Okay, here's my two cents on the heading/section issue...
The element <h1> can be used in HTML4 multiple times. Therefore, it is not by default the title of a document. The most natural thing to assume is that <title> is the title of the document, regardless of how people might abuse it. At best, you could say that when there is only one <h1> element, it might be the title. Even that is not a certainty.
There are already <meta> values for much of the stuff that people abusively put in the <title> element. For example, "author" metadata is pretty much a standard already:
| <meta name="author" content="Space Dog">
Therefore, why not just standardize it and deprecate the use of such values in <title> while giving <title> more specific semantics? Something like this:
| <meta name="author" content="Ian Hickson"> | <meta name="sitename" content="WHATWG"> | <meta name="publisher" content="Hixie Industries">
The <link> element can be used for the stuff that requires a URL or hyperlink of some kind.
As for sections, in my opinion there are two types of sections. The first is what we have now, an implied section system that is difficult to define but essentially goes from the beginning of one heading to the beginning of the next. The second type of section is one defined by markup such as <section>. This type should be the only type that can be styled and affect rendering.
Generally, implied sections should begin at the beginning of a <h#> element and end at either the end of the document or at the beginning of
another <h#> element. Allowing exceptions makes describing implied sections much more complicated. If developers need flexibility with regard to this, we can use language that says developers *should* start and end sections in such a manner rather than *must*.
Importance (given by the number of the <h#> element) determines styling, and it implies structure, but the level of importance should not create missing sections within the document outline. If the importance level of a heading implies a missing parent section, the outline tree should first be constructed as if the implied nodes existed, then the children of missing sections should be collapsed into the position of their parents.
That said, this is how I would process the sample markup:
<body>
<p>...</p> <unnamed section>
<h1>A</h1> 1 A (importance level 1)
<h2>B</h2> 1.1 B (importance level 2)
<h3>C</h3> 1.1.1 C (importance level 3)
<h2>D</h2> 1.2 D (importance level 2)
<h3>E</h3> 1.2.1 E (importance level 3)
<p>...</p> E
<ol> E
<li> E
<h3>F</h3> 1.2.2 F (importance level 3)
</li> F
<li> F
<h3>G</h3> 1.2.3 G (importance level 3)
</li> G
<li> G
<h3>H</h3> 1.2.4 H (importance level 3)
</li> H
</ol> H
<p>...</p> H
<h4>I</h4> 1.2.4.1 I (importance level 4)
<h2>J</h2> 1.3 J (importance level 2)
<div> J
<p>...</p> J
<h2>K</h2> 1.4 K (importance level 2)
<p>...</p> K
</div> K
<p>...</p> K
<h3>L</h3> 1.4.1 L (importance level 3)
<h2>M</h2> 1.5 M (importance level 2)
<h4>N</h4> 1.5.1 N (importance level 4)
<h3>O</h3> 1.5.2 O (importance level 3)
<h1>P</h1> 2 P (importance level 1)
<h1>Q</h1> 3 Q (importance level 1)
<h2>R</h2> 3.1 R (importance level 2)
</body>
This would be the outline:
Document
|
+-- <unnamed section>
|
+-- 1 A
| |
| +-- 1.1 B
| | |
| | +-- 1.1.1 C
| |
| +-- 1.2 D
| | |
| | +-- 1.2.1 E
| | |
| | +-- 1.2.2 F
| | |
| | +-- 1.2.3 G
| | |
| | +-- 1.2.4 H
| | |
| | +-- 1.2.4.1 I
| |
| +-- 1.3 J
| |
| +-- 1.4 K
| | |
| | +-- 1.4.1 L
| |
| +-- 1.5 M
| |
| +-- 1.5.1 N
| |
| +-- 1.5.2 O
|
+-- 2 P
|
+-- 3 Q
|
+-- 3.1 R