By now, many of you may have seen François Briatte’s recent survey of 10 web sites he reads on a regular basis. My props to François for assembling an insanely detailed, and very well documented and explained study.
Perhaps you’ve also already seen responses by Jon, Dave, and Eric. There’s no need for me to rehash anything they’ve already stated. I’ll just add a few notes relative to Stopdesign’s position within the survey, as well as my overall perspective of the results.
What I found particularly interesting was Stopdesign’s placement as the supposed “least dissident” site among those surveyed. I’m not entirely sure what to think of this. I wonder if being the least dissident is a good thing or not. The practical side of me says “Ah-ha, I’ve found a pretty comfortable middle ground compared with my peers.” The creative side of me wonders if I’m pushing the boundaries enough, or if I’m simply over-conforming to the norm.
The results aren’t really conclusive for anything beyond simple comparison data. But for some of the criteria, it’s certainly interesting to see who implemented what. If anything, it gets a few more people discussing the value or benefits of one approach over another. Being dissident does not imply doing the “wrong” thing. Maybe only one or two people have seen the light, the rest are still in the dark. Sometimes the least dissident folks are a good model of what’s currently accepted and practiced within an industry. But the dissidents might be the people we need to study more closely. Are they dissident by intent or by ignorance?
François spent a lot of time already putting together the data for just ten sites. What he’s done is a great start. Imagine if there were a means of using a Technorati type of service to rapidly collect this data automatically for thousands of sites? The results could be plotted over graphs. We might start to see more reliable trends in the areas surveyed. Some people could choose to follow the trends. Others, to purposely break from them.
As it stands now, the data is a little blurry in a few areas to draw any absolute conclusions. The sampling size is small (it was only put together by one person, so of course it’s small), the means of collection (by a human, and over the time span of a month) is not fool-proof, and a few of the questions were slightly vague. For at least one of the questions, François has old/inaccurate data (of which he is now aware), and another question’s results are debatable depending on interpretation of the question.
Question #18 asks “Is there a print style sheet?” The answer for Stopdesign is recorded as “no”, but the answer is now “yes”. Stopdesign used a print style sheet before the redesign, but didn’t for about a month after the redesign. I added another print style sheet for Stopdesign on July 2, which is right in the middle of his survey period, so it’s not necessarily François’ fault. He easily could have correctly recorded a “no” if he checked Stopdesign before that date. But since the survey period spans an entire month, any of the results for any of the sites could have changed during that period, as they obviously did with Stopdesign.
The second question I point out might end up more misleading due to a basic lack of specificity in the question. Question #19 asks “Does the navigation bar use image rollovers?” In this case, the recorded answer for Stopdesign is “yes”. But it’s only partially accurate, depending on the question’s interpretation. François clarifies that he was looking for CSS image rollovers. But does that question imply the navigation text is image-based, and a new image appears (or a new portion slides into view) for the hover state, changing the appearance of the text because a new image is revealed? Or does it simply mean some type of image appears behind (or around) pure HTML text? For one of those questions, Stopdesign’s answer is no, the other, yes. The same alternating answer is true for other examples cited in the survey, e.g. Dive Into Mark and SimpleBits. On the flip side, Zeldman.com and Mezzoblue use image replacement techniques to hide the HTML text and replace it with image-based navigation, so they answer yes either way.
Both of the questions mentioned above are currently at 60% consensus, which is a close call. Since there are only 10 sites in the survey, change the answer for just one of the sites, (as needs to be done for the print style sheet question) and the results all of a sudden become 50%. If there’s a 50/50 split, which direction is classified as dissident? Alter one value, and the dissidence percentage could change for every site in the survey. Tough call.
One note: We should be thanking François for doing this work in his own free time, and for starting the dialogue, rather than critiquing his selection of sites, or the number of them. He’s human, and only has so many waking hours in the day that aren’t spent on other critical work and personal tasks. Obviously it would be great to have lots more data, as I suggested above. But we gotta start somewhere. Thanks François.
14 comments
Seth Gunderson 4 years ago
What a fun little survey. Very hard for one person to do a semi-large pool over a month and NOT keep up with the web. Things change too quick. Like you said, it’s easy to whip up a print style sheet and if Fran√ßois has already passed you by, then you’re out in the cold.
Very interesting results, though I’m not sure what “least dissident” means in the overall scheme… so you don’t very that much from everyone else. I’d say that’s a sign that web standards are evolving to a point where it makes more sense to do it one way, rather than another - although doing things differently is not wrong.
Thanks François!
monkeyinabox 4 years ago
It’s a great chart for making design decisions. Should I underline links? Should I use rollovers? Should I have to make any decisions at all? Of course a little dissidence never hurt anything CREATIVE. :)
Andrew Phillipo 4 years ago
A few of my thoughts:
a) Designers sites are generally continually worked upon throughout their lifetime as a complete ‘design’ - this constantly moving target of admittedly both trends as well as ‘missing’ features is impossible to judge.
b) These are designers personal sites - hence they are done when the people involved have time. There are no commercial pressures to complete that print stylesheet or get that 404 up to scratch.
c) How long it might take an individual to change some of the features of a design. Many, many of the things on that list can be changed extremely quickly because of the wonders of CSS. For example in most cases I would imagine that you could underline links in 5 minutes, or even move away from image based navigation links.
d) The bias in the questions of the sites. The writer of this survey has clearly looked at every single one of the designers sites before writing the questions. In my opinion this has influenced him to ask some very in vogue questions.
Perhaps I’m being very harsh :-) but I don’t think that this survey does anything particularly helpful - IMO I can’t help but think that we should discuss the issues at hand as opposed to worrying about what the leading lights of design are doing and blindly following rather than understanding why they are important.
Howard 4 years ago
I happen to agree with many of the comments posted by
Andrew Phillipo especially where the “population” of sites is biased to only include sites that François frequently visits.
It’s like asking Republicans if they’re going to vote for Bush in the next election — they all are going to agree to some extent. A better approach would take a random sample of at least 100 sites and apply the same 25 questions.
My guess is that all of the sites mentioned in the study are very dissident from the “normal population.” But, that is why these sites stand out.
Douglas 4 years ago
When I saw Stopdesign as the “least dissident”, it was a small surprise. As Andrew says in #3, they are in many ways “in vogue” questions. Stopdesign takes the things in vogue, and uses them. Fixed width. Graphic header. Sans-serif font. Things which are good to have.
But then it goes on to do different things: different colour schemes for each section, split top menu (the top half of wich I only recently noticed), no vertical drop shadows. These are things which are so dissident that it isn’t even worth asking the question.
Douglas
phnk 4 years ago
“Things change too quick.” (#1) You said it, Seth ! Tantek also modified his layout during July, print stylesheets appeared here, Jon Hicks abandoned “Steal These” buttons… I was absent middle of July, when I came back I just corrected a few typos and wrote down a few notes, and published.
Andrew is right (#3), these web sites are moving targets. Did you notice that, on July 28th, a copyright appeared bottom of the latest Nielsen alertbox ? :)
Howard (#4), beware: the survey is not about making an inductive statement for all the Internet. I was focusing on web design gurus. That’s a microcosm (”how many really influential web design sites can you list?” was the question I asked myself in the first place), so 10 web sites seemed okay to me. I’m working on a larger base now, though.
All these comments are useful for writing an updated, more complete and broader survey. Thanks!
Jon Kennedy 4 years ago
Tell your creative side to chill.
As far as the trends show, you’re doing exactly the right thing. The redesign, though now old news, is by far the best looking design within the “circle”.
The fact that I, you, and anyone else can come here and USE the page with absolutely no problems, no learning, and no problems encountered, should be enough to appease the creative side.
Trends in design tend to show what works, and what is accepted as standard. With all the talks about designing with standards, one would think implementing these standards in a standard way would be a logical progression. Be proud.
It’s still fun to see where you stand though, ain’t it? I applied the conditions in Fran√ßois’ survey to some past sites I’ve worked on, and now see I have a bit of work to do in future endeavours.
Thalwil 4 years ago
How can the 40% of a 40/60 spit be regarded as dissidents in such a small sample (or in any sample)? Surey only an 80-20 or 90-10 split sould be regarded as true dissidence - ie the 20% who are ahead of the curve (or perhaps behind it).
One question which I’ve always struggled with: how far to conform to tried and tested design patterns and how far to dissent because I want to try something new or even just to mark an individual identity for a website. The latter path is always more tempting (too may website look alke these days) but usually quite hard, often ending in frustration and the realisation that patterns are as they are because they’re easy and functional. But I still want to be different.
Richard 4 years ago
interesting survey - but is it really dissidence to decide to not use underlined links?
to me the dissidence-issue isn’t that interesting - but many of the questions being raised are interesting.
- so i didn’t think about using accesskeys before and i had the impression that links should be differentiated (e.g. by prepending a symbol for insite-links and appending one for offsite-links) but not necessarily underlined.
- 404: yeah, many sites miss that
something like a poll and discussion on these questions would be cool…
Scott 4 years ago
Your site is incredible, dissident or not.
Steph 4 years ago
Imagine if there were a means of using a Technorati type of service to rapidly collect this data automatically for thousands of sites? The results could be plotted over graphs. We might start to see more reliable trends in the areas surveyed.
Yes, indeed. I came across the following site which had looked at major e-commerce sites in May 2002 and compiled interesting statistics on the positioning of navigation and that kind of thing.
http://www.webdesignpractices.com
The thinking seems to go that ‘dissidence’ from a well-observed norm risks diminishing usability, which makes sense for major e-commerce sites, I guess.
padawan 4 years ago
Statistically speaking, the error margin on such a tiny sample is +/- 10%, i.e. 20%, so I wouldn’t try to draw definitive conclusions about the dissidence column.
phnk 4 years ago
Gallup in the US (and Ifop in France) get 3% error margins on 1000-smaples, so I guess the margin won’t move unless I can find a way to send a clone of myself at work.
Derek Pennycuff 4 years ago
a very interesting idea.
it wouldn’t be hard at all to expand on this. the list of questions could be revised and expanded to say 35-40 questions, and a group of 20 people could each quickly run down the checklist for 5 different sites [for a total of 100] every 24 hours. i can’t imagine that would take more than an hour out of the day of each volunteer. the data could be collected and plotted and trends could be tracked as they emerge. such and such design element first appeared on this site on this date and over the next 3 months it became a common feature on 45% of the sites surveyed. not sure how useful such information would be. but it would definately be cool.
someone with a decent programing ability could probably even write a bot that could troll a given number of specific urls and determine answers to simple questions, especially if the answers are yes and no (or in the case of font family, serif and sans-serif). we could even take a lesson from diebold and have the program cache screenshots, html and css files so that there would be a verifiable “paper trail”.