W3C home > Mailing lists > Public > public-html@w3.org > May 2008

Re: alt and authoring practices

From: Henri Sivonen <hsivonen@iki.fi>
Date: Sun, 4 May 2008 20:50:08 +0300
Cc: HTML Working Group <public-html@w3.org>
Message-Id: <9348D2D4-F820-4222-9C98-49BFC895F546@iki.fi>
To: Smylers <Smylers@stripey.com>

On May 4, 2008, at 18:29 , Smylers wrote:

> Henri Sivonen writes:
>
>> On May 4, 2008, at 11:19 , Henri Sivonen wrote:
>>
>>> When a user of Nvu doesn't provide a textual alternative to Nvu, is
>>> the result better or worse for the user experience of the page
>>> reader than Nvu omitting alt and letting UAs indicate the presence
>>> of an image in a self-consistent UA-specific way?
>>
>> What I'm trying to get at is this:
>>
>> Alt has three states:
>> 1) Not available.
>> 2) Specific text available.
>> 3) Conceal the presence of image from non-graphical rendering.
>>
>> A single text field without an accompanying checkbox can only handle
>> two states: #2 and either of #1 and #3.
>>
>> I think it's reasonable for authoring tools aimed at 'average users'
>> to only have a single text field, but I think in that case the  
>> overall
>> outcome is better if leaving the text field empty maps to case #1 (no
>> alt) than to case #3 (alt="").
>
> Why?

Because the generator doesn't know what the image is, so letting  
readers of the generated page to know the presence of the image is a  
safer alternative than concealing the presence of the image.

> The no-alt case is only permitted where the image is unknown to the  
> author;

 From the point of view of the markup generator, the generator doesn't  
have the alternative text when the user doesn't cooperate. The spec  
can say in strong words that the user must/should cooperate, but  
software written for 'average users' *will* face uncooperative users.

When the situation is that the user didn't cooperate, placing the  
blame doesn't really help the readers of the generated pages. However,  
compared to concealing the existence of images, letting the presence  
of the images be known might make the rest of the page make a bit more  
sense thereby making things a bit less bad.

> In particular if the page-generation software is being used to  
> insert a
> specific local image file into the page then the image must be known  
> to
> the author, and there's no excuse for not providing alt.

There may not be an excuse that you agree to be a good excuse  
according to your values, but a piece of software cannot coerce its  
user to provide *useful* supplementary data, since the user him/ 
herself doesn't need that data to accomplish his/her goal (in this  
case getting an image out there).

> The only
> decision is whether alt="" or some text is the best alternative, so
> I'd've thought it made sense for those to be the two possibilities the
> software presents.

If the software displays a text field and an unenlightened 'average  
user' hits enter and leaves the field empty, the fact of the matter is  
that a textual alternative was not provided--not that the user made a  
decision to conceal the existence of the image in non-graphical  
rendering.

A piece of software cannot extract an informed decision out of  
uninformed users, and software for 'average users' needs to deal with  
uninformed users.

-- 
Henri Sivonen
hsivonen@iki.fi
http://hsivonen.iki.fi/
Received on Sunday, 4 May 2008 17:50:55 UTC

This archive was generated by hypermail 2.3.1 : Monday, 29 September 2014 09:38:54 UTC