Re: validator - how true is it?

On Sat, 15 Jan 2005, olivier Thereaux wrote:

> How would you suggest we could improve our message that "validation is
> good, but it is neither flawless nor an ultimate goal"?

May I? The critique was probably based on a) not understanding what a
markup validator is and b) bad experiences from people who don't
understand that either but wish to throw validator reports at other
people. But the question you raise is good, so I will make a few comments
(not that I wouldn't have made similar comments earlier).

To improve that message, first remove explicit statements and implicit
references to the contrary:

1. At http://validator.w3.org/ the first sentence says that the service
"checks documents like HTML and XHTML for conformance to W3C
Recommendations and other standards".
The concept of "standard" aside, the statement, as understood by common
people, is false. Passing the markup validation test does not ensure
conformance to W3C recommendations, not to mention other standards.
Think about WAI recommendations, for example. Passing the test does not
even ensure conformance to HTML recommendations. I'm sure you know this,
but the statement says otherwise. People do _not_ read it as you meant it,
as checking _some_ aspects of conformance.

(So what _should_ it say? "Checks documents like HTML or XHTML for
conformance to certain syntactic requirements". You can't use much simpler
language without saying something that simply isn't true, given the normal
meanings of words to normal people.)

2. In the validation reports, the exclamation mark in (very emphatically
presented) statements like "This Page Is Valid HTML 4.01 Transitional!"
is not only childish but also seriously misleading. It gives the
impression that validity is something to be really proud of, something to
be _shouted_. - Besides, the formulation is misleading in a deeper sense
as well, especially when the DTD used is not one of the DTDs given in
specifications but a customized one. Moreover, "This Page" is an odd
reference; in a document, you would expect it to refere to the page
itself, wouldn't you?

(Issuing simply "No errors detected in validation" or "N errors detected
in validation" would be better, though I know from a long experience that
people so often think that "No errors detected", e.g. from a programming
language compiler, means that there are no errors. But it would be an
improvement.)

3. Remove the Tips Of The Day. While most, if not all, of the tips
themselves are important and largely correct, this is not the place to
present tips. People keep misunderstanding them as relating to the
validation and even to their specific page being validated. Then they
start wondering e.g. why the tip tells them to use the alt attribute when
they have no elements that allow such an attribute, or when they have
used an alt attribute for every element that needs one. And there aren't
even tips _of the day_ (but apparently picked up at random).

4. Stop recommending that people add icons on their pages. I've explained
the reasons earlier, see
http://www.cs.tut.fi/~jkorpela/html/validation.html#icon
This is the most important single improvement. It also removes much of the
irrelevant or misleading content from the validator's report, thereby
promoting the idea that reports shall be simple, concise, and to the
point. In the current wordings, the following is especially bad:

"To show your readers that you have taken the care to create an
interoperable Web page - -"

In addition to recommending that people show off, it relates validity to
"taking care" and to interoperability. I think you know perfectly well
that validity does not guarantee interoperability. So don't say it does.
(The current formulation claims that, when normal people read it,
assigning normal meanings to words and expressions.)

But the entire text about validation icons should be removed. That's the
real cure. I know there will be difficulties because people now have icons
that link to the validator page, and they might be used to using them
for re-validation of their page, and the change in policy might raise some
questions. Such things can however be fixed by adding a link to an
explanation of the policy change, perhaps including instructions on
re-validation shortcuts that don't spoil one's page.

-- 
Jukka "Yucca" Korpela, http://www.cs.tut.fi/~jkorpela/

Received on Saturday, 15 January 2005 08:55:35 UTC