And, actually, CSS creates content-independant presentation that HTML largely lacks. When you use <b> tags, for example, you're saying, "present this content in bold format." But what if I'm visually impaired and using an audible browser? CSS allows the page designer to say "on the screen, make this bold, on a printer, make it italics, and when spoken, surround the text with these words." That, Ytrew, is content-independant presentation.

Oh, and about blind people - I know quite a lot of them, including an ex-girlfriend. They all use IE to browse webpages, and use their screen reader to hear what's on the screen. Simple webpages is what they prefer, any meaning added to pages using CSS is lost on them. And forms and image maps can drive them to madness.

Not being blind, nor having a browser capable of interpreting CSS's audio, I would not have noticed. And CSS should not be adding meaning - only presentation.

CSS allows your web pages (content) to be much simpler (and smaller), shunting presentation off to another file. This should allow your screen readers to function easier, and, as time goes on, audio browsers will likely come about (I can imagine IBM wanting to extend Firefox for this, and then donating the code back to the community, in exchange for the ability to throw Firefox on all of its products' CDs - that would be worth millions to a company like IBM). CSS is looking forward to that day with current standards.

Don't get me wrong - CSS isn't perfect. But it's a heck of a lot better than pure HTML.

Oh, I very well know what CSS should be used for. I also know what HTML should be used for, and what the intended usage for the Internet was (sharing of (computing) resources).

But only the most naive person would escape the notion that the web is mostly about presentation - whether that's being delivered by CSS or HTML attributes. Most web authors, and an even greater percentage of clients only care how something looks - and not about being good citizens.

As for screen readers, they just do that: read the screen, or rather, text in widgets. It doesn't matter for a screen reader whether CSS is used to color a text red, or a FONT element. It may be tuned to distinguish between bold/normal font, but again, it doesn't matter what caused the boldness. It does pay attention to links though. Pages that consist purely of clickable images (and no ALT text) sound like "LINK LINK LINK LINK" (but then very very fast).

If you want to cater for blind people, you'd be wrong to think that CSS will do the trick. You should focus on the content - not the presentation, and not the separation of presentation and content (of course, that's still a good idea, but mostly for a programmers point of view. It would help if browser actually allowed for user supplied style sheets, but that isn't commonly found on browsers. Even something simple as setting the fonts and fontsizes for P and H1 to H6 seems to be lost after the Mosaic browser).

Don't get me wrong either. I'm not saying CSS is useless. But from the moment the first style-sheet aware browsers were available (1994/1995 - using DSSSL, no CSS available at that time), I predicted that it wouldn't change the web. It would still be about presentation, and little content. And CSS only makes it easier to make webpages that are hard to understand if you don't confirm to the presentation the author intended. And unfortunally, history hasn't proven me wrong.

Actually, what CSS generally adds is faster downloads (CSS markup is generally smaller than HTML markup),

As for Firefox, I recently installed it, noticed that using the middle mouse button no longer opened the link in a new window, didn't find a setting or another easy way to open a link in a new window and ditched Firefox. I know people get all excited about tabs, but windows are under the control of my window manager, while tabs aren't. And since I can do important window operations (delete for instance, or cycling) with a single keystroke without having the move the mouse, using separate windows is much faster than tabs.

RTFM (Read the Firefox manual, keyboard shortcuts, mouse shortcuts):

1. Next tab - ctrl-tab
2. Previous tab - ctrl-shift-tab
3. Close tab - ctrl-w or alternatively ctrl-F4

Also, if you have a special/"multimedia" keyboard, you can assign the extra keys to do fun stuff in Mozilla, like close tabs, etc.. Another huge advantage of Firefox is the Adblock extension, which allows you to selectively disable graphics from loading.

Still, while Firefox and CSS is a really nice and convenient idea, I don't think that dropping tables and NS4 support is a good idea for PM, as the browsers of mobile phones etc. don't work well without tables.

Update: Reworded keyboard shortcuts for closing a tab - it requires a chord of two keys to close a tab, but there are two alternatives.

I'm well aware of those commands, thank you very much.

That's exactly why I prefer to use my window manager. I can cycle between windows using one keystroke instead of one, and delete a windowing using one keystroke instead of four.

And no, I don't find having to use both the mouse and the keyboard to open a link in a new window convenient.

My pages generally dropped in size from about 50k, to 10k. Maybe that's atypical. But it is definitely something in favour of CSS here. And since the CSS is the same on all pages, downloading it once means you don't need to download it again, even if it is an extra connection - a connection can be faster than 5-10K of repeated text, especially on a 28.8k modem...

You start saving significant bytes once you say:

<h1>Heading</h1>
[download]

With some CSS for the heading, compared to:

<p><font size="+2" color="red">Heading</font></p>
[download]

On every single heading. On any modern (HTTP/1.1-compliant) HTTP server and client, the connection will be left open, making cost of the extra request for the cache-check nearly zero.

"There is no shame in being self-taught, only in not trying to learn in the first place." -- Atrus, Myst: The Book of D'ni.

Let's see. Just assume the perlmonks.org is the fastest server in the world, and it replies to requests instantly (as we all notice). I still have a roundtrip time to perlmonks of about 100ms. That's the time that's always lost to do a request. (For the sake of the argument, I leave off the processing time of the proxy I'm using). I also get a throughput of about 75kb/s. That means, in 100ms, you could have added an extra 7.5kb. That would be more than 200 of those headings.

Note that I'm not suggesting to not use CSS. But using CSS to "save bytes" sounds like premature optimization. If you want to save bytes, turn off the nodelets for AMs. Remove the stupid image in the top right hand corner (approx. 60 CSS headers worth of savings). And most of all, turn off the ads. (Well, not for me, I've taken care of that on a much lower level (aka /etc/hosts)).

I do prefer:

    <h1><font size = "3" color = "red">Heading</font></h1>
[download]

over

    <p><span class = "big-red">Heading</span></p>
[download]

with a style sheet that says "big-red" should be displayed in a big red font.

I do not think that many web authors who didn't use <h1>, but a <font> instead will if they go to CSS, suddenly will use logical markup. They'll just replace their font elements with span elements.

If I'd get a penny for every CSS enabled webpage that was hard or impossible to read when CSS is turned off in the browser, I'd be rich.