MikeNoLastName - 11:10 am on Mar 7, 2013 (gmt 0)
It is so sad we have to even discuss the possibility that G may or may not like bolding.
In so far as the "almighty user experience" is concerned, I believe a lot of <B>ing often has MAJOR utility. Consider a club newsletter with multiple different subjects covered in separate paragraphs. Bolding keywords allows the lazy skim reader (in this time-challenged world) to skip over the subjects they are not interested in and zip right to the topics they are.
Likewise, if you have a <UL> list of, I dunno say, movie theaters in the metro area and what they are showing that evening. <TABLE>ing everything adds SO much unnecessary code overhead to the page (increasing download time) and wastes a lot of screen space requiring more user scrolling. Bolding the theater names which are immediately followed by the movie list and times in an <LI>, can certainly make it easier to quickly distinguish the theater name from the list of movies there. Or alternately perhaps bolding the latest new movie releases people are more likely to be looking for vs. the older movies. This would naturally result in the same movie being bolded many times on the same page if it is playing at many different theaters.
Another example: simple newspaper classified ads listings. Who would NOT naturally bold the name of the item being listed at the beginning of the listing and maybe the phone number?
Pick up any magazine or newspaper you have laying around the office or house and look at the myriad of things they bold just to get attention WITHOUT them considering how it might help them on a search engine and then think how G would PROBABLY penalize this 'natural' bolding if it were on a website.
There are a MILLION examples exactly like this, which I'm sure G has not considered, as actually being a GOOD or natural thing for the user, and instead try to generalize with an ALGORITHM which can only fail. I'm sure the plex whippersnappers don't consider that a lot of the ever-increasing, aging, computerized population find bolding easier to scan through and read on a screen. Or execs in a hurry who just want the factual 'keyword' talking points.
So it boils down to, people are NOT algorithms and G will have to realize that they can not ever generalize even the majority of the examples. Unless they have every page objectively reviewed by multiple humans, specializing in the particular field of the website, they will never have the ideal SE or anything close to it. Since such human review is not infinitely scalable for finite dollars, as they wish it could be (so they can just sit back and watch their results database and profits swell infinitely), they will inevitably fail.