Anne van Kesteren

The inaccessible Google Accessible Search

Now I guess Google Accessible Search probably works more or less in most screen readers and such (I assume they tested that), but the markup is just crap. Perhaps it’s slightly better than google.com given that it has a DOCTYPE. Then again, the DOCTYPE is wrong. Furthermore, they don’t declare the language of the page, use a bgtcolor attribute on the body element, interchange <br /> and <br>, use font and b instead of hx elements and I haven’t even looked at the result page yet.

Looking at that now I’ll guess I just leave it at this. Besides, Karl Dubost is already hunting them down. Not that I’ve high hopes they’ll respond…

Comments

  1. Yeah, it always surprised me that for such a simple layout, with just Google in general, there markup is haywire all over. Google Accessible doesn't look very different to me at all. Does it even provide any benefits?

    Posted by Ethan Poole at

  2. Well, the idea behind is that you can find accessible pages. That’s also the irony here, of course.

    Posted by Anne van Kesteren at

  3. Most accessible search at this moment is search.msn.com. Home page validates as XHTML strict. Results page has only one issue with use of custom attribute gping.

    Also they are using pretty semantic markup.

    Good job there by Microsoft.

    Posted by FataL at

  4. I've come to the conclusion that google use shit markup on purpose purely to piss off standardistas.

    Posted by Robin at

  5. The rumour goes like this:

    1. With the amount of traffic they have, every byte shaved off their pages translates to quantifiable dollars.
    2. They have the resources to test their markup in dozens of browsers and maybe hundreds of different versions thereof, and they do. When you can actually test all implementations, standards compliance is meaningless.

    But that may all be just lore. I have no sources or citations for any of this.

    Posted by Aristotle Pagaltzis at

  6. Working for a company that produces a browser I can tell you that two is certainly not true. And one, looking at the code, isn’t either.

    Posted by Anne van Kesteren at

  7. What's not accessible about it? It's XHTML and uses tags included in the Transitional spec, within bounds of interpretation by browsers that would be used in the context of the product (that is, browsers by people with disabilities). Single quotes and font tags aren't of any meaningful disadvantage in this case.

    Beyond that, they followed through in the SERPs with usability standards like bright blue links, bright red visited links, less results on a page, less space between results, less description text, less paraphernalia associated with a result, etc. It's a massive improvement, accessibility-wise, over the standard Google, which is already getting quite crowded for such a minimalist engine.

    Posted by Rahul at

  8. For years, Web Standards and Accessibility are just search results for Google. But, there always is hope. (Above all since Ian signed there...)

    Posted by Jens Meiert at

  9. What's not accessible about it? It's XHTML…

    You must be new here.

    Posted by Anne van Kesteren at

  10. What's not accessible about it? It's XHTML

    Yep, all you need is a doctype to make your site accessible. I put 2 doctypes on some of my pages to make them ultra accessible.

    This is how the majority of developers think.

    Posted by Jake Archibald at

  11. Hmm, I think there's some confusion here between standards and accessibility. A standards compliant site is by no means necessarily accessible. Likewise, a site can be accessible without complying to standards.

    Thus as long as Google Accessible Search works in all browsers, screen readers, pda's and what have you it necessarily IS accessible. Of course I can't tell whether it does work in all user agents, but it works in the ones I use and the results it brings back are definitely more accessible sites than in a standard search.

    It would be nice if the results pages were standards compliant as well. It would seem that with such a basic page it can't be very difficult. However this would not per se make it anymore accessible.

    Posted by Robin at

  12. I find it very sad and a good bit of sameful that Google produces such incredible bad code. If they can't even produce standard-compliant HTML, what guarantee do we have for all the other code they are churning out (like Google Toolbar, GMail etc)?

    What reason do we have to believe that their only lack in competence is on the HTML they produce, but that every other aspect of programming, including security, cryptology and whatnot, is of the highest possible quality?

    What's even more shameful is that they continue to produce some of the worst HTML available (at this level and field at least) even after Ian Hickson joined their forces. I understand that improving Google's websites code-wise isn't his job, but not utilizing his expertice on the area is just foolish imho.

    Posted by Asbjørn Ulsberg at

  13. This made me laugh out loud! Not only do the pages look almost identical to the original Google search page, but all of the “accessibility” could have easily been achieved by giving the user the choice of an alternate stylesheet…

    Posted by Martin Payne at

  14. It's a laugh but my guess is that they aren't too serious about it themselves either. The copyright notice at the bottom (when viewing the results) says "Copyright 2000".

    Looks like they took an old car, polished it and sold it as new again...

    Posted by Ulco at

  15. What's not accessible about it? It's XHTML and uses tags included in the Transitional spec, within bounds of interpretation by browsers that would be used in the context of the product (that is, browsers by people with disabilities).

    Accessibility is not about building web sites for people with disabilities.

    It's about creating web sites for all kind of users, children, old people, the disabled, the blind and visually impaired, the color-blind, the inexperienced, the impatient, the pda and mobile phone users, the search robots - creating web sites for all kind of browsers and platforms and with all data and information accessible.

    Posted by Klaus Hartl at

  16. When I use Lynx, I get a HTTP 400 Bad Request-response when I hit the search button.

    The bad code of Google's web sites is something I still don't get. Google started as a web search service, so they should know the benefits of semantic code on web pages.

    Posted by Niels Sijm at

  17. Google started as a web search service, so they should know the benefits of semantic code on web pages.

    Umm. A big part of Google’s core competence is making sense of full text without the page authors jumping through semantic hoops and spoon-feeding Google everything.

    Posted by Henri Sivonen at

  18. I Still love the irony how it places "Flash only" pages near the top of the listing for several search results.

    Posted by Robert Wellock at

  19. I think the reason there is fustration at Google is because they don't justify their use of the languages used to produce their applications. Why should they? They answer to searchers not developers. The best way to get the message across is to search elsewhere, as will anyone that finds an issue with accessing the facility.

    As for the lore stated by Aristotle, it has to be the case, why else would they write the way they do? The single biggest infuence in terms of output is the cost of bandwidth.

    Robin made a good point. Standards doesn't equal accessibility. As far as I know, taking the example of visual impairment, validation means nothing but clarity of information does. Linearity and contrast go a long way. Google is the search engine of choice in this respect.

    Posted by Edward Clarke at