Does anyone have a link to this "Google Paper"? I've never read it.

Gracias

de nada

I would hazard a guess that based on what we see from reverse engineering rankings the Google may have added H tags to the fancy hits list, which would be easy to do without amending the data structure.

Another thing which I would like to highlight is this passage:



Viewing all the optimum page size, keyword density theories in light of this statement would seem to indicate that they are all wet.

I'm not so sure about that Mel. The paper states that "Google keeps track of some visual presentation details such as font size of words. Words in a larger or bolder font are weighted higher than other words." That seems to say that bold words are moved up the 7-value size scale. Nothing is specifically stated about H tags, and I'd guess that the same is done with words in H tags. There were no spare bits at the time the paper was written, and they could easily have left it that way - ar they could have added another byte of bits for things like that.

Certainly, tapering the weights to nothing does away with the need for an ideal number of words in the text, and densities. It just means getting enough searchterms in to fill the count, and the nearer the top, the better (position).

All that needs to be done is to add the H tagto the fancy hits list in the parser, along with anchor text, page titles etc. No need to change the encoding just the scheme which determines if a hit is fancy or not.

They may have already included the H tag in the fancy hits. They stated that fancy hits "include" certain things (URL, title, anchor text, and meta tag), which may not have been a full list. And they had 4 bits for the type of fancy hit - that's 16 possible varieties of them. If they saw H tags as being "fancy", they could include it there, but it couldn't have been included as a plain hit without some significant rewriting, because there were no spare bits.

Darren's one-off test was interesting, and it makes me think that H tags may simply be given a boost in text size - like it appears to be with bold text.

and quite possibly italics

yep