sufficiently complex that letting one adjective carry the whole idea is not the right approach. So I propose to open a separate issue. |
This is where @chaals and @darobin lose me: In what sense does W3C "determine what they can choose or not"? W3C has done a lot of stuff that was ignored by the larger web community (XHTML as opposed to "HTML5" is the obvious example). Implementers choose which bits of a W3C Recommendation to use , and popular profiles / partial implementations of W3C standards such as Mastodon become de facto standards. There are certainly examples of W3C Recommendations that "determine what [very large groups] can choose or not" because they are enshrined in government regulations (such as WCAG). But it wasn't the small group at W3C that determined that websites around the world MUST support WCAG, it was the very large group of users who benefitted and the large number of government entities that found it useful to mandate what W3C "Recommended". I had thought of "voluntary consensus standards" as a well-understood term of art among standards professionals (in the US anyway ... see the examples in #177 (comment) ). But I'm not wedded to the world "voluntary" in this context, given the pushback. But I do have to ask why some think this term is worth the bike shedding effort it is consuming ... it seems to be standing-in for some larger issue. If someone can identity that larger question, be explicit about whether you feel strongly enough about it to block consensus on a Vision statement, and can propose language we could plausibly get broad consensus on that provides a vision of a Web or W3C that embodies that issue, please clarify. |
A small group at W3C discusses, say, adding an interface to browsers to support DRM modules, and agrees on how to do it. A slightly larger group of people implements that. And the world toes their line or builds their own ecosystem. A small group at W3C decides that allowing geolocation to stay on in a browser, and follow a user looking at a different window or none at all, is a bad idea. So you can't do it (unless you write an app for one of the two "native" platforms whose browsers can't do that). ...
I suspect "voluntary consensus standards" is a term of art known to standards professionals in the US - thousands and thousands of them. And perhaps thousands and thousands more across the world. That's a very narrow slice of the people for whom we work, and to whom it would probably be helpful to explain ourselves. I doubt that most people who work on web technology (as in vendors other than content developers) understand the term clearly. Certainly my experience is in working in well-known W3C member orgs where most employees can't really explain what it means - nor why it matters.
I don't think it is standing for a larger issue. I think it is about how we represent what we do, especially since the stated aim for this work is to be good enough to make decisions about what work we (won't) do in the future. There's plenty enough there to keep people concerned about framing of statements, accuracy, and things that seem fairly small on their own.
I have suggested we say that we "work on standards" without calling them voluntary (nor anything else). I don't know if that could get consensus, but it seems the least-difficult path I could find. It leaves out @darobin's suggestions to talk of enforcement on purpose, in an effort to find consensus. |
As discussed before, 'voluntary' has a well-established meaning in standards and the law. Rather than trying to change the meaning to suit our intuition -- thereby creating confusion in any discussions with policymakers and other parties who use the term correctly -- we should clarify what it doesn't mean so that people who share our intuition are aligned with the wider world. To work from @fantasai's proposal, this might look like: Voluntary Conformance: Our standards are voluntary to implement. This means that implementers decide what to ship: we cannot require anyone to adopt our standards. Other parties (e.g., authors, browser users) are constrained by these decisions, and so W3C's role is to assure that they do not have unintended consequences, through mechanisms like horizontal review. |
This implies that "a slightly larger group of people" implemented DRM for videos in the browser because W3C told them to. This is close to being the exact opposite of what happened: a number of people had decided on implementing DRM for Video on the web, so W3C (reluctantly, judging by FOs and appeal) agreed to be a venue for writing related specs, based on the fact that it was happening anyway, and our multi-stakeholder system was likely to help mitigate the concerns.
This, again, is backwards. W3C isn't allowing or disallowing anything, since it's permission is not needed here. Implementers, through their own research and through discussions in horizontal review and with multiple stake holders, have come to realize handling geolocation that way is bad, so they don't. They coordinate about how to do that at W3C, which does broaden the discussion to many more stakeholders, but there is no W3C decision which compels either way those who had the ability to allow or disallow that behavior in their implementations in the first place. It isn't about W3C decisions which gets enforced by browsers. There are implementer decisions which are observed and documented by W3C. Along the way W3C provides a venue and a system for a broad set of people to provide feedback, but in the end, implements ship what they want, and W3C's specs can reflect that or call for something else or ignore the matter altogether. People, who independently of W3C, have the power to make the choice of what to implement, retain the ability to make that choice, regardless of what W3C say. That's what makes conformance voluntary. People who are consumers rather than providers of those technologies do get a venue and a process for providing input, do not get to make W3C enforce their preferences onto the providers of the technology. @mnot I feel that that definition seems a little narrower than what is usually mean by this, but I think it does capture what is essential about it. +1 |
I don't see much philosophical musings going on, though referring to other people's participation that way is certainly offensive. You are describing a process in which a small group of powerful people get together to remove rights from another group of people, and then to enforce that removal of rights. One moment they are "implementers" or "authors", with the right to voluntarily adopt technology, the next they are "attackers" and have those choices made for them. Dear me, what gray areas could there be? Is it a problem if the browsers agree to enforce Rick Rolling Prevention but one of those browsers shows a rick roll to its users every five minutes? Is it a problem if it puts entire industries at risk? Could this one raise censorship concerns?
If you write a prevention (or similar) spec and it's not enforced by browsers, you can't get to Rec. I don't understand what's a "philosophical musing" about this, it's just the Process. |
I think that this is a step in the right direction, but it still hides the little shell game in which the implementer set includes browsers and authors in some cases and only browsers in others. With this definition, we can make anything "voluntary" just by changing the list of who it's voluntary for — which was the issue to boot. Iterating on that: Preferring Voluntary Conformance: Our standards are mostly voluntary to implement. This means that all implementers decide what to ship: we do not require anyone to adopt our standards. In some cases, however, one group of implementers (e.g., browser vendors) will impose adoption on other parties (e.g., authors, browser users). The W3C's role in such cases is to assure that they do not have unintended consequences, through mechanisms like horizontal review. |
I think that for some things that's still not quite true. Especially in the "horizontal" areas - making things that are easily internationalizable (is that a word?), ensuring that it is possible to build things accessibly, ensuring there aren't obvious fixable security issues - we would prefer the ability to impose conformance, even encouraging third parties (organisations, regulators) to impose enforcement. |
@chaals I think there is subtlety involved. We can look at different cases:
All three of these are clear-cut cases of rules that get enforced — though there's a gradient of distance of the editing to the enforcement and how core to the Process proof of enforcement is (in 1 it is taken for granted, in 2 it needs to be proven to proceed, in 3 it's just a consideration, in 4 no one cares). I think that it might be true that generally people prefer (4) unless it wouldn't work, in which case they prefer the others (not necessarily in order). Put differently, we wouldn't use 1-3 if 4 worked. Note that this is an important distinction to draw out. Some members (e.g. quite vocally MOW) have been advocating for us to only do voluntary standards (i.e. group 4) and to never impose standards on authors (2) or even on our other standards (1). I disagree with their position, but it's coherent. We should be equipped to discuss it. |
@darobin I don't think a 'preference' comes into it. Iterating: Voluntary Conformance: Our standards are voluntary to implement. This means that all implementers decide what to ship: we do not have the power to require anyone to adopt our standards (although others might). However, when successful our standards do constrain what other parties do (e.g., authors, browser users), so W3C attempts to avoid unintended consequences through mechanisms like horizontal review. |
Open-stand.org says "Standards are voluntarily adopted and success is determined by the market." |
The document currently states:
(Emphasis mine.)
Voluntary means "done, given, or acting of one's own free will." A good example of a voluntary standard would be
border-radius
. If I want to useborder-radius
it's right there, if I don't then I just don't use it. You can't make me.But that is not true of all W3C standards. If I don't want my website to support standards for mixed content or same-origin policy, for instance, then I have no say in the matter. Maybe those standards are arguably good things to impose upon websites, but they aren't voluntary in any meaningful sense for the estimated 1 billion websites out there.
I think that it's problematic for this document to actively mislead people about how the W3C operates. It would be problematic if we advertised our work as being about voluntary standards only for someone to join and discover that that isn't always the case.
An easy fix would be to define the term. I don't have a good suggestion to offer here because I find the notion confusing and misleading, but that is no indication that there isn't a good solution! In previous conversations, @cwilso has stated that W3C standards are voluntary because "no one can force Google or Apple" to implement them. Maybe that provides a good starting point to work from?
The text was updated successfully, but these errors were encountered: