Web Standards in Estonia vol 2

Table of Contents

• 1. Preparations
  • 1.1. Web pages included
• 2. Results
  • 2.1. Excluded pages
  • 2.2. Valid pages
  • 2.3. Document types
  • 2.4. Encodings
  • 2.5. HTML elements
• 3. Conclusions

1. Preparations

Half a year has passed from the time when I conducted the before mentioned survey. So, it’s interesting to know, if there have been any shifts in the quality of the web pages in Estonia. Let’s just pull out the good‐old script and see...

But the first web site I noticed to have been changed is the site of W3C Validator. And it came out, the script does not work any more. Happily, few minor changes seemed to fix everything. Although, after the survey was completed, I noticed the script didn't identified when a web page was tentatively valid – instead it gave error-message "A Mystery Error". Does this ruin the results of this survey?

At the previous survey there were 94 tentatively valid pages and error‐log contained 14 error-messages titled "A Mystery Error" (I still don’t know what is the exact reason for this error – that’s why the name for the error). At the current survey there were 116 error-messages titled "A Mystery Error". But as 94+14=108 which is only 8 pages less than 116, then it looks like all the number of tentatively valid pages hasn't probably changed a lot. Was there a bit more tentatively valid web pages or a bit more pages which caused the error, doesn’t really matter.

I think it shows, that these small errors in the script do not really alter the results of this survey, but it’s completely yours to decide – no‐one should be forced to believe the results gathered with obviously buggy software.

1.1. Web pages included

22,735 pages took part from this survey (taken again from the Neti.ee Estonian servers list), which is 830 pages more than in previous survey.

The survey was conducted between 2005-08-27 11:41 and 2005-08-29 15:52 (Estonian Time).

2. Results

This time I only mostly list the differences between this and previous survey.

2.1. Excluded pages

The program excluded 2542 pages before validation, so that 20,194 pages only made the W3C Validator test. In previous survey the number of pages excluded was 2250 (292 pages less). About hundred of the excluded pages were those which were tentatively valid. So what about the growth of 200?

The main reason for excluding was (as in previous survey), that the page was redirected. The second main reason was not being able to connect to the server (also as in previous survey), but this time there were 867 of those pages, which is (compared to the previous 693) 174 pages more.

I don’t have a good explanation for this – maybe there was some server down? I remember an incident from the end of spring, when one Estonian company, who offered free web hosting with the ability to use personal domain name and a lot of other things, suddenly quit offering the free service – maybe this has something to do with the number of pages out-of reach. Or it could also be that the server of this very same company is just down – very possible indeed...

2.2. Valid pages

The number of valid pages has risen from 436 to 440 (only 4 pages). At the same time the number of invalid pages has risen from 19,125 to 19,753 (628 pages!). And so the proportion of valid pages has fallen from 2.22% to 2.17% (or from 2.23% to 2.18% if we exclude the data of tentatively valid pages).

In all the following comparisons I exclude the tentatively valid pages explicitly, it clearly doesn’t effect the outcome.

When looking at the numbers above it might look, that the set of valid pages is fairly static – after all, only four valid pages more this time. But actually the changes in this set were not limited with those four pages. Compared to previous survey, 186 pages are not any more in the set of valid pages, and in the place of those are 190 new valid pages.

You can take a look at the list of new valid pages.

2.3. Document types

If previously 35.29% of invalid pages had doctype specified, then now it has fallen to 34.98%. Minor change, but indeed.

The distribution of document types has mostly stayed the same. Only HTML 3.2 has dropped from forth place (481 pages) to fifth (213 pages). And HTML 4.01 Frameset has taken his place.

When we look at only the valid pages, then the XHTML 1.0 Strict has climbed from forth place (23 pages) to third (31 pages) and therefore switched places with HTML 4.0 Transitional

As can be seen from above, the use of XHTML has grown and the use of HTML dropped.

2.4. Encodings

The number of pages, which specify encoding has risen from 68% to 72%.

The encodings popularity-table looks almost the same as in previous survey:

Interestingly all the windows-family encodings have lost popularity and ISO-family encodings gained it. But as iso-8859-1 does not de jure allow all Estonian characters and windows-1252 supplements the latter with those characters, this is not a good sign at all.

Happily the use of utf-8 has increased quite a lot.

2.5. HTML elements

The amount of pages with tables and images is growing. But the most remarkable growth is in the amount of pages who use <script> and/or <link>. At the same time the popularity of <div> and <span> has grown the most. Contrary to the rise of <link>, <style> has become less popular. This clearly indicates the growing use of style sheets, and also the growing use of scripting.

The use of frames (elements <frameset> and <iframe>) has fallen.

The use of elements <center> and <hr> has fallen most.

Although the use of all the other deprecated elements has fallen, the use of <font> has got quite a growth.

For some strange reason there are now more pages without <html>, <title> and <head>. Also strangely the use of <h1> has dropped, but at the same time the use of <h2> has gone up.

3. Conclusions

It is good to note, that the use of stylesheets, UTF-8 and XHTML is growing. The use of old HTML document types (like HTML 3.2) has dropped significanly, and frames and center‐tags are being used less.

On the dark side, the use of scripting is growing with a really fast rate, and the practice of using tables for layout and images for decoration has gained even more popularity. Also the use of the terrible font‐tag is still growing.

This growing use of bad practices seems to confirm the numbers – although there are more good sites now, done entirely in the best spirits of valid HTML and CSS, even greater amount of pages continue to use all kinds of obsolete techniques.

Is this the trend in the whole world? Or does it only reflect the situation in Estonia? (Which sounds quite possible, as we have no laws that require government sites to pass accessibility tests as many other countrys do (on the good side: the home page of the President of Estonia is valid!))

Hopefully further studies will provide answers to these questions.