Posted by Jukka K. Korpela on 06/02/07 20:52

Scripsit dorayme:

> If, whether for
> good reasons or bad, you have buttons that go to different pages
> as exampled in cwdjrxyz's page, it is entirely appropriate to
> have exactly the same alt text to help the user who sees no
> image.

No, it is not. It violates the basic principle that different links require
different link texts. There are serious usability and accessibility reasons
to this principle. When an image is a link, the alt text acts as the link
text in essential ways.

Try using a speech browser in "links mode", and you'll see. Er... I mean
you'll understand.

> I remind you that in his example, the go buttons were just
> devices that took you to answers to questions that were already
> displayed in text on their left.

The _links_ still have the same texts, and that's what matters to programs
that work on links. The "go buttons" are just pointless and childish. That's
why there's no reasonable way to write alt texts for them.

> There is no short "alt text for dummies" book,

There's an "alt text for all" page,
http://www.cs.tut.fi/~jkorpela/html/alt.html
which begins with a short summary that helps you write good alt texts in
most cases, and the rest helps with more complicated cases.

>> It would appear from the comments received that W3C approval can
>> still be obtained by simply choosing a non-descriptive ALT tag -- so
>> what is the value of W3C approval in this instance? Or have I totally
>> misunderstood the reasoning?
>
> About this business of alt="" for some situations.

To me, the question seems to be whether one could write alt="" just as a
method of getting away with validation, with no regard to the meaning and
purpose of the image. The correct answer is that it would naturally satisfy
validity requirement and would be either undescribably stupid or
disgustingly dishonest.

> You can do this if there is a clause in your contract to
> supply validated source.

Cheating is possible. You would be cheating if you did something wrong to
satisfy the letter but not the purpose and intended meaning of a clause in a
contract.

> Or if you simply cannot bear the sight
> of being rebuffed by a report from W3C. You can cheat and fudge
> to get over this line if you want.

That would fall into the undescribably stupid category. It's like cheating
when playing solitaire.

--
Jukka K. Korpela ("Yucca")
http://www.cs.tut.fi/~jkorpela/

[Back to original message]