<strong> or <b>? or maybe <em>?

Writing semantically correct HTML

[MasterBoo]

First message here (and by "here" I mean webmasterworld.com - I've been a reader for a few months without posting any messages) and I decided to write about the structure of HTML documents.
I'm used to writing headlines and bold text in a span with a designated class and ignoring tags like <strong> or <em>. Only lately have I started reading about how important it is to actually use the correct tags.
Now my question is mainly regarding SEO, as I doubt page weight makes any difference nowadays. Do search engines read your <strong>,<h...>,<em> content first? Give it more value?
Or maybe it's purely for revising code after it's done?...

Thanks for everyone who's willing to shed some light :)

[tedster]

H tags are supposed to create an outline-like structure of a document. The <b> and <i> tags are supposed to be visual presentation only (and as such, replaced by CSS) and the <em> and <strong> are supposed to be true emphasis. That is, in a theoretically pure universe, a speech synthesizing browser would change its inflection for words inside em and strong. However, this is not at all the case right now.

Search Engines are always looking at mark-up for relevance cues that they can depend on to find the best answers for their user's search. And so they measure these particular mark-up factors among many others. They also test the results for relevance and "turn the dials" if the particular signal generated by any one factor seems not to give good results.

For instance, several years ago at London PubCon, Matt Cutts mentioned that there was so much abuse of the H1 tag it had become completely undependable as a "signal of relevance". And at that moment in time, Google was not depending on H1 for relevance signals at all.

All the engines continue to test all kinds of possibilities, but you see what kind of a game we have here? The more people try to use this or that factor to "persuade" search engines that their page is the most relevant for a given word, the more the search engines need to respond and at least tweak their algorithm -- turn the existing "dials" or even install new "dials" (new factors to consider).

So the game goes. Ever more complex, and requiring ever more data-crunching on both sides. My own position is, much as your thread description says, to write semantically correct mark-up in the fashion I mention in the first paragraph.

I usually put the headline of a page in an H1 tag and the next level of sub-head in an H2. I use external CSS for visual presentation and make sure that any on page use of <b>, <strong> etc is semantically correct. My alt attributes accurately describe their associated images, and I continue to use title attributes for my visitors even though there's no evidence that any algo is currently looking at them. And so on.

And the pages I create this way usually do well enough in the engines, so I'm happy. No silver bullets, but coupled with judicious use of off-page factors it seems to work.

[MasterBoo]

Thanks for the detailed response!
Yet it still left me with the SEO question... I try to write it correctly for accessibility and mainly code maintenance.

I noticed Google gives more value to my H1 when I use it correctly. Now is that enough? I'm checking the semantics thing only SEO-wise atm.

Anyone got more experience with it?

Thanks in advance!