Originally Posted by jayeff

I have to declare a vested interest, since I make money from switching peoples' pages/sites to "pure" CSS, but I really do believe your points are off-target.

Point #1 in particular isn't logical, because the HTML should be more simple with the styles and dimensions out of the way and with the ability to reduce table:tr:td simply to div.

It is possible, even likely, that someone setting out with CSS is going to still be thinking in non-CSS terms and therefore they will emulate the structural "vision" with which they are familiar. Then, not only would a client be asked to get used to something new, but also something nearly as complex. But if you strive to use semantically correct code and inheritance, you usually can reduce your instruction to the client to "paste the whole lot in here"...

Dealing with cross-browser incompatibilities also takes some experience. But once you learn where, for example, margins will cause problems but padding will not, the number of hacks you are forced to use declines dramatically. Most of my conversions use 2-4 hacks, some none, and I cannot recall any which have needed more than 6. So there certainly isn't any risk of losing the "advertised" advantage of having less code to deliver.

There are a couple of further points. All tables are not born equal. Coded well tables can load quickly: but most are not coded well and therefore they load slowly. When people switch to table-less (especially if they enforce on themselves the discipline of strict doctypes), not only do they gain the inherent SEO and speed advantages (because browsers handle non-table code more efficiently), but they will be forced to learn how to code properly and often be doing so for the first time. That by itself makes a big difference in many cases.

The trap I see people fall into regularly, is that of not adjusting their perspective into CSS terms and that is what can throw the file-size advantage away. I quite regularly see HTML+CSS pages more than 30% bigger than need be, because their old structure has been replaced more or less directly instead of being re-thought with CSS in mind.

Last, there is the purely pragmatic view that the 'net is never going back to the days of bloated code. People can persist with plain HTML, so long as browsers can recognize it. But in the end they will have to learn the "new" way (new in 1997/8 that is) and in the meantime, not only are they losing out on the many features and benefits which CSS has to offer, but they are denying themselves skills which their competitors are learning.