The Abdication of the HTML Standard

GMGruman writes "The end of numbering for HTML versions beyond HTML5 hides two painful realities, argues Neil McAllister. One is that the HTML standards process has failed, becoming a seemingly never-ending bureaucratic maze that has encouraged the proliferation of draft implementations. That's not great, but as all the wireless draft standards have shown, it can be managed. But the bigger problem is that HTML has effectively been abandoned to four companies: Apple, Google, Opera, and Mozilla. They are deciding the actual fate of HTML, not a truly independent standards process."

Comments

Those Who Ship Win

[eldavojohn]

But the bigger problem is that HTML has effectively been abandoned to four companies: Apple, Google, Opera, and Mozilla. They are deciding the actual fate of HTML, not a truly independent standards process.

This reminds me of something that was promoted in a book I reviewed:

those who ship win

It's that simple. If this armchair talking head who wrote this article chastising the standards process were to magically code up a browser that better empowered me, a software developer, to deploy code to users that ran to my satisfaction then his standards would be realized first. And I might be tempted to use it and ask my users to use it so we can get good functionality.

Duh.

Back when the standards were still in flux (and still are) I was using Google Chrome to enjoy an Arcade Fire experiment that used many HTML5 elements. And guess what? I started using Chrome and the implementation of their perspective of the standards gained just a planck constant more marketshare.

This guy can sit around and complain all he wants but for better or for worse: those who ship win.

Re:Those Who Ship Win

[Saint Stephen]

Gee, that sounds like - a De Facto standard. Like MS Word .doc format! Guess evil is in the eye of the beholder.

Re:Those Who Ship Win

[Anonymous Coward]

While I hate eldavojohn as much as anyone, I don't see anything in his post that was saying this fact is 'good' or 'evil', just that it's a fact. And honestly, he's right. If you design by independent committee, that committee needs to move at the pace of development or it will be ignored. If software companies are putting out releases faster than the committee is putting out standards, then the committee is worthless. This is ultimately the reasoning behind the move to non numbered releases, as it at least gives the standard a chance in hell at succeeding. You can't expect companies to slow development to wait for some voluntary, independent group to tell them what they can and can't code, you need to speed up the group to match the developers.

Re:Those Who Ship Win

[bberens]

It sounds to me like we've finally gotten around to admitting the truth about how the system works instead of wasting a bunch of time and money with the standards process. It's not as if a particular browser saying it was HTML# compliant was really meaningful. You still had to test every feature and work around the bugs on a per-browser basis. IMO nothing really has changed except the illusion.

Re:Those Who Ship Win

[rjstanford]

Yes, yes it is.

The difference being that the group behind the de facto standard sees high value in being consistent, predictable, and having that pseudo-standard very well documented, because without those facts nobody can create content for them to consume.

With the .doc format, there's high value to Microsoft in obfuscating the "standard" documentation as much as possible since they both create and consume the documents.

Big difference.

Re:Those Who Ship Win

[Geoffreyerffoeg]

ODT is as much of a de facto standard. If you give me an ODT file that conforms to the standard but triggers bugs in OpenOffice.org, what good is it? I'm not going to spend days setting up an OOo build environment, learning whatever awful framework they use, and bisecting this bug in order to read your few paragraphs.

The problem with .doc is not that it's a de facto standard -- all standards that are worth anything must be de facto at least as much as they are de jure -- but that it's a bad one, because it's hard for any program that doesn't share MS Word's internal data structures and algorithms to implement (because a .doc is, to first order, a memory dump of Word's data). HTML doesn't work like that, and it's hard to make it work like that if you tried.

Re:Those Who Ship Win

[theaveng]

>>>HTML has effectively been abandoned to four companies: Apple, Google, Opera, and Mozilla.

Sounds like a lot of FUD to me. It used to be:
- 1999 and earlier: No HTML standard existed and Mozilla Netscape just willy-nilly added new features (blink tag for example).
- 1999 and later: Ditto Microsoft once their IE became dominant. IE5 and 6 were browsers that complied with nothing, and even today still cause problems for web designers.

Better to have four companies talking to one another and hashing-out HTML5 and HTML6, rather than the old (a) chaos of Netscape producing non-compliant features or (b) Monopoly of MS-IE. We don't want to have another Format war like HD-DVD v. Bluray on the internet. We want consensus first, even if that slows progress a little.

Re:Those Who Ship Win

[drinkypoo]

Seriously, why do I even bother with this site?

Because you know that the majority are both more temperate and less vocal than the baseline stupid douche?

Re:Those Who Ship Win

[morgauxo]

So, if I made a browser that used a flavor of html which made your job easier you would automatically begin coding your html for it? Really? And who would your customer be?

The sad thing for web developers is that it doesn't matter whose html standard is techically better or which one better enables development. It's which one/ones are being used by your target audience that matter. Otherwise you are coding a site just for yourself! It really comes down to a browser marketing issue, not an html standards one. Whoever markets their browser better gets to set the standard.

Re:Those Who Ship Win

[unwesen]

Nothing wrong with de-facto standards, if they're fully open.

Look at e.g. the Python programming language. The CPython implementation is the de-facto standard implementation, and the language specs actually refer (or used to) to the implementation saying if in doubt, that implementation wins.

Yet there are other, mostly compatible Python implementations out there, and nothing - not patents, nor secrets - stops you from starting a new one.

Re:Those Who Ship Win

[gig]

It's nothing like Word get a grip.

Apple created canvas, submitted to W3C, Mozilla submitted changes, canvas was standardized, then Apple invested significant engineerig resources into changing their canvas implementation to match the standard.

If you want an academic standard with no real world use, XHTML 2 is available for your masturbatory needs. The Web needs a practical HTML standard that documents how you DO write HTML, not how you theoretically SHOULD write HTML.

Re:Those Who Ship Win

[gig]

It's not 4 companies, that is BS:

1) in this context, Apple is the WebKit open source project

2) dozens of vendors use WebKit, including Google, and there are many contributors

3) Mozilla is a foundation

4) Microsoft and Adobe are also part of W3C, although they sometimes had to be dragged kicking and screaming, but that just shows that standardization works

Re:Those Who Ship Win

[Nadaka]

evil? no. Beauty? Death? Petrification? Anti-magic? sure. Evil is in the alignment of a Beholder.

Re:Those Who Ship Win

[jc42]

Having browser makers defining the "standard" is the mess we needed to get out of in the first place. What, do we want to return to browser makers making up things as they go along, like goddamn IE?

There's a long history to this that explains it fairly well. The concept of a "standard" was invented to handle the problem that vendors have always cheated customers by having their own definitions for units of measurement. Thus, food seller have always tried sell a "pound" (or whatever the local weight term was) of produce that was short weight.

Governments finally blocked this by making legal definitions of such units of measurement, and prosecuting vendors who used a unit that was different from the standard. This is, historically, the only thing that has ever worked. Anywhere in the world, if you sell, say, 200 grams of meat or rice, and it only weighs 180 grams, you will be found guilty of consumer fraud. If you try to say that you use your own definition of a gram, that statement would be used in court to show that your fraud was intentional. (There are occasionally parts of the world where such things aren't enforced, and the result is always the same: Vendors start cheating their customers.)

The real problem here is that HTML is not actually a "standard". That is, it has no force of law behind it. Any vendor can violate it at will, call their product "HTML", and have no fear that the government standards body will prosecute them for consumer fraud.

The phrase "de-facto standard" is in fact a statement of knowingly committing consumer fraud. It should lead to prosecution. But it doesn't. That's the real reason that we have such problems. If we had an actual HTML standard, enforceable in court, we wouldn't have such problems. (Actually, we'd still have them, but they'd lead to prosecution. ;-)

How has backwards thinking like this become so prevalent in modern technology? I don't get it.

It's because "When a computer is involved, all precedent is forgotten, and everything has to be relearned." I've forgotten who said that, but it applies here. The computer industry uses the term "standard" for things that aren't standardized at all. Vendors can freely claim their products "standard" when they don't follow any actual standard, without fear of prosecution for fraud. We have "industry standards" bodies like W3C that have no enforcement power, and thus aren't actually defining standards. They are merely defining pseudo-standards, marketing terms that vendors are free to violate at will without fear of any legal consequences.

The entire history of commerce tells us how vendors will behave in such a situation. They'll define their own "standards", using the same terms as the published pseudo-standards such as HTML. They'll do this knowingly, to persuade the public to buy. Without any legal enforcement, they know they can get away with it.

If we want an actual standard for HTML, it has to have legal import. Without this, vendors will continue to behave as they've always behaved. But there is no sign this is being considered, here in the US or in any other country.

So any HTML "standard" is just a pseudo-standard, a marketing term that vendors can violate at will. We can discuss it all we like, but this will have no effect on the vendors. The only way this can change is if government get involved by having their standards bodies declare actual legal standards that are enforced. We have millennia of experience showing that this is the only approach that actually gives us a useful standard. Computers haven't changed this fact; they only led us to ignore it and then complain that vendors are violating a "standard". Of course they are; they're vendors who are trying to sell us something, without any legal constraints on their use of terminology.

Re:Those Who Ship Win

[samuraiz]

Profit motive.

Re:Those Who Ship Win

[guruevi]

Foundation is not for profit, company is for profit. As you may have noticed from recent history, if a company for profit makes a 'standard' it will only be possible to get implemented either by said company or with the blessings (patent licenses) of said company (DOC, ActiveX, MSOffice Open XML, Silverlight, Flash, DECNet, Skype...). When a foundation makes a standard it will be possible to get different implementations without any restrictions (OpenDocument, HTML, WebM/VP8, Ethernet, ...)

Re:Those Who Ship Win

[Nadaka]

A standard that says how things are done is not a standard. It is merely documentation.

A standard sets a standard, an expectation of behavior. If you do not meet that expectation, you are not compliant with the standard. Period.

You can not have a standard that contradicts itself because then, nothing could ever actually implement it. This is one problem with the HTML5 standard, it is unimplementable due to self contradiction. This problem is made worse because now even theoretically compliant browsers will eventually loose compatibility over time as the standard drifts.

XHTML2 not perfect, but it was far superior. I put a lot of effort in preparing for it. Sunk hundreds and possibly thousands of hours developing tools for working with it. and it all came crashing down in the end when W3C made the mistake of abandoning it.

The Standards Really Never Have Been the Standards

[salesgeek]

The W3C has never really had complete control of HTML. Those who write the browser effectively can extend or cripple HTML features at will. Netscape added many new features and everyone simply had to live with the results. IE did some nasty things to CSS and we all had to live with that, too.

HTML *was* simple

[drumcat]

Remember when it was ok to use a "b" tag, and no one scoffed? How about table layouts? It's funny, the new standards aren't always better. This is why a format "of the people" isn't going anywhere. I could teach my grandparents how to edit HTML 10 years ago. Now, not so much. Is that better? I'd argue, no. It's not that editing is hard; it's not. The problem is that we're turning the browser into an application-level container. HTML should be more focused on making layouts easier, and faster. It should not be focused on animation. This is where MS Word has fallen off a cliff. If you want more adoption, focus HTML on what actually is important - layout that's understandable to the masses.

Re:HTML *was* simple

[Anonymous Coward]

Any HTML your grandparents wrote 10 years ago still works fine today, so what are you complaining about?

If anything, HTML5 represents a shift back towards the 'vernacular' - for example, the B tag is officially a-ok for bolded text.

Re:HTML *was* simple

[Rui Lopes]

Remember when it was ok to use a "b" tag, and no one scoffed? How about table layouts? It's funny, the new standards aren't always better.

1. 1) Download the NVDA screenreader
2. 2) Learn about the problems induced with your comment
3. 3) Spread the word!

If you still think it's actually not better, sorry, but you should have 10 blind persons hit you with their canes...

Re:HTML *was* simple

[Hatta]

Remember when it was OK to use an "i" tag, and it worked on Slashdot?

Re:HTML *was* simple

[BitZtream]

We replaced B with STRONG and I with EM which are effectively the exact some thing. If you think its different, sorry you're confused.

Tell me why exactly we needed to change tags?

Explain to me how change the name of a tag from B to STRONG actually made it so your screen reader worked better. I'm pretty sure if it knew the difference between B and STRONG it could do whatever it want with text and wouldn't care if the tag was B or STRONG.

Some douche's in a commitee sat down and decided the B was a display thing and STRONG was a speech thing and they are different and should be handled as such ... except ... they are used for the exact same thing. Same with EM and I.

An intelligent solution would have been much simpler ... 'when people use B they mean to make that part strong and stand out, lets treat it as such.' instead of 'hey, lets create a new tag, that is visually identical to B out of the box, but has a different name, so that way we can identify it ... unlike bold where we have no way to identify it at all!'

EM and STRONG are shining example of retarded changes to a protocol because some people involved in the discussion don't have a clue.

Eh?

[Anonymous Coward]

HTML has effectively been abandoned to four companies: Apple, Google, Opera, and Mozilla.

And Microsoft is where?

Their Internet Explorer is used by most Internet users today ( http://marketshare.hitslink.com/browser-market-share.aspx?qprid=0 )

Re:Eh?

[johnnysaucepn]

Microsoft haven't been active in developing the latest version of HTML5, while the others have. That's all they mean - it's no reflection of the size of the company or their reach to customers, but in their work on the standard.

Re:Eh?

[KingMotley]

They are still there. The article just fails to mention them. Microsoft has contributed a LOT to the HTML 5 specification process. A very large number of test cases were submitted by them, and they contribute during the discussions as well. It's just the author obviously has a very anti-microsoft bias. And for the purposes of the article, the lack of any one company doesn't really matter the principle remains the same.

Re:"not a truly independent standards process"

[Waffle Iron]

Isn't that par for the course? It seems a lot of standards are driven by a few big players who have a strong interest in it.

True. When I read the summary, I thought that four players seemed better than the early days of the web, when HTML was driven by just the pair of Netscape and Microsoft.

finally

[FuckingNickName]

Finally, people are starting to realise (and argue) that today's HTML is no more "open" than Flash. It's just a cartel between a few major tech companies to promote particular implementations of particular technologies in their medium term interest. Apple's canvas is the most obvious culprit. Rather than freeing people from Flash, it gives such a seductive but incomplete alternative (to an already subpar platform) that developers are encouraged to write native Cocoa apps. It's msjvm deja vu all over again.

Re:finally

[BHearsum]

Regardless of who is setting the standard, it *is* an open standard, implementable by anyone who reads the spec. Flash is not. Big difference.

Re:finally

[Henriok]

There's a WORLD of difference between a dictating singularity making a "standard" and then hardly documenting it after the fact (you can include Google and WebM here too) and a small committe comprising four of the world's five implementors doing an openly documented iterative evolutionary development, while doing at least three different implementations in parallel.

Bad Thing?

[ArhcAngel]

Hey! At least a certain monolithic juggernaut ISV that is known for hijacking ALL standards isn't in the top four.

Could be worse

[Thelasko]

At least the standards aren't determined by Microsoft.

What's with all the hate?

[graveyhead]

Last I checked, anyone could submit ideas, corrections, feature requests *RIGHT THERE ON THE HTML5 WORKING DRAFT*. "Feedback Comments" right at the top of http://www.w3.org/TR/html5/

Now, if they ignore your idea, that's almost certainly because it sucks and is badly written. No really, it does suck. Follow the instructions there *carefully*, really think about this feature or tag or whatever you're requesting, and your ideas will get consideration.

Welcome to corporate feudalism

[gestalt_n_pepper]

What's happening with this issue is a microcosm of what's happening in the world. Democracy and the rule of law wither, while wealth, in the form of organizations or a few super-rich individuals control outcomes.

patents, MS

[bcrowell]

It seems to me that everybody is moaning and groaning about what a bad job WHATWG is doing, when in fact WHATWG is just doing the best it can in an extremely difficult environment created by patents and Microsoft.

The confusion with respect to audio and video codecs only exists because of patents. A certain patent-encumbered codec shows up that's good enough, so it gets widely adopted, and then it's impossible to displace it because of network effects. This is not WHATWG's fault.

The html 5 feature that I really care about is mathml, and here it's very, very clear that MS is the bad guy and W3C and WHATWG have just been trying, unsuccessfully, to work around MS. Mathml worked fine in xhtml years ago, but MS never bothered to support xhtml in IE, which would have been technically trivial to do. They stated that their policy was to have independent vendors supply support for mathml rendering via plugins, and Design Science did their best to do that, but MS made it impossible for them to do that in a standard way, because the standard depended on xhtml, which IE didn't support. So xhtml died in the crib, and WHATWG decided to pour the svg and mathml namespaces into the flat html 5 namespace. Kind of an ugly solution, but they had no other choice. Now for the first time it is theoretically possible to write a web page coded in a standard way that has mathml in it and that might render properly in some future version of IE. But meanwhile big institutions are still sticking to IE 6 because they need compatibility with all its bugs, and preview versions of IE 9 have broken mathml support.

The big problem is that commercial entities have interests that oppose the interests of their customers and internet users at large. MS wants users to be locked into their browser through proprietary plugins and bug-compatibility, and they don't stand to profit by supporting features like mathml, which are only used by a relatively small proportion of their users. (Never mind that blind people can access mathml but not bitmapped renderings of equations. Blind people aren't economically important to MS.) Owners of patents on codecs want to harvest licensing fees, and they don't care if that screws everybody else up and makes a mess out of audio and video on the web.

McAllister complains that WHATWG is dominated by a clique consisting of Google, Apple, Mozilla, and Opera. But that clique is basically a list of all the browser vendors, and doesn't that kind of make sense? These are the people who acually need to implement the standard, so of course they should be the ones with the most influence. The only browser vendor missing from the list is MS, which is only interested in subverting standards.

Re:The Standards Really Never Have Been the Standa

[CynicTheHedgehog]

It never had complete control, but it did its job. It established a level playing field and and brought parity (more or less) to four different browser engines. Now that there *is* competition, all four vendors are busy as bees trying to add new features and mimic the new features added by the other vendors. So we don't need a standard per se, as long as we have users that have iPhones expecting that a web page will work the same way on their desktops.

So kudos to the W3C for making it viable for other browsers to come to market and compete, and kudos to Mozilla for making that dream a reality. Now it's all about innovation and compatibility.

just being realistic

[boxwood]

The lack of version numbers is just being realistic. No browser is 100% compliant even with HTML 4.01, which has been around for how long now? And when is HTML 4.02 coming out? Seems to me they've abandoned the versions a long time ago. Everyone just uses HTML 4.01.

They can make a HTML 5.00 standard, and have most of the browsers implement 99% of it and then they release 5.01 and the browser makers will get to work implementing that, but totally abandon implementing that last 1% of the HTML 5.00 spec... because they would be too busy implementing 5.01, 5.02, etc. So a Web developer sets a HTML 5.00 doctype, uses a feature that isn't implemented yet hoping that someday browsers may support it. But there is no guarantee they will. So the web developer will just change the doctype to 5.01, 5.02 (or whatever the latest version of the spec is) every time he makes changes to a web page or CMS.

So they're just being realistic. No matter what standard they come up with, it will never be implemented fully by all browsers. Their standard won't be the law, it will be more of a guideline. Having version numbers is pretty pointless when all browsers aren't going to render a HTML 5.01 document exactly the same. Its easier for the web developer to tell the browser that this is a HTML 5 doc and the browser will use its latest code to render the page.

Re:just being realistic

[boxwood]

ok so you target HTML 5.00 which isn't implemented perfectly. You test it and notice a few problems on a couple of browsers. You do a work around for those browsers.

Then a couple of years from now a developer for one of the glitchy browsers notices a bug in how it renders HTML 5.00. What should he do? if he fixes the bug then your workaround is going make the page be rendered in a way that you didn't intend. A couple of more years later, that browser gets a completely new rewrite of its rendering engine. Now they have to make sure that it will still render HTML 5.00 the same way (the incorrect way) that the old rendering engine did? What if they didn't know that there was a bug that caused HTML 5.00 to be rendered improperly? Thats right, your page isn't going to work right on that browser. You have to go back and make some adjustments.

Thats the thing, no matter what you do, as a web developer you are going to have to make little tweaks to your web pages to make sure they work right on newer versions of all browsers. If the popular browsers do things right you won't have to do this very much. If there is a popular browser that makes big changes to how it renders pages between versions (*cough* IE), then you might have to make a lot of changes with each version. It sucks, but have version numbers on the spec isn't going to change that.

From what I can see the biggest difference between how pages are rendered now is dependent on the browser version, not the HTML version.

And if they do a good job on the spec, they won't have to remove/change how elements get rendered as time goes on. They'll just add new features and keep the existing featureset static. So HTML 5.01 would end up being the same as HTML 5.00, only with a couple of new features added.

Re:CSS is horrible for table layouts

[Stooshie]

If you actually want a table of data, then use the table tag. It's not deprecated, just assigned to be used for tables. If you are doing layout use divs or spans with floats and clears.

Re:HTML compliance is for wankers

[Stooshie]

Remind me to never hire you for a project. You sound like you are a nightmare to work with. I suspect you have never worked on a real site that needs to be used by a wide range of people across a wide range of circumstances. Blind people, colour blind people and people with upper body problems have to be able to pay online for their council tax, apply for planning permission etc.... Standards are vitally important for that.

Canvas patent

[tepples]

Apple created canvas, submitted to W3C

And a lot of people complain about Apple's patent on the <canvas> element. Apple isn't required to license this patent until <canvas> becomes part of a W3C Recommendation.

Re:HTML compliance is for wankers

[unapersson]

That's funny, I've encountered lots of people using Microsoft Word as their HTML editor who say just the same thing. So you're in good company.

Validators make a good sanity check/diagnostic tool when something isn't working correctly in Foobar browser, but they're not a crutch. Once you've got a solid working knowledge of HTML they're not really going to teach you much but might find a few typos.

Once you move beyond HTML and into CSS, valid HTML can certainly make a difference, but if you're sticking with HTML3.2 and not bothering with CSS you can get away with a lot.

Re:Reference implementation

[Rockoon]

WebKit is not a reference implementation. Its certainly an implementation, but it is neither complete nor authoritative.

Re:The Standards Really Never Have Been the Standa

[CynicTheHedgehog]

Not really. A "rolling" standard allows the periodic adoption and standardization of a single tag or attribute, which would allow progress on that front while the vendors continue to bicker about the other proposed changes. It's like of like the JCP process was (sans incorporation into a major version release). Different ideas are proposed: some get formalized and adopted and some languish. Nothing wrong with that. The result is that you get a browser or update with support for (for example) WSR-1234 which might be embedded video or canvasing or something, and after a (short) while they all will pretty much have it. And because it's incremental, it's easier to implement and release updates quickly, rather than rewriting the entire engine every two years.

Re:Have you got an odt file like that?

[pavon]

I have been on the sending end of that problem. KOffice and OpenOffice generate ODT files that differ wildly in their interpretation of parts of the standard that aren't specified. For spreadsheet files in particular, there are more things that are unspecified than are specified. Trying to collaborate using KOffice and OpenOffice is worse than using different versions of Excel.