Random Acts Of Validation
February 17, 2004 | 
49 Comments
Over the last few weeks, when I’ve had a chance, I’ve been conducting a less-than-scientific, potentially useless, but rather interesting, study.
After reading the rather heated discussion that popped up over at Andy Budd’s site on the use of validation buttons I decided to put some well known sites to a simple test.
I tested 30 sites at random, as I came across them, to see how well they stacked up against the various validators. I only chose sites that were either run by an advocate of either CSS, Accessibility or Web standards, we’re referred by one of those advocates or proclaimed in some other way to be valid.
I wanted to make sure that the sites I validated were from a pool of sites that I figured would be striving for validation as a goal.
The sites I chose run the gambit. I’ve checked blogs, portfolios, online magazines, corporate sites and everything in between. Some were personal and run by one person, some had multiple authors.
I tested each site’s homepage and 10 random pages throughout the site for (X)HTML validation and, if applicable, the site’s main stylesheet for CSS validation. On a few sites I also checked Section 508 compliance.
This study is hardly complete or scientific, but the results were interesting, and seemed to show some surprising patterns. Here are some of the more interesting results:
- 16 of the 30 sites had a valid homepage. This was much lower than I’d expected.
- The average number of pages per site that came up without error was Five out of 10. Not too good.
- I only found 3 sites out of 30 that validated on everything I tested. And two of those were Zeldman, who wrote the book on Web standards, and The Web Standards Project.
- Two more of the 30 would have been 100% valid, but for comments. Some of this might have been luck of the draw.
- Blogs seemed to have more valid pages in general, especially if you take out the comments.
- Of the 23 sites who’s CSS I checked, only 14 were valid.
- The most common errors were un-encoded ampersands and nesting errors.
- Comments, forms and links were the most common place for errors.
All in all it seemed that validation was strived-for on most sites, but not fully achieved. I was a bit surprised at the level of valid support, just because the sites I tested were all sites that I expected to be valid or nearly so. The average was right around 50% valid by page. That seems a bit low.
What does this tell us? Not a whole lot, but it confirms in my mind that:
- It’s very hard to keep a Web site valid.
- At the end of the day, even Web standards and CSS advocates can live with some validation errors.
- Striving for a valid site is worthy of praise.
- Validation isn’t the end-all be-all of Web standard support.
- The validators aren’t very useful as an indicator of quality. Every single site I checked is highly regarded.
- If your site is 100% valid or really close, you deserve some serious props.
- If your site doesn’t validate 100% you’re in very, very good company.
Just FYI, I tested my own site as well. My homepage was valid, as was my CSS. Eight out of my ten pages were good, the two that weren’t being made invalid by comments, both nesting errors. I put quite a bit of time into eliminating errors, so I was pretty pleased. After conducting this study I’ll be more comfortable with the errors that creep in that’s for sure.
Which makes me wonder if there is much use trying to keep a valid site. The validators are very useful in nailing down CSS bugs and display errors, but aside from that I’m not sure they provide much in the way of practical information.
I think it’s less important to please the validator than it is to please as many browsers as possible. I mean, hell, even if you have valid (X)HTML and valid CSS, your site could still have rendering problems in just about all of the major browsers.
I know this site has a few. The validators have been no help with those. ;)
Filed under: Web General
Comments
1. Dave S. said:
I’d be curious to hear if you included my site in this, and how well it fared if so.
I’ll say this now instead of later: I’ve gone to ridiculous lengths for the sake of validity. We’re talking server-side plugins and ugly RegExs a-plenty, all to make the validator smile.
Posted on February 17, 2004 11:29 PM | #
2. Roger said:
After reading this I went to check my own site. I always make sure it’s valid when I make any changes, so I didn’t think the validators would find any problems.
The first few pages I checked were valid, and I started thinking I’d pass the test, but then I found a couple of errors caused by unencoded characters in comments. So now I have to find a way to make MT take care of that.
So, yes, it is hard to keep a site valid. But it’s not impossible.
Posted on February 17, 2004 11:51 PM | #
3. Chris Vincent said:
It’s easy to create a valid site. It’s difficult to keep it valid. That much has been said before, and this little survey has supported the statement. Props for going through all that for the sake of our interest. :)
Most validation errors are simply overlooks. The errors my site has on a few pages involve something simple that I mean to fix when I get the gumption to, but it really doesn’t bother me that much for now. Validation only really becomes an issue on sites that are invalid due to errors inherit in their coding, with obviously no effort put towards validation.
I liken validity to programming syntax. A program written in C simply won’t compile without 100% valid syntax. On the web, there is a syntax, but you don’t necessarily need to follow it to get the desired results. However, the closer you keep things to the spec, the better. After all, as we know from some recent hubbub over XML error handling, you never know how one user agent will recover over another.
Posted on February 18, 2004 12:35 AM | #
4. Pete said:
Chris is absolutely correct; I don’t know how many times I’ve forgotten about ampersands in URLs or heard horror stories from friends and collegues about handing over completed totally vaild sites only to see them slowly become uncompliant by their clients.
It’s not a surprise that the majority errored due to un-encoded ampersands and nesting errors, blogging tools generally don’t enforce W3C compliance or have automatic tag closing or encoding ampersands in urls.
Posted on February 18, 2004 01:48 AM | #
5. Mike P said:
“Validation isn’t the end-all be-all of Web standard support.”
In my own experience I’ve found that while validation errors can be corrected, markup *can be* a bit more difficult to change - therefore from the ground up one needs to focus not on validation but on acheiving a level of “proper semantic markup”.
I don’t know whose blog it was on, but someone said “strive for proper semantic markup and validation will take care of itself”.
Knowing that little validation errors can sneak in, This quote seems on the money (imo!).
Posted on February 18, 2004 02:55 AM | #
6. Dave Meehan said:
I know there has been the debate before about what if user agents actually rejected non-conforming pages. We all realise that the web would shrink enormously if this was to be implemented. What might be more interesting though is if user agents rejected pages that claim to be valid, ie. through links to validators.
Personally I don’t see the point of web standards if someone, somewhere doesn’t actually start to make conformance mean something to the average joe in the street.
Posted on February 18, 2004 03:10 AM | #
7. Ste Grainer said:
Heh, I can definitely say that my site attempts to be valid, but every once in awhile the little things creep through. (I just checked and found that I forgot to encode an ampersand in a URL.)
The other day, I embedded a quicktime movie in an entry and that forced me to invalidate that page; I wasn’t particularly upset because validation isn’t an end, it’s just a means to an end. The end is a website that strives to be compliant to standards and forward-thinking, and I think most would argue that even getting halfway there is a significant achievement. When you throw in the important aspects of usability and accessibility, any website that can be valid, fully accessible, and easy to use is like a needle in a haystack. Ultimately, that should be the goal of any dedicated web designer, though.
Posted on February 18, 2004 03:22 AM | #
8. Richard Allsebrook said:
Once you make the switch to XHTML strict (as I did), it becomes CRITICAL that your pages validate (or at least are well formed). Mozilla for instance wont even attempt to display a badly formed page. Opera does have a go then spoils it by inserting an ugly error message at the point it all went wrong :-/
I have spent quite a lot of time an effort on my site ensuring that all the pages validate. Even the comments. I run each comment post through a XHTML validator to ensure they are well formed, then I enforce the correct encoding. Since the updates to my commenting system, I’ve not experianced any invalid pages. I have an article in progress that explains the steps neccessary to ensure user comments don’t break your site. I’ll post a link once its done.
Mistakes do creep in from time to time though - especially on dynamically generated sites like mine. That is why my site DOESN’T have the valid sticker on it, even though (to the best of my knowlegde) the site is 100% valid.
If your site proudly boasts its validity and then quite clearly isn’t - it looks bad on you - not the person who posted the offending comment - and it makes a mockery of the w3c’s advocacy of standards too.
Also, it would be nice if the browser could inform the server that it experiance a problem trying to render the page. You could tie it into a automatic webmaster notification script.
If you don’t think you can enforce valid comments, either don’t allow comments - or take off the sticker.
Coo… I get a loverly view from on top of this high horse but I think I’ll get down now before I get a nose bleed :-D
Posted on February 18, 2004 06:38 AM | #
9. Milan Negovan said:
I checked my site stats and it appears that people do hit my validation buttons to see how it plays out since the W3C validators show up as bots. :)
Like Dave and Richard here I run every page through a filter which cleans up the outgoing HTML stream and tweaks it for compliancy. Some platforms (ASP.NET in my case) take a lot of control of how everything is rendered. I guess there’s no right or wrong approach of enforcing standards compliancy then.
Posted on February 18, 2004 07:40 AM | #
10. Suzanne said:
I find it’s not hard to keep a valid site once you’ve started. Using Mozilla and a strict doctype generally keeps me on the straight and narrow. However, my older pages on my blog, well, as I said when I switched over, I’m not going back through time to fix those on purpose. I do, however, go back and fix them as errors come up, or as people report them.
What trips me up repeatedly is not encoding urls properly. If I could remember that when I enter things (because I do things predominantly by hand, so I can’t rely on software to help me), I’d be golden.
Of course, I have no idea if my comment pages are valid as they are separate for now from the rest of the site.
Posted on February 18, 2004 08:14 AM | #
11. Matt said:
In response to Dave:
Is it really worth it? Going to those lengths to make sure a site is valid makes it seem more like a competition to beat the validator, rather than to actually serve well-formed documents.
I had the same feeling when the uproar about the bug in the CSS validator popped up a few weeks back. Yes, the bug should be fixed, but did it really matter that suddenly CSS didn’t completely validate? The CSS files were still technically valid.
I guess that the validator, to me, should be used as more of a guideline than a bible. If your pages are well written, then what is the problem? Is it necessary to implement numerous hacks, just to get around some of the idiosyncrasies of the validator?
Posted on February 18, 2004 09:11 AM | #
12. Suzanne said:
In response to Matt: My CSS doesn’t validate because I’ve used -moz bits in it. But that’s supposed to be valid, so I say, well, pfui on you, validator! :)
I think the goal (for me) is to have well-formed documents, marked up sensibly. That way I don’t have to touch them all the time. I really dislike working with the same documents endlessly.
Posted on February 18, 2004 10:03 AM | #
13. Arve said:
Dagfinn Parnas did a rather voluminous examination of page validity in his master thesis How to cope with invalid HTML.
Even if this thesis has been available for quite some time, I believe it still has validity. He sampled a total of 2 500 000 documents from Dmoz. The results were quite apalling: 99.3% of the documents failed validation.
Posted on February 18, 2004 10:51 AM | #
14. Dave S. said:
In response to Matt (#11):
“Is it really worth it? Going to those lengths to make sure a site is valid makes it seem more like a competition to beat the validator, rather than to actually serve well-formed documents.”
Well in my case, I spent all the time I *would* have spent double-checking my hand-coded XHTML, and instead used it to build a dynamic system so I never had to worry about it again. Unless I make a dumb error in my post content, all Movable Type-generated pages on my site are valid, guaranteed. It’s a nice feeling to just *know* your code is valid, rather than wait around for the inevitable storm of emails, or in the case of serving XHTML with the app/xhtml+xml, wait around for your site to completely break due to well-formedness errors.
“If your pages are well written, then what is the problem?”
I think Keith’s point is that a lot of pages he checked weren’t well-written. Whether that’s an artifact of a CMS or a result of hand-coding, the code comes with nesting and encoding errors.
All this is pretty meta though. One day, hopefully sooner rather than later, the systems we use will just generate valid code and we won’t have to worry about this. I can’t see any valid argument being made that a human should fix an unencoded ampersand, and not software.
Posted on February 18, 2004 11:42 AM | #
15. Keith said:
I’ve got to agree with Dave. It’s just not practical for a person to try and keep up with all the errors that creep into many sites.
Take this post for example. It was valid yesterday, now it’s not.
It’s very ironic, and somewhat funny, that when Richard said:
He broke my page with his comment.
This isn’t entirely his fault, he coded in paragraph breaks which causes a nesting error and really I should have something in here explaining that while XHTML is ok in comments, line breaks are added. Maybe I should disalow paragraph tags?
It’s hard though because I’ve got a wide variety of users, with varying skill levels that like to comment and who do you cater to? I’d bet that regardless of what I did comments would eventually break the site as far as the validatior.
Is it worth my time to find an ultimate solution? Nope, don’t think so.
Posted on February 18, 2004 12:48 PM | #
16. Scott Johnson said:
“If your pages are well written, then what is the problem?”
Even when your pages are well-written, a stubborn CMS or unvalidated user comment postings can quickly break your valid work.
Posted on February 18, 2004 04:21 PM | #
17. Devon said:
I’m sorry to say, but anyone (who does coding regularly) who can’t simply nest tags correctly and properly encode their &’s and <’s as entities, is lame. Wait no, I’m not sorry for saying it. It’s too damn simple. On the other hand, I can understand having validation problems, they’re a bit tougher to keep at.
Posted on February 18, 2004 07:34 PM | #
18. Keith said:
Devin, you’re killing me. That was a joke right?
I had to re-encode the & and < that you left in that comment. I sort of hope you did that on purpose – even though that would have been pretty lame.
Maybe I should just disallow XHTML in comments. Would save me a bit of grief.
Posted on February 18, 2004 09:44 PM | #
19. TOOLman said:
Validating your markup (and CSS) is good, but as someone said, it’s a means to an end, not the end itself. Making a page validate when you allow comments can be very hard, especially if you allow XHTML. The blog author shouldn’t be held responsible for such errors.
However, displaying a “this is valid XHTML” icon on a page that doesn’t validate is hypocrisy. In that case, you should use another phrasing in your link, such as “validate XHTML”.
/Tommy
Posted on February 18, 2004 11:04 PM | #
20. Matt said:
Keith, that’s not necessarily a bad idea.
To Dave (#14):
I misinterpreted your original post. I was under the impression that the “riduculous lengths” you had gone to were not on-the-fly XHTML validators (which are a very good thing), but instead a series of hacks that made your code look good to the validator, even if it made the pages less well-formed than they otherwise might have been. I’m not quite sure why I got that impression, but I agree with what you’ve done 100 percent.
The fact that so many pages are NOT well-formed is certainly an issue, one that I didn’t mean to poo-poo in my last comment, if that was the impression that I mistakenly gave. I believe that the proliferation of these kinds of errors can mostly be blamed on WYSIWYG editors (think back to </p>that Frontpage ad</p> and laugh), and unfortunately I don’t see this changing anytime soon. HTML is too widely used by too many authors that don’t care what the code looks like that they are writing, as long as it looks good on whatever browser they happen to be using.
Posted on February 18, 2004 11:10 PM | #
21. Richard Allsebrook said:
Oops - sorry Kieth *blush* - I seem to be making a habit of breaking pages with comments - earlier this week I broke one on Ryan’s site: http://www.ryanbrill.com/
In my defence, I did preview my post before I posted it and it previewed fine.
Here’s one idea I have been toying with:
A basic and advanced comment option. Basic allows NO hand coded markup and squirts it out as a <pre> tag. The Advanced option would allow markup (which would be validated before being accepted). That way you would keep most level of user happy.
Please note - no markup was used in the construction of this comment ;-)
Posted on February 19, 2004 01:51 AM | #
22. Robert Wellock said:
Well, I chose application/xhtml+xml which says it all really if I don’t do well-formed XML, in Mozilla you’ld see “Yellow Screen of Death”.
Like Tommy said it’s good-practice but not a necessity though I have used eXtensibility on various pages thus it made sense when I created the DTD I use valid XHTML.
It’s also good if you broke one of Ryan’s pages he would thank you so he could improve his software blog system, well he did when I purposely broke it to show him some possible flaws.
Posted on February 19, 2004 04:14 AM | #
23. Tim said:
I’m new to the validation scene, and while I’ve used the W3C Validator, is there any way to check the validity of an entire site?
Posted on February 19, 2004 04:34 AM | #
24. Anne said:
See also: The XHTML 100 and X-Philes: Reloaded comment 3.
Posted on February 19, 2004 04:40 AM | #
25. Martin Lucas-Smith said:
In response to Dave Meehan (#6):
I think what would be extremely useful for many webdevs is a browser mode in which only valid pages were rendered. Such a mode would have to be explicitly switched on by the user, who would probably only be a webdev testing their site as they develop it.
Any Mozilla developers reading this and would care to consider such a feature?
Posted on February 19, 2004 05:04 AM | #
26. zeldman said:
Don’t let visitors post XHTML markup in your comments and you won’t have to worry about comments breaking your validation. (Not to mention that you also won’t have to worry about malicious scripts, images, etc.)
Posted on February 19, 2004 10:53 AM | #
27. Keith said:
Well, yeah, obviously. I am thinking (as I mentioned up there in the comments somewhere) about not letting people use markup in the comments.
The thing is I’m not sure I care all that much about the minor errors comments cause. But then again…
Maybe I’ll put it to the readers.
Posted on February 19, 2004 11:09 AM | #
28. Keith said:
Actually, now that I think on it, preventing markup in the comments wouldn’t nessicarily clear up the problem by itself. I’d also have to put something in place to clean up unencoded characters as people like to cut-and-paste into the comments.
I imagine MT has something built in for that though…
Posted on February 19, 2004 11:12 AM | #
29. pete said:
In my own home-grown CMS/weblogging tool, I’ve added hooks to check all new entries for well-formedness using tidy, and to check each feed for validity using a local install of the feedvalidator. This helps to ensure that I find any errors before you do, and I correct them.
Of course I’m one of those standards-weenies you hear about on the web… ;)
I think if the tool vendors made it easier to prevent bad markup from getting into the system, it could go a long way. So if Six Apart, Blogger, Live Journal, etc added an *option* to check and warn users about invalid markup, and even corrected, or offered to correct it, we might see a lot more valid pages. Note that I said the “option” as there are some people who will never care. We’re after the people who care a little, and might care a bit more if things were easier.
Posted on February 19, 2004 11:57 AM | #
30. pete said:
Tim asks: “…is there any way to check the validity of an entire site?”
See this url: http://www.htmlhelp.com/tools/validator/
Posted on February 19, 2004 12:08 PM | #
31. David said:
For those using Movable Type and find that it’s generally comments that break their valid markup, there is the Validable plugin which converts ampersands to &, correctly closes <img /> tags, etc.
[It will also work on your MTEntryBody text, but personally I think it’s better training to not rely on the plugin, and I just try to remember to encode things correctly myself, until it eventually becomes second nature.]
Posted on February 19, 2004 12:31 PM | #
32. Matt H. said:
The Amputator plugin for MT auto converts ampersands. I’ve found it quite handy.
Posted on February 19, 2004 02:24 PM | #
33. Keith said:
I installed the Validable plugin as David suggested and that cleaned up many of the errors to seeminly no ill effect.
I still have a problem with a few apostrophe’s creeping in from when someone cut and past some text, but aside from that it looks pretty good!
We’ll see how it goes.
Posted on February 19, 2004 02:32 PM | #
34. Johan Svensson said:
MT can properly encode ampersands with encode_xml=”1” in your MT tags. That’s how I keep my site validating since I would always forget to fix links manually.
The only part of my site that doesn’t validate is the individual blog entry pages – which also happens to be the largest part of the site. MTs default template has some icky javascript that breaks validation, and I haven’t gotten around to fix it yet.
Posted on February 20, 2004 04:05 AM | #
35. Pat said:
“I still have a problem with a few apostrophe’s creeping in from when someone cut and past some text…”
Here is a php function I wrote to take care of those apostrophes, accents, and other pesky entities. It will encode any illegal character into an html entity, and will leave all properly encoded characters alone.
I hope someone out there finds it useful. And if anyone can think of a way to improve it, by all means let me know.
function entityencode($string)
{
$chrs = get_html_translation_table(HTML_ENTITIES, ENT_NOQUOTES);
foreach ($chrs as $key => $value) {$chrs[$key] = ‘&#’.ord($key).’;’;}
unset ($chrs[“<”]);
unset ($chrs[“>”]);
unset ($chrs[“&”]);
unset ($chrs[“’”]);
unset ($chrs[‘”’]);
$output = strtr($string, $chrs);
return preg_replace(‘/&(?!(\w+|#\d{2,4});)/’, ‘&’, $output);
}
(just incase the form screws up the code, you can also get it from here)
Posted on February 20, 2004 04:43 AM | #
36. Trisignia said:
I’ve recently decided to keep all of my sites valid, at least as far as the XHTML is concerned.
It’s hard, though, especially when rogue comments louse up things.
Posted on February 20, 2004 12:23 PM | #
37. Steven said:
I personally don’t get the point of validation, maybe I’m still carrying scars from the ‘old days’ when ‘valid html/javascript/css’ still didn’t mean it came out the way it was supposed to look (thanks to browser bugs).
By the time one of web applications is finished and we’ve spent all our time getting the functionality right, making our database queries go fast enough and working our way through the backend code I frankly couldn’t give a d**n if my html comments are improperly nested or I haven’t provided ‘alt’ tags for my images (and neither does the customer)
The whole flaw of a computer is that unless you tell it exactly what to do in exactly the right way it doesn’t work. We have a medium (which like reality) accepts a certain degree of inaccuracy, why break your neck trying to achieve a perfection that no one will notice?
Posted on February 20, 2004 07:23 PM | #
38. Jeff Walden said:
Pat: what’s the difference between your code and that of htmlspecialchars()? I don’t have a blog or a website, but I’ve written one for someone else, and htmlspecialchars() seems to work fine for me (after I do some regexing to convert line breaks to <br />s, that is).
Posted on February 20, 2004 10:05 PM | #
39. Pat said:
Unlike htmlspecialchars(), my function leaves your tags alone, and avoids double encoding characters. So you don’t get any &amp; type mistakes, no matter how many times you run the text through the entityencode() function.
Posted on February 21, 2004 12:45 PM | #
40. Gez Lemon said:
Sorry to join this discussion late.
For people who dabble in web development, validation errors are excusable. If you’re a professional web developer aiming to construct a standards compliant website, there is no excuse for failing. I’m amazed that some of the responses in this thread have the audacity to blame their visitors. The visitor didn’t build the site, so how can they be held accountable? If developers allow visitors to use markup, validate the markup before allowing it to be submitted. Simple.
Web development must be the simplest profession in the world. Learn a few tags, and that’s it. Accountability in web development is almost non-existent. It doesn’t validate? That must be the visitors fault, or an error with the validation program you’re using. Push the issue of validation, and you’re a nerd. These excuses wouldn’t cut any ice in other professions. “Your house fell down when you were having a dinner party? Sorry, the warranty doesn’t cover you accepting visitors.” Design seems to be the only part of building websites that has any sort of professional ethic.
These comments aren’t aimed at anyone in particular in this thread. I think you’re all the most wonderful people I’ve ever had the pleasure of reading. I’m just playing devil’s advocate, as the devil seems to be light on friends at the moment.
Posted on February 21, 2004 04:32 PM | #
41. Kyle Fox said:
I sure hope you’re playing Devil’s advocate Gez, beacuse when you imply that you can’t be a professional Web developer and have validation errors you’re way, way off the mark.
I work on large (500+ pages) content heavy site with multiple authors and developers, all of whom I consider to be professional. Our site started out with no validation errors but over the course of time a few have crept in.
Like Keith says, it’s very hard to keep a valid site valid.
Didn’t he just point out that 90%+ of the sites he looked at, all professional I imagine, were invalid?
As far as comments go, you’re right that a visitor shouldn’t be blamed for that. I don’t think anyone was affixing blame, just point out that that was the reason for invalidity.
If not having a valid site casts aspersions on your professionalism I see a real problem…with your logic.
Posted on February 21, 2004 08:28 PM | #
42. Gez Lemon said:
Reading that again this morning, it does sound harsher than I intended, even in Devil’s advocate mode. But staying in that mode…
In other industries, professionals work to prescribed standards and their work is checked off against those standards. If testing is an integral part of the development stage, how do errors creep in? If testing isn’t part of the development stage, it doesn’t appear (to an amateur) to be a very professional approach.
Could it be that we have skilful and professional confused? It seems to me that developers spend more time on coming up with reasons as to why they can’t make a standards compliant website than addressing the issues and making a standards compliant website. I have absolutely no doubt that developers are capable of ensuring a website validates; it just doesn’t seem to be an important objective.
The number of invalid websites isn’t a new issue. This topic comes up time and time again, and most discussions on the topic degrade to a consensus that standards aren’t really that important. Does that sound professional? I agree that validation services don’t give the final verdict on the standard of a website. Validation services merely check that the rules of the chosen DOCTYPE are used correctly, but can’t determine that the semantics of the document are well chosen. From that, I would assume that a validating site was the very least a professional would expect to achieve. The fact that validation is so easy to perform, particularly if it’s done on an ongoing basis, makes it even more difficult to understand why professional sites don’t validate. Some industries that work to prescribed standards don’t have the privilege of an automated checker.
I don’t doubt for a second that the sites Keith tested that failed were produced by talented developers, and I certainly don’t mean to cast aspersions on any individual’s professionalism. I just find it difficult to understand the notion of a website built by a professional that doesn’t meet subscribed standards. Yet it seems to me, most developers are happy with that outcome.
Maybe you’re right, and there is a fundamental problem with my logic. Maybe there’s a very good reason why professionals produce websites that don’t validate, and I’ve just overlooked the reasons. What are they? Is it that the people who developed the website are talented, and we must be careful not offend them? Is it that standards just aren’t important? Although I could follow either of those two arguments, neither gives the impression of a professional. I didn’t miss Keith’s point that 90%+ of the websites he tested were invalid. I also didn’t miss the point that he only chose sites that were in some way proclaimed to be valid, but it doesn’t help explain the issue to me. If anything, it adds weight to my point about the lack of accountability. If there are other reasons that I’ve overlooked, I’d be very interested to hear them.
Posted on February 22, 2004 03:43 AM | #
43. Keith said:
Gez, I’m trying very hard not to be offended by your first comment. I’ve been a working Web professional for over 8 years. I’ve done quality work at every stop and on every project.
I didn’t start validating my sites until a few years ago. Not one. Ever.
I don’t really want to get into a debate about semantics but at it’s most basic, the word “professional” means you get paid for your work. I realize more is implied there. It’s just that there is so much more that goes into “professionalism” than whether or not you can keep your Web site valid.
There are many reasons why a site might not validate. Many (most) of them have nothing to do with the professionalism, skill or talent of the site’s developer.
I wonder if you’ve ever had to maintain a valid, Web standards based site. My guess would be no, as you seem to think it’s easy. It’s not.
When you have to make updates on a large Web site on a daily basis errors creep in. I validate Seattlechildrens.org as much as I am able. Can I test every page before it goes up? No way. And that is just one of many sites I’ve got to maintain.
I’m not the only developer on the site, some might not be as skilled or experienced as me, but they are professional. They don’t have a full handle on Web standards yet, but I need them to help me maintain the sites. To be honest, it’s not the best use of my time to maintain these at all.
We also have non-developers that provide content. I can be reasonably sure my own code is clean but I there is only so much I can do for the rest.
Right now the homepage doesn’t validate because when one of our media people added a link to a press release they cut and paste some unencoded apostrophes. Should I check every update before it goes up? Should I ask them to do it?
The answer is no. It’s not that big a deal to have a few unencoded apostrophes on the site. Now that I’ve seen it, next time I’m working on that page, and if it’s still there, I’ll fix it.
It’s a matter of resources, usually. I’m responsible for much more that coding that site. Maybe if, and this is a big if, we had someone dedicated to coding and only coding, this would be possible. It still wouldn’t be easy.
I’m responsible for code, design, usability, accessibility, project management, content management, editorial, documentation, testing and lots more. You’ll understand that pleasing the W3C’s validator isn’t my top priority.
There are also times when it’s not practical to make sure your site validates. If you’ve never come across a situation where you had to choose to break a page (as far as the validator is concerned) you’re lucky.
I’ve put way more time than is practical into making sure Asterisk* validates. I very often ask why I bother. Only the pedants really care and if it’s not validation it’s something else, a spelling error perhaps.
I’m not worried if I have a few errors. I realize that it is a lack of effort and knowledge on my part that errors come in via comments. I don’t blame anyone but myself for that, but at the same time Web design and development is about balance and I’m only one person who would rather focus my time on quality content and other things that will help make my site successful and worth my time. Encoding ampersands and apostrophes doesn’t seem to be time well spent if it takes away from other, more important, aspects of my site.
Maybe we should be talking about “successful” rather than “professional”. I feel my site is very successful when laid against the goals I set out for it. More than I ever thought it would be.
I think validation has very little, if anything, to do with the success of a site.
Look at Ebay, ESPN and Amazon. Are those sites successful? Are they built by professionals? I would hope so.
Not one of those sites validates. They don’t even come remotely close.
Saying, in any way, that a Web developer who doesn’t produce valid sites isn’t a professional is, frankly, a load of shit. There is so much more that goes into it and really, the validity of a site doesn’t show the whole picture.
If you’re able to dedicate the resources to every project you work on so that you can build and maintain your sites in a valid state, good for you. I’m not able to do this.
Maybe, if some day people start to give their Web sites the attention they deserve and provided the money and resources it takes to create and maintain the type of site your talking about, this could be true. But today most “professional” Web developers have lots more, and bigger, fish to fry.
Posted on February 22, 2004 10:09 AM | #
44. Gez Lemon said:
Keith, I have a great deal of respect for you, and a lot of other people who have posted in this thread. If I’ve offended you or anyone else, I’m really sorry.
If standards aren’t important, then validation isn’t an issue. If they are, then I’m struggling to understand why someone trying to meet those standards wouldn’t achieve them. Quite a few of the comments to this entry are claiming that it’s either too difficult, or not practical. Developers have complete control over what goes on a website, and plenty of tools to ensure they can’t be broken by visitors.
You’re quite right that you shouldn’t have to ask people to ensure their markup is valid, and there’s no reason why you should check every update; it should just be an integral part of the submission process. Working this way means you don’t have to put any resources into ensuring the site is valid, leaving you more time to focus on quality content. It’s a relatively simple task to write a basic script that ensures that entries made by contributors are valid. With a little effort, these kinds of scripts could be extended to correct mistakes rather than relying on contributors fixing them.
I take your point about the success of a site not depending on the quality of markup, and that the overall quality of the site depends on much more than web standards. I’ve concentrated on validation in my responses, as that’s the essence of your entry.
Posted on February 22, 2004 11:14 AM | #
45. Keith said:
Your tool sounds promising.
It didn’t seem to catch the cut-and-paste apostrophes, though, which is my biggest problem.
And it doesn’t help me at my day job at all. There is no form like that really, and I couldn’t expect our marketing folks to use it anyway.
So, it doesn’t work exactly right for what I’d need it to do, but I understand where your coming from and when, and if, I have time, I can look into something like that.
I feel that standards are important and I feel that validation is an issue. I can see how it might be hard to understand that I don’t really care about certain kinds of errors, but that is the way it is.
I do the best I can with what I’ve got. It can be a challange to make something of quality, that changes on at least a daily basis, when your making something out of nothing.
Thanks for the comments and discussion, and don’t worry, I’m not offended. You need a thick skin in this business, especially if you put your thoughts out there all the time like I do.
Posted on February 22, 2004 12:08 PM | #
46. Gez Lemon said:
Thank you for highlighting the cut and paste apostrophe problem, I’ve corrected it.
I’m glad I haven’t offended you, as I was just playing devil’s advocate. Along with Zeldman, Andy Budd, et al, you do a great job of highlighting issues that are important to web developers. I tend to watch from the sidelines on this sort of debate, but decided to get involved in this one, as there were some interesting views in the comments.
Posted on February 22, 2004 02:37 PM | #
47. Vlad Alexander said:
A large portion of the blame for sites that do not validate has to go to tool vendors, since it’s a lot to ask developers to make sure their sites are valid when they don’t have the right tools (code editors and WYSIWYG editors that produce clean code). But it’s not an unreasonable expectation that developers will do their best even if the tools don’t exist. However, developers really share part of the blame because they don’t demand more from software vendors. This is where professionalism comes in. You as a developer have to be professional enough not to settle for less from vendors. Demand better tools that do the job for you according to standards. In our standards-based approach to product development, we at XStandard are trying hard to make debates like this a thing of the past.
Posted on February 23, 2004 09:31 AM | #
48. Christopher Mahan said:
Can we have an icon that says: “Trying to be XHTML1.x” compliant?
Posted on February 25, 2004 02:11 PM | #
49. Dagfinn Parnas said:
The thesis How To Cope with incorrect html has now been moved to elsewhat.com. Please note that the page is work in progress, but the url structure is set.
The major new feature is that the each page of the thesis is converted to an image and is viewable through a browser (the old pdf was horrible and not to many people have a postscript reader). Hopefully I will get it converted to html soon.
Posted on October 26, 2004 12:18 PM | #
Comments are now closed