Comic Update: This is not a Reference
Posted by Kyle Weems on July 13, 2009[Update: The CSS3 Attribute Selectors article in the Reference was updated just prior to this post going live. So my ranting about that section is largely out-of-date and can be summed up now as "Took much longer than I'd anticipated".]
When I was first hired by Mindfly in 2007 I was not what you’d call “web standards aware”. Upon seeing the table-based layouts, font tags, and massive collection of inline-styles that stampeded through my pages like wild buffalo, grown men would gnash their teeth and wail in torment and mothers would hide their children.
It only took a few crying infants for me to realize something needed to change if I wanted to keep this career. My infovore nature led me to consume as much information on the topic I could muster, starting with Andy Clarke’s Transcending CSS and continuing through dozens of online tutorials and references. Learning the errors of my past, I spent a bit of time quietly taking my old sites out to the woods, instructing them to dig their own graves, then whacking them with a shovel before burying them for all time.
After the evidence had been destroyed, I went about trying to make compliant, pretty sites using the best practices in HTML and CSS. By 2008, I had friends who thought making website was the bee’s knees, but they didn’t know where to look for learning CSS, etc. At that point my bookmarks of handy sites had grown enormously, so I heartily recommended several.
One that I mentioned time and again was the Sitepoint CSS Reference, which was (to my knowledge at the time) a very complete, useful reference to the wonderful world of CSS. It even included tasty tidbits about CSS3 support. The main reason I sent each person who asked to this reference was explained in Sitepoint’s announcement: “…the reference contains a bunch of features that make it stand out from the pack — things like cross-browser compatibility charts and user feedback to ensure that it is accurate, up to date, and best-practice. If you’re building sites with CSS, this is a reference you’ll keep coming back to again and again.”
With the speed at which this industry changes, who wouldn’t want access to a constantly updated reference that even incorporated outside feedback?
There is a problem, unfortunately. As near as I can tell, the reference isn’t updating. Since its launch, it seems to be sitting still, failing to modernize its information as browsers march on. Every single browser on their compatibility list has had major updates since its launch, putting much of the CSS3 support information well out of date. Never mind that Google Chrome has been out for quite a bit of time now (as the Internet sees such things) and has no compatibility information present despite it’s higher browser share than Opera in most markets.
In a book, this situation is a necessary reality. Books, by their static nature, are out-of-date typically before they’ve even been published, requiring later editions, etc. But for a web-based reference, which claims specifically to have the benefit of staying up-to-date and incorporating user feedback, this isn’t terribly cool.
For me, the situation is exasperated by their promise to incorporate feedback (or even claiming to do so) when they (at least in situations I’ve experienced) clearly are not. To back my claim, I’d like like to direct you to my own experience I’ve had, which I’ll call Exhibit A. If you examine the page on CSS3 Attribute Selectors, you’ll find that it erroneously claims that Internet Explorer 7 completely fails to support these little gems.
I’m no IE fan, but I can tell you my friends, that this is a falsehood. And because I was foolish enough to take that advice at face value (who doesn’t trust Sitepoint?) I created IE-specific workarounds in a project where I first included these selectors, workarounds which ended up costing me a decent chunk of time. It was only later that I decided the best experience is personal experience and I actually tried the selectors in IE7, only to discover that they work perfectly fine. I’d wasted time (translate that: money) fixing a nonexistent problem they claimed existed.
Not being the type to hoard information, I shared the fact that they were mistaken in a comment on May 2, 2008, complete with a link to a test page to confirm that I wasn’t full of hot air. (The old test page has disappeared, so you can see what I’m talking about if you check this test page in IE7). Eventually they marked that they’d incorporated my comment into the article… only they hadn’t. It still incorrectly stated IE7 support didn’t exist. Sometime much later (aka, this year), I commented in annoyance at the mistake again on Twitter. They responded multiple times over Twitter to me, asking for clarification (which I gave) and then promised to update their Reference (which they didn’t).
That was a couple months ago.
Today’s comic shares my view on the so-called Reference, albeit in a somewhat abstract sense. So let me make it clear: I don’t think the Reference is what they claim: a reference. Rather, much like René Magritte’s unpipe it is not what it appears, merely the image of it. The unreference, if you will, is something that claims authority and completeness but increasingly lacks both as time moves forward.
One erroneous page isn’t worth tearing down their entire reference. However, with a complete lack of modern CSS support information on every major browser, Sitepoint’s “up-to-date” CSS Reference has become largely useless as a source for web designers living and working in 2009. I’m upset at this, because I sent literally everyone I knew with an interest in learning CSS to their site, saying “Hey, these guys know their stuff.”
Now… well, now I tell people to avoid it. I’ll repeat that for anyone reading: Don’t bother. They mean well, but they’ve failed to live up to their mandate of keeping updated. In March, when I’d commented (again) on my disgruntlement with the lack of updates, I received the following pair of tweets from Kevin Yank (@sentience).
April 6, 2009 4:19pm: “What erroneous compatibility info did you find? We are planning to refresh the Reference for the latest crop of browsers in May.” (after which I gave a summary).
April 6, 2009 4:50pm: “Thanks! Will get that corrected ASAP.“
It’s July now. I think we may have different definitions for “corrected” or “ASAP”.
Tags: Comic, CSS, rant, reference, René Magritte, sitepoint, this is not a pipe
I love the Sitepoint reference, but I’m equally frustrated that its compatibility listings are horribly out-of-date. I can forgive lag time, usually, but they don’t even cover Firefox 3 on there, and that’s been out for over a year.
This is a real problem with putting up a reference to web technology. As Ian Hickson has stated: There is no stability here
The moment you do you are either on the hook to keep it up to date forever (and updates start from the moment the site goes public) or you become obsolete. It happened with quirksmode, webdevout and others.
Over on the openweb-group mailing list we’ve talked a bit about this problem and one option is to en-masse start contributing to Google’s Doctype A wiki-like format is one way in which the constant update problem can be ameliorated.
A wiki-like format is only desirable if you trust all of your contributors, Jeff. Otherwise you have a metric ton of constant updates that have to be checked for accuracy. I talked at length about different methods of doing compliance charting back in 2004, in case anyone’s interested in reading that.
I’m not going to bust on SitePoint too much, because I can say from long and deep personal experience that creating and maintaining support information is really, really hard (see previous link). They probably deserve a few dings for not living up to their high-flown rhetorical promises, but they’re hardly alone in that either. How many times have we seen and heard big splashy announcements of sites and groups and resources that just fizzled out?
@Erik Vorhes – I agree on the scope of time involved. I’m not expecting Firefox 3.5 on there already, but you’d think Firefox 3 has been around long enough for it to show up on the charts.
@Jeff – A wiki is a neat idea, but…
@Eric Meyer – …I agree the major obstacle for a wiki is trust. There’s no reason to believe a wiki would be any more accurate (unless, I guess, it had high volume of obsessed editors that constantly are prowling the recent changes list). I’ll admit, though, that I thought about suggesting a “CSS Wiki” as a possible alternative to Sitepoint’s model.
My major beef is their failure to incorporate feedback as they indicated. Still, considering the level of upkeep such a resource requires, I’ll give them a bit of slack if they do get around to updating eventually (which I’m hoping is merely delayed and not canceled.)
Thank you for the link, Eric! It really helps put some great context on the challenges for such a project.
It’s true – just like Wikipedia, you need a large support staff whose energy has to scale with the number of contributions. But unlike Wikipedia the scale of what needs to be documented is much smaller. My own suggestion was that browser vendors can foot the bill for one support person a week to help administer – that would at least cover 5 days a week [1]
The reason I bring up a wiki though is because we’ve seen the pattern before: Someone gets a passion to document support scores for a spec. It has a burst of popularity and the author starts to be considered an expert. Maybe the website author goes on to write books and speak at conferences and what happens
Of course the website becomes obsolete. It’s happened over and over again in little data silos all over the web for a decade. Eric’s pages, quirksmode, webdevout, Fyrd’s When Can I Use? This is not to blame anyone! It is simply the nature of the beast.
The beauty of putting the data into an open system is that when person A loses interest, person B can take over. The bad part of this is a loss of whuffy.
Jeff
[1] http://groups.google.com/group/openweb-group/msg/832e59bf5c5be898
Hooray, we’re in a comic! Oh, wait…
Your points are well made, Kyle. The team here at SitePoint, as well as our amazing authors Tommy, Paul, Ian and James, worked tirelessly to assemble an up-to-date (at the time) reference site which, apart from a major redesign several months back, has sat largely unchanged since its launch. We’ll wear that.
The plan when we designed the SitePoint Reference was for the reader comments at the bottom of each page to bridge the gaps between updates to the site. Without sacrificing the editorial polish of the main content, we hoped, these reader comments would serve as a public errata/update list that we could periodically roll back into the reference.
Unfortunately, the SitePoint Reference didn’t generate the level of initial traffic (and, by extension, income) that we had hoped. The resources that would have normally gone into updating the Reference were therefore diverted to finding ways to generate additional traffic (like building our CodeBurner reference extension for Firebug), and to other more profitable pursuits. Reader comments like yours have contributed a certain amount of updated information to the site, but we’ve learned that authoritative-looking content like browser compatibility tables aren’t especially inviting to comments.
In a bizarre coincidence, the fix for the specific issue you raised (support for attribute selectors in IE7) went live on the Reference yesterday—mere hours before you published your comic and this blog post. It took much longer than it should have, but we finally got around to it. Another update—a clarification to the “HTML versus XHTML” page in light of the recent termination of the XHTML2 effort—is being deployed as I write this. So we have been making small updates here and there, but we could be doing a lot more.
When we exchanged tweets in April, the release of IE8 had just made it clear that the Reference needed a major update. At the time, our plan was to quickly update the compatibility tables for the latest browser releases, including FF3 and IE8. It seemed like something we could get done in a month, but competing priorities conspired against us. With FF3.5 and Opera 10 looming, we decided that the Reference as a whole needed a refresh to cover newly-implemented features of HTML5 and CSS3. As I write this, we are in the process of commissioning authors to make those updates, which we hope to publish in real time, putting up new content as the authors write it.
All this to say we know the SitePoint Reference isn’t the living document we had hoped it would be. Much of that comes down to business decisions we have made based on the level of traffic and income it has generated for us so far. We are working on ways to increase those so that we can justify devoting more resources to its maintenance, but in any case we are determined to conduct a wholesale refresh of the site’s content sometime this year (hopefully sooner rather than later).
I’m not sure why your comment was marked as “integrated into the reference” when it wasn’t; we might have to chalk that one up to human error, with our apologies.
@Kevin – Thanks for taking the time to address my concerns!
Eric’s earlier comments helped highlight the fact that such an undertaking isn’t a small one, so in that light I can understand why you’d not devote heavy resources to a rigorous update schedule when business realities indicated that it made more sense to concentrate elsewhere. I admit I easily divorce the reality of business costs associated with feature support in my head when I’m discussing these kinds of things.
I’m very glad to hear that the reference will be getting updates in the future, as it was a very valuable resource for me when I was starting out.
@Jeff – I still find whuffy to be a funny word.
Perhaps with browser support charts, the “keys” could be given to someone who represents each browser (such as me for Opera) and each rep could update that browser’s support info on a wiki/ reference?
And yes, Kyle, whuffie is a very silly word.
@Bruce – That’s an interesting concept. I imagine Opera and other standards-obsessed browsers would be quite diligent with this. I’m not sure, however, that Microsoft (as an entirely random example) would be keen in investing time into documenting what they’re not doing right with CSS. If everyone could be somehow trusted to be forthright with such information, then it’d be a great idea.
Au contraire, Mr Weems. Microsoft have been diligent in documenting their features against two of their competitors http://www.microsoft.com/windows/internet-explorer/get-the-facts/browser-comparison.aspx. Ahem.
@Bruce – Well then, I stand corrected.
*cough*
Looks like it’s finally been updated to cover all the modern browsers! Huzzah! …Although for some reason it’s covering Chrome 2.0 when they’re up to version 4.0. Meh, I’ll take it.