Rebuilding The Web

Articles, advocacy, discussion and debate about the many problems of the Web and the challenges of rebuilding it.

Will HTML5 make the Web even more invalid?

Currently, each Web browser silently auto-corrects invalid HTML in a different way. HTML5 will harmonize the auto-correction of invalid HTML, so that all browsers will fix errors in the same way. Uniform auto-correction of invalid HTML behaviour across all browsers can only lead to good things - right? Or will this create even more invalid HTML on the Web?

Hey, my Web pages work in all browsers!

For most Web site creators, the only feedback that there is something wrong with their Web pages is the different rendering in different browsers. With HTML5, all browsers will fix errors in the same way, which means invalid HTML will render the same way in all browsers and fewer people than ever will know that they are authoring invalid HTML.

Validators?... We don't need no stinkin' validators!

With uniform auto-correction behaviour across all browsers, many Web site creators who currently use validators will rely on them less. Why? - because the two main reasons for using validators become irrelevant with consistent auto-error correction in all browsers:

Only valid HTML behaves consistently across browsers
This reason for using a validator becomes irrelevant because with HTML5, all browsers will render invalid Web pages the same way.
Only valid HTML future-proofs your work
Future browsers are likely to be backwards compatible by supporting the HTML5 auto-correction behaviour, making this reason for validating irrelevant.

With HTML5, all browsers will fix errors in the same way, which means there are fewer reasons to validate HTML, and fewer people than ever will know that they are authoring invalid HTML.

Wow, publishing on the Web has got a whole lot easier with HTML5!

Since all five major browser vendors signed off on the HTML5 auto-correction behaviour, it must be pretty good right? Better auto-correction of invalid HTML means less skills and effort required to create something that Web browsers will recognize/render as a Web page. Invalid HTML that did not render correctly before might render adequately now, and fewer people than ever will know that they are authoring invalid HTML.


Uniform behaviour across all browsers is a good thing only if the behaviour itself is good. However, silent auto-correction of invalid HTML is a bad behaviour and if all browsers implement uniform silent auto-correction of HTML per the HTML5 spec, this will lead to more invalid HTML on the Web.

Public comments

1. Posted by Rimantas
on Tuesday 2009-12-22 at 03:36:10 PST

Ok, let's see:

first, there always will be many many more html producers which don't care how html looks like, and even less so how
does it look in different browsers — that's why it is important to precisely define how browsers deal with invalid markup.
The most important are the end users, after all.

second — people who _do_ care about validity of their markup _never_ do judge it by how it is rendered—they use validators for that, not to mention that many many things that can make markup invalid will not affect the rendering in any way. So your assumption that those who do use validators now will stop doing that is plainly wrong.

third—there is a growing trend of using progressive enhancement, which means that even valid markup can and _will_ look differently in different browsers, based on their abilities.

fourth—silent autocorrection of HTML errors was there from the first browser to show up and will stay there till the last one. Making it consistent can only be a good thing.

Does anyone really believe, that lack of unified error correction somehow did affect the quality of web markup for the better?

2. Posted by Yannick
on Tuesday 2009-12-22 at 06:11:02 PST

Validiators... -> Validators...

3. Posted by Justin Sinclair
on Tuesday 2009-12-22 at 12:13:04 PST

I think consistent auto-correction is positive. There are many scenarios where you the original author have no control over markup quality. Content contributed by the client after you've left, comments on a blog if it uses a rich-text-editor. In all these scenarios it will be great to know that any issues will be consistently handled. Will it stop me caring about well structured, semantic and valid HTML? No way.

4. Posted by Jonas
on Wednesday 2009-12-23 at 05:05:51 PST

The assumption is interesting...wrong direction to think about it, but interesting. There will always be messy code but blame the developers, not the language. ;)

5. Posted by mattur
on Wednesday 2009-12-23 at 05:50:56 PST

Great Historical Questions to Which the Answer is No, an ongoing series brought to you by Vlad Alexander... :-)

6. Posted by Richard
on Thursday 2009-12-31 at 09:35:31 PST

Consistent auto-correction is a good thing. As a result, fixing problems will be easier.

7. Posted by Stuart
on Monday 2010-01-11 at 15:24:37 PST

I agree that harmonizing the correction of invalid HTML will result in more "invalid" HTML, but I don't see that as necessarily a bad thing. It sounds to me (without the benefit of knowing what corrections will be done) as if in effect HTML5 will just be specifying alternative ways to write valid HTML. It just that some ways will be "valid" and others will be "auto-corrected".

Now that I think about it, wouldn't it be simplier to change the spec so that the auto-corrected HTML is valid? Why specify something as invalid and then specify a way to convert it into something that is valid?

Comments are closed for this article.

Main menu