What Would Louis Do?
Last Thursday Google revealed a new search product called Google Accessible Search. This is an encouraging development, but it needs work. I have praised Google in the past for contributing to the web standards movement and promoting accessibility, mainly through the positive effects of its search algorithms on the rankings of standards-compliant pages, but it seems that the organisation is more ambivalent on these questions than I had realised (clever marketing graphics notwithstanding).
Ben Buchanan’s recent post, Google Accessible Search Isn’t, presents some reasonable criticisms. Why should the world’s premiere web search brand, develop a secondary search page to promote to users with disabilities when they are in such a strong position to champion the universal appeal of top-to-bottom accessibility? Why use tables for layout? Why use the
<font> tag? (There’s something a bit John Bender-ish about coming to Accessibility class and then littering your source with
<font> tags when nobody is looking.)
So what does Google’s new tool do? It filters query results to privilege accessible web sites. This provides visually-impaired users with hits that are easier to navigate than those provided by the regular Google search. I think that this is a great idea, but I agree with Ben that this technology should be targeted for inclusion in Google’s front page, rather than being one of it’s many side-projects. Accessibility is an issue of universality, not particularisms.
I suppose that extreme minimalism could account for the fact that Google’s own pages do not provide a DOCTYPE declaration, but this behaviour flies in the face of the W3C’s accessibility recommendations, of which Google themselves are official proponents, at least when they are advising others on how to improve their search rankings. But the Accessible Search results page that I just checked uses no less than 20 tables to display non-tabular information. This practice may slip past unnoticed by anyone using a common browser on a personal computer, but is clearly sub-optimal for non-standard user agents, which category includes not only screen-readers or braille displays but also devices such as mobile phones and PDAs. In an earlier post Buchanan summarised the situation perfectly:
Users with devices other than desktop PCs have the same requirements as disabled users. That means some of the most technology-savvy, youngest and brightest users are actually the ones at risk of hitting problems. The user base also includes key decision-makers like managers, CEOs and so on who like to have the latest and greatest toys.
That’s a substantial and growing base of users that don’t all fall into the category of “Visually Impaired” who could benefit hugely from the functions of the new accessible search, as long as their devices can wade through the tag soup surrounding the search results themselves. It’s still early days for this product and the fact that Google is soliciting feedback is encouraging. But the choice to recognise accessibility as a question of universal access, rather than political correctness, is a obviously a tough sell, even for Google. It’s a message that bears repeating: making accessibility a central goal of core services, and following the technical standards as closely as possible, will advance everyone’s interests in the present and future of the web.
Let me start by congratulating you on another excellent article. As usual, I agree with your sentiments, however, again as usual, I’m going to attempt ask you to shift your perpective a little.
Google Accessible Search is a long time coming and I believe that it is the first offering from a mainstream search provider which attempts to provide a quality service to those with with accesibility requirements. I’d like to point out that this is a Google labs project, that you can provide feedback and how I see this product fitting in with whole suite of Google offerings.
Google Labs was established to test and play with some interesting and perhaps non-mainstream ideas. It allows users to see what is happening at the leading edge of search technology and I think that it is important to remember that Lab offerings fall into one of two categories: not-ready for mainstream use or never going to be in mainstream use. From labs.google.com:
Google Sets is a great example, it has been a member of Google labs since 2002 and while it seems quite mature and reliable, its utility is probably questionable for the majority of users.
The whole point of Google Labs is to provide the general public with access to play with new technologies, and for Google to get feedback in return. As far as I can recall, only Labs products provide anykind of feedback mechanism. Being something of an optomist, or perhaps I’m just naive, I feel that this feature is meant to be used. I’ll be writing a note to them once I finish this comment, and I hope that yourself and other critics from the blogosphere join me.
While agree that using non-standards compliant html is not cool, especially when creating an application for accessibility, I feel the argument that this will alienate all users of non-standard user agents is invalid. I don’t want to discount that it is an interesting argument, just that it isn’t applicable to this argument.
Something Google does well is to segrigate their products so that they attempt to address different segments with different start pages. This is one of Google’s strengths and one of the reasons that they got ahead in the original search engine wars, where their competitors generally tried to fit as many features into one page as possible. To demonstrate my point, may I refer to the following:
The list could go on, but I think that makes the point. Google attempts to address each niche seperately. What works in one niche may or may not work in another. Imagine having your image search and content search integrated? Try using google.com effectively on a cell phone. It just doesn’t work as well as it should, thus the niches.
Using the argument that because Accesible Search renders results in tables, first adopters and other users of new technology will be alienated is false and perhaps foolish. I wish I have the knowledge to give this falacy a name, but alas. This scenario is like complaining about a lack of performance in your sports car when filled with standard gas when the car was designed to use high-octane. The car will work, just not as well as it should.
I can ask this question, why oh why would a first adoptor use an accessible search rather than one tailored for pda and cell phones? I’ll hazard a guess that the results from pda.google.com probably render in a similar manner, and therefore a similar argument can be made against the useability of that service. This can’t be confused as being the same argument as the one against Accessible search.
So, my tirade is over for now. Yes, its bad that Google don’t use standards compliant code themselves, yes it is especially bad (evil?) when offering an accessible search. But, it is a trial, provide feedback – they want it.
July 26th, 2006 at 2:24 pm #
I wish mes could right better english. Far to many errors in the comment above, I’m sorry…
July 26th, 2006 at 2:28 pm #
Thanks for your comments Dan. Yes, you’re right. I will send them my feedback and as I said, I am encouraged that they have a mechanism for it.
On your point about my tables criticism, what I said was:
Stripping it down to basics:
1. You don’t have to have a disability to run into accessibility problems, just a non-standard user agent;
2. The functionality of Google Accessible Search should be good for such users since it helps them avoid sites with deprecated code;
3. Deprecated code will not always stop you dead in your tracks, but it may slow you down to a crawl.
I take the last point to be implicit in my statement, but there nevertheless since I said “as long as their devices can wade through the tag soup.” So I don’t see anything fallacious in that. It’s actually in agreement with your objection!
One final thing. Google can do it right: http://www.google.com/pda is valid HTML 4.01 Transitional, and the markup uses semantic principles. Google Accessible Search does not validate as HTML anything (since it has no DOCTYPE), contains errors that would prevent it from validating if it did have a DOCTYPE, and mostly ignores semantic principles.
Validation is not the be-all and end-all of accessibility, but it’s a pretty good baseline. And in this case it supports the argument that using Accessible Search with a PDA might be a pain in the arse. In truth, Accessible Search and PDA Search should accomplish the same thing, because if they did the users would all be better off.
July 26th, 2006 at 4:38 pm #
I’ve read your post again, just to make sure that my comment was within scope. A better answer would have been: That’s disgusting, you’d think that Google would know better. Have you left feedback? It seems that Google consider only people with some, but diminished vision as potential users of an accessible search engine.
Let us agree that all searchs should render results that are consistent and standards-compliant. However, to me it is not logical to get the same search results from both accessible and pda and I therefore have an issue supporting anything which makes them the same.
The reason that I take this stand is that accessibility on the web has nothing, if anything to do with the use of browsers that aren’t IE on a computer that may or may not be Windows based. For a user of a minority browser, a poorly rendered webpage is merely an annoyance. This kind of user probably has some choices available to them – they can wait until they get home and use their pc or go to a library or net cafe.
On the other hand, the kind of people who just plain can’t make use of mainstream technology are the people for whom accessibility is all about. For them, a poorly rendered page excludes them from society and discriminates against them. They have no choice.
The fact that ‘accessible’ websites are the ones which will work on my Mac, PDA or RiscPC is by the by. ‘Accessibility’ has become a catch all for all minority users, not just those who need accessibility. I can, and have, made web apps which work on all browsers and all most platforms, but will fail completely for a special needs user agent.
So, saying that some kid sitting on the bus playing with their myspace page has needs that are the same as person who has a genuine need for accessible sites BOILS MY FUCKING BLOOD. They aren’t the same.
Ok, I’ll give you that their needs have some overlaps. Good code benefits all. In fact very similar methods would yield positive results for both groups. But we are morally obliged to help one group.
Now I’ve got that out of the way, let me add that the needs and the optimal search results of someone using a pda for convienence should be very different from the results of a user who relies upon an user agent with an accesibility focus. And to Google’s credit, the two different niche searchs render different results for the same search terms.
I think that the need for differences is fairly obvious, so I will not insult you by rabbiting on about them here. But I will attempt to find out what the differences between the three search types are.
For a search for “wireless networking”, accessible returns http://www.apple.com/airportexpress/ where both google and pda returns http://computer.howstuffworks.com/wireless-network.htm. Now, in an ideal world, one would have to assume that link to Apple would be somehow of more relevance to someone with limited accessibility. So I used Silktide.com to measure the two sites. The results for the accessibility tests are 5.6 and 7.8 respectively. The exact opposite to what should have been. I’d love to know the reasons behind the differences.
Ok, to wind up. I agree that google should really know better. I disagree that pda users and disabled users should be treated the same when the latter are already so poorly treated on the web. There should be and is a difference in results for the two searchs, but the science that Google is using is different to that of Silktide.
Your homework, if you want it, is to determine the effect of web 2.0 methodologies such as DHTML and AJAX on special needs user agents. How do you verbally represent a DHTML rollover menu?
July 26th, 2006 at 11:25 pm #
I can’t tell if we are arguing at cross purposes here, or whether we each take slightly different messages away from the Accessibility issue. In several posts now I’ve stressed the point that Accessibility is a question of universal access, and I mean to stand by that assertion. The most important concrete effect of universal access should be that the most limited user-agents can access most of the available content on most sites with relative ease, especially where that content is primarily text (i.e. minimal discrimination based on user-agent). Deprecated code won’t always prevent such access but it can slow it down — I think we agree on that. As for search results: that Google PDA and Google Accessible Search are different makes sense given that the stated objective of the latter is to filter out sites that would be troublesome for people with vision impairments. What I was saying about other users, such as people with Blackberries, is that if the accessible search effectively returns only the most accessible pages from a regular result set, then on the whole the service would probably be beneficial to those users too (because hits would contain fewer graphics, have simpler page formats, utilise good keyboard navigation, separate style from content, etc.).
July 27th, 2006 at 12:43 pm #