On the question of authoring style, there's some very instructive early discussion in the Style Guide for Online Hypertext at the W3C, with particular reference to Don't format for a particular browser and Avoid talking about the Mechanics.
After all, the author of a book does not normally try to tell the reader how to read it, what light to read it in, how to use the fold-out insert etc. unless there were some very special reason to do so. You'd assume they already understand how to do that, or have the freedom to choose for themselves, wouldn't you? So, I would ask you to kindly not patronise your readers by implying that they don't know how to use their own WWW browser.
While it's true that at a time of explosive growth of the WWW, many readers will be newcomers and maybe not fully familiar with their browsers yet, your article on wild mushrooms isn't going to be improved by turning it into yet another browser-use tutorial. If there's some special requirement on your page, then I'd recommend you politely suggest your users visit a tutorial on the use of their particular browser. Bear in mind that they might not be using the same browser/platform that you use, and if you try to tell them how to use your browser, it's quite possible that you will confuse them.
I make no claim here to be offering the one and only correct answer, or even necessarily the best answer, to any particular situation discussed. In each case, there will be quite a range of solutions that would fulfil the requirements of getting the desired message to readers of all kinds. There will also be a wide range of purported solutions (as seen repeatedly on the WWW) that quite unnecessarily fail in a significant range of browsing situations: no-one can reasonably object in a situation where the nature of the material itself precludes it being available - what is being objected to here is material that is inherently textual in nature being pointlessly made inaccessible.
My chief aim has been to bring the subject into the open; to
provoke further reasoned discussion; and to address the many
ALT texts that I am seeing on the WWW.
In doing so, I hope also to demonstrate that the often repeated
claim that "authoring also for text-only users would cost us an
unreasonable additional effort which we cannot afford" is a
mistake: this claim seems plausible only if you don't know
what it is that you are trying to say to your readers, and have
confused the presentation details (which in any case cannot be
closely controlled by writing HTML for the WWW) with
the content of your message.
I can't resist mentioning here that the HTML3.0 draft
recognized the shortcomings of the
IMG tag by
introducing a new and much more versatile tag,
which was supported by one or two advanced browsers, though
not by the mass-market ones;
finally, the HTML4.0 recommendation offers a re-worked version
of that as the
However, within the scope of this note I am specifically dealing with
the appropriate use of the
OBJECT is still not well supported by browsers
at the time of writing.
There are many possible reasons. If you, as an author, assume that you know the reason (for example, if you assume that text mode is only used by people with a low disposable income, and therefore of no interest to you), then you will very likely exclude - and annoy - some people that you really would like to have as customers, even though you didn't realise it. Apart from that, if you are making available some textual information, then what logical reason could you have for fencing it away behind an impenetrable thicket of graphics-only material?
One of the most important readers of your WWW page is the indexing robot: it is, in effect, blind, and cannot understand your images. The ALT text is an excellent way to help the robot and ensure that your page will be indexed appropriately. Some authors seem to prefer spending lots of effort on composing META tags for the indexers - which is a reasonable enough idea in itself - but that produces no visible spinoff for your text-mode human readers, whereas effort spent on your ALT texts produces benefits for both kinds of "reader". And there has been so much abuse of the META tags by some web-page providers, that certain indexers now ignore the META entirely, or apply strict disqualification rules to avoid getting fooled by that kind of trickery. With 30Mio+ URLs on the WWW, you really don't stand much chance of an interested reader finding your page by their own unaided efforts: everything you can do to get it indexed properly is giving you an extra chance to reach readers who are specifically interested in what you have to offer them. These are your "quality" visitors, each one far more valuable to you than an army of readers who might stumble onto your URL while "surfing".
1. In Europe, at least, the digital cellphone (GSM or DCS-1800) is now a commonplace. Increasingly, these digital cellphones are coming equipped with a digital data socket, to which a more elaborate display, for example a laptop or palmtop, can be connected, either for display of SMS messages or for a fullscale dialup digital data path to an Internet provider etc. However, the data rate that is available is quite limited, compared with a wire telephone, due to the broadcast nature of the transmission path. Within the defined transmission standards, it is inconceivable that any large scale browsing of images is going to be achieved with an access time that would be acceptable for routine browsing.
2. In the Hitchhiker's Guide to the Galaxy by Douglas Adams, The Book is described as a handy device having a display about four inches square, capable of displaying any one of a million pages of information. It's clear that The Book is conceived as a self-contained system, as would be appropriate for the intergalactic traveller. Here in the real world, we have palmtops (e.g the Psion, the Newton) that are getting better all the time; and we have digital cellphones, that are getting hooked up to palmtops. See for example the On-line with your Psion page at Steve Litchfield's 3-Lib site. So, something rather like The Book, but giving access to not just a million pages of information stored within it, but the 30million-plus (and still counting) documents on the WWW.
This will only be practical for text, and simple graphics. Those humungous images will definitely not be retrieved routinely: this is a fundamental limitation of the transmission path, not just a minor technical obstacle to be overcome.
In fact, subsequent to my writing this section, several products consisting of a cellphone with integrated web browser, or cellphone/palmtop combo, have appeared, e.g the Nokia 9000 series and the Psion 5
3. The Speaking Machine - not only for the blind. Busy executives might have a speaking machine read out stuff to them while they are driving between business meetings. Also dial-a-web-page telephone services etc.
4. Some sceptics on the WWW usenet groups, when an unusual or futuristic presentation device is mooted, appear to believe that they gain some benefit by deliberately shutting out those unusual devices, rather than exploiting HTML's ability, when used appropriately, to put their information into every reader's browsing situation. This shows a very narrow minded and short sighted view, unworthy of the high principles of the WWW. Why "author for" a restricted reading situation, when HTML (at least where textual material is concerned) enables you to "author for" all reading situations? The oft-repeated phrase about "target audience" refers to those people to whom you are aiming your "topic of discourse": applying yet another layer of selectivity according to their browsing situation (except where the content by its very nature compels you to) seems to be a misunderstanding of what "target audience" or "target readership" is about.
Even 0.01% of internet users is still a significant number
of individuals, especially if they just happened to be the
people who might be interested in your product: how can you be
sure they're not?
Truly it has been said that many WWW pages have
for arguing with customers,
and things seem to be getting worse rather than better.
Anyway, here's my very incomplete list of why readers might choose to use a text-only browser, or a graphics browser with auto image loading turned off.
Right, you clever authors that say text mode readers aren't worth bothering about: what was the common factor in all those users, apart from the fact that they use text mode (whether from choice or from necessity)? I reckon there wasn't one. Just because you think you know the "profile" of your target audience, does not mean that they will all of them fit that profile, all of the time. When you advertise in the wrong magazine, you merely miss some of your potential customers: no active harm is done. When you offer WWW readers something which, it turns out, they cannot read because it's been designed for a limited range of platforms, you have been actively rude to a potential customer. You only have to annoy them once, to send them to your competitors. Just one disgruntled customer can generate a lot of negative advertising, and lose you other potential customers.
I have to confess to having rather little experience of writing WWW documents specifically for blind readers. The issues are a hot topic at the Web Accessibility Initiative area at W3C.
Frankly, many of the pages that I see on the WWW are so obviously in need of a degree of care and attention that would help all text-mode readers, and so often I have received the response from unsympathetic authors who refuse to spend their time adjusting their pages for what they see as a small minority group; therefore I am concentrating on techniques that seem to be beneficial for all text mode readers, and by implication also for blind readers as well as for indexing robots, rather than convincing such sceptics to spend additional time enhancing their pages specifically for blind readers. I hope my intentions in proceeding this way will not be misunderstood: I applaud those authors who would additionally like to take the time to make their documents even more accessible, and who are willing to follow appropriate guidelines.
Note that in HTML4.0 there are three or four attributes available for their respective purposes (I'm paraphrasing the HTML4.0 recommendation):
How best to use these mechanisms is still under active discussion in detail, but we can see that there is even less reason now to use the ALT attribute for mere descriptions of the image, in general.
For legacy browsers, the forerunner of the LONGDESC attribute is the D-link, pioneered by WGBH.
I received an interesting email from someone who was attempting, unsuccessfully, to use FONT COLOR to get their ALT texts to contrast properly against their dark-ish background.
It's been my experience that the ALT text shows up in the same colour as is specified in the TEXT attribute of the BODY tag, or in some browsers if the image is a link then it will use the LINK or VLINK colour as appropriate.
FONT COLOR is dangerous in Netscape anyway, because it cannot be disabled by choosing the "use my colors" option. Thus, the reader who is in a difficult browsing situation and trying to make matters better by controlling the colours themself, will get the author's chosen FONT COLOR displayed against the reader's choice of background. In some cases this makes text totally invisible. This issue is discussed in more detail in Warren Steel's FONT rant.
MSIE and Opera don't have this bug: when they are set to use the reader's colour scheme, it overrides FONT COLOR as well as the other colour attributes.
Because of Netscape, I would recommend avoiding FONT COLOR whenever possible, and only using it to specify colours that are expected to be an effective contrast against any plausible background colour. Use of the colour attributes of the BODY tag, by comparison, is harmless, because all browsers can override it when the reader finds it necessary. Of course, if you use a background image then its colour needs to be consonant with your BGCOLOR attribute.
I recommended my correspondent to set the basic colour choices for their page in the BODY attributes only. In my experience this gives good results with the ALT texts on a good range of browsers.
The contents of this article were originally published at http://ppewww.ph.gla.ac.uk/%7Eflavell/alt/alt-more.html, where they are currently maintained.
Original materials © Copyright 1994 - 1999 A.J.Flavell & Glasgow University