RE: Fw: Announcing WebMetrics 1.0 from NIST

This is a quick summary of my cursory experience with the new WebMetrics
tools from NIST. I tried out the WebSat Static Analyser Tool, which can be
run from their website - the others must be installed on a server to run.

The new webpage analysis tools are very interesting, since they test the
"useability" of pages on several broad principles, of which accessibility is
only one. Performance testing is not unusual but the readability test is,
and I don't think I've ever come across a "maintainability" test before.

The presentation of the results is a little cryptic. Some rules report
'Yes|No', some a number; some results display a value when the rule is
passed while most are left blank. The explanations as to what the results
mean are on several other pages which are NOT linked to on the results page.
I recommend reading these explanatory pages before running a test.

The accessibility tests are not based on the W3C-WAI's draft accessibility
guidelines. For example the use of tables to create multi-column text is not
flagged as a problem for screen readers (Thanks to those that pointed this
out to me not too long ago!).

Some of the other tests can return some strange results. For instance, one
rule calculated the average number of words in text links, and raises a flag
if the number is less than 2 or greater than 5. A typical navigation bar
will run into trouble here. The explanation mentions the common "click here"
mistake but the rule doesn't check for it.

The "Maintainability" rules will flag a page if it contains an absolute link
(ie a link containing a root domain, as opposed to a relative link to a page
within the same site). The rule is supposed to remind the designer that
links have to be checked and maintained regularly, but it's hardly an
accessibility error. My pages were flagged for containing a link to the HTML
4 definition in the DOCTYPE tag.

The Readability rules also flag the use of horizontal rules, which
apparently may be confused for the end of the page. The page containing this
explanation uses four horizontal rules.

Overall, it's an interesting tool, but like most HTML validators it flags
elements that in practice cause no problems, and I think the explanations
are inadequate in defining what problem it's testing for. Like all HTML
validators, the results should be taken with a grain of salt, and more than
a grain of HTML knowledge.

I'd like to see the option to turn off various rules, especially if I was
testing across multiple pages, or a whole site.

Sean Lindsay

Received on Thursday, 4 June 1998 19:59:50 UTC