I'm looking for an objectively based survey of websites which use css-based positioning versus tables as layout.
June 12, 2008 6:42 AM Subscribe
I'm looking for an objectively assessed survey of websites which use css-based positioning versus tables as layout.
I'm not looking for inspiring sites or well-designed sites, just a recent survey of websites using css-based layout. Ideally it would be something like this w3schools statistic chart on Browser statistics and ideally it would also offer a comparison of table-based layout sites. My guess is anyone doing this would be using a representative sample as the sheer number of sites is overwhelming.
I'm not looking for inspiring sites or well-designed sites, just a recent survey of websites using css-based layout. Ideally it would be something like this w3schools statistic chart on Browser statistics and ideally it would also offer a comparison of table-based layout sites. My guess is anyone doing this would be using a representative sample as the sheer number of sites is overwhelming.
I'm not sure that this exists. I don't think there's a programmatic way to tell if a site is using tables or css for layout.
And besides, the numbers for these sorts of things are never the complete truth. For example, the w3schools browser stats are based only a sample of people who visit that site. And who visits that site? People who make web pages, which is hardly your average person. So you can't trust those numbers. It might get you in the ballpark, but it might also have a 20% margin of error. The same would be true for these css stats you seek.
Since we don't know what you need this information for, my recommendation is that you do a survey for yourself. Pick a bunch of sites, maybe a good mix of famous ones, maybe unknown ones, and look at the source code. You could even make a project out of it and publish your results (on the web of course).
posted by kpmcguire at 9:07 AM on June 12, 2008
And besides, the numbers for these sorts of things are never the complete truth. For example, the w3schools browser stats are based only a sample of people who visit that site. And who visits that site? People who make web pages, which is hardly your average person. So you can't trust those numbers. It might get you in the ballpark, but it might also have a 20% margin of error. The same would be true for these css stats you seek.
Since we don't know what you need this information for, my recommendation is that you do a survey for yourself. Pick a bunch of sites, maybe a good mix of famous ones, maybe unknown ones, and look at the source code. You could even make a project out of it and publish your results (on the web of course).
posted by kpmcguire at 9:07 AM on June 12, 2008
It'd be possible to programatically check to see if a given site contains table tags or CSS declarations, but I'm not sure the results would be meaningful, because it's not a strict either/or question: most sites use a combination of the two techniques.
Sites which use no CSS at all either haven't been updated in years, or were built by amateurs. (Not that there's anything wrong with that.)
Sites which use no tables at all were generally designed that way to prove it can be done, or by somebody who writes for A List Apart, or would like to. (Not that there's anything wrong with that, either. :)
Nearly everybody else (eBay, Amazon, NetFlix, MeFi, Yahoo, Google, Apple, Microsoft, for a start) is somewhere in between, either because they've been updating their HTML only incrementally as needed, or because some layout tasks can still be done more cleanly using tables than pure CSS. (Unfortunate, but true.)
posted by ook at 1:28 PM on June 12, 2008
Sites which use no CSS at all either haven't been updated in years, or were built by amateurs. (Not that there's anything wrong with that.)
Sites which use no tables at all were generally designed that way to prove it can be done, or by somebody who writes for A List Apart, or would like to. (Not that there's anything wrong with that, either. :)
Nearly everybody else (eBay, Amazon, NetFlix, MeFi, Yahoo, Google, Apple, Microsoft, for a start) is somewhere in between, either because they've been updating their HTML only incrementally as needed, or because some layout tasks can still be done more cleanly using tables than pure CSS. (Unfortunate, but true.)
posted by ook at 1:28 PM on June 12, 2008
or because some layout tasks can still be done more cleanly using tables than pure CSS.
Or... because they are displaying tabular data, which is what tables were intended for.
posted by DarlingBri at 1:36 PM on June 12, 2008
Or... because they are displaying tabular data, which is what tables were intended for.
posted by DarlingBri at 1:36 PM on June 12, 2008
Response by poster: Yeah, honestly I'm looking for anything, I was hoping someone had already done the survey.
posted by jeremias at 4:17 PM on June 12, 2008
posted by jeremias at 4:17 PM on June 12, 2008
I think you'll find a number of people with strong objections to the word "cleanly" used to describe table layout, ook...
Nthing the idea that it's not possible to even do the count.
Devil's-Advocate-Filter: It might be possible to guess at the use of tables for layout as opposed to their use for, well, tables, by algorithms like "is there a <table> tag directly after the body tag?" or "what proportion of the page is contained inside <td> elements?" but it would only be a very very broad guess and subject to a ton of false positives/negatives. And you'd have to cross-reference it over time to make sense of it anyway. All sites before a certain date had table layouts, or strictly linear content, so what would be the point of counting them?
So, to even try and do the survey would be impossible.
But what would numbers prove either way?
The use of CSS/tables for layout is essentially a moral question. It's wrong to use tables, it's right to use CSS. You're not going to make a convincing argument about right and wrong by quoting numbers.
posted by AmbroseChapel at 8:25 PM on June 12, 2008
Nthing the idea that it's not possible to even do the count.
Devil's-Advocate-Filter: It might be possible to guess at the use of tables for layout as opposed to their use for, well, tables, by algorithms like "is there a <table> tag directly after the body tag?" or "what proportion of the page is contained inside <td> elements?" but it would only be a very very broad guess and subject to a ton of false positives/negatives. And you'd have to cross-reference it over time to make sense of it anyway. All sites before a certain date had table layouts, or strictly linear content, so what would be the point of counting them?
So, to even try and do the survey would be impossible.
But what would numbers prove either way?
The use of CSS/tables for layout is essentially a moral question. It's wrong to use tables, it's right to use CSS. You're not going to make a convincing argument about right and wrong by quoting numbers.
posted by AmbroseChapel at 8:25 PM on June 12, 2008
Response by poster: Hmmm, I'm going to chalk this up to poor question framing on my part. I think I cast too wide a net.(Sorry for the pun)
Trying to do an exhaustive automatic count of the entire world wide web is not my goal. I was hoping for any sort of data, even an informal survey: i.e. of the top 50 sites on Alexa, 30 primarily use css positioning for layout, 20 primarily use tables.
guess I 'll go do it myself (shuffles away)
posted by jeremias at 5:20 AM on June 13, 2008
Trying to do an exhaustive automatic count of the entire world wide web is not my goal. I was hoping for any sort of data, even an informal survey: i.e. of the top 50 sites on Alexa, 30 primarily use css positioning for layout, 20 primarily use tables.
guess I 'll go do it myself (shuffles away)
posted by jeremias at 5:20 AM on June 13, 2008
« Older Garlic and onion growing and harvesting advice | Call and response where we all know the response Newer »
This thread is closed to new comments.
posted by amtho at 7:20 AM on June 12, 2008