[uf-discuss] Re: Precise Expansion Patterns
bhawkeslewis at googlemail.com
Sun Dec 16 10:01:34 PST 2007
Martin McEvoy wrote:
> The crux of what I am trying to explain is that at the moment empty
> anchor text links mean nothing as far as SEO is concerned, bots will
> either ignore or simply erase them from there index.
> If we as a respected community say that empty anchor text DO mean
> something, then bots and other applications that crawl the web will have
> to take this into account in order to correctly represent their indexes.
> Black Hat SEO's will undoubtedly exploit this to their own means
> rel="nofollow" is a classic example of where a microformat has been
> exploited by SEO's to do something its not meant for and thus may be
> regarded as an Anti design pattern.
> I do not wish (although the intentions are good) to be responsible for
> opening the floodgates on any such behavior, and as a community we have
> a responsibility to steer well clear of hacks, and half hearted
> solutions that may end up causing more damage than good.
Let me go through Martin's argument, as I understand it, to explain why
I don't find it persuasive. My comments are in parentheses.
1. Search engines currently "ignore" TITLE on non-linking A. (Does
anyone has any clear evidence to confirm this? Does that evidence hold
for all major engines, or only for Google? I can't find anything solid.)
2. If microformats made use of TITLE on non-linking A for
machine-readable data, search engines would want to use it. (This seems
3. Black hat SEOs would then keyword-stuff TITLE on non-linking A.
(Given the above premises, this seems plausible).
4. Search engines would not be able to distinguish keyword-stuffed TITLE
attributes on non-linking A from microformat data. (rel="nofollow" is a
standalone pattern and it would be difficult to tell correct from
incorrect use without human inspection. But these data-hiding patterns
are intended for use within super-patterns like hAudio that tell parsers
what to expect. Distinguishing data-hiding patterns of this sort from
keyword stuffing sounds like a trivial programming task, so I think
search engines surely would be able to distinguish between them.)
5. This problem only occurs with non-linking A. (But by its very nature,
markup that hides content from most users, most of the time — and that
includes /all/ patterns that hide data in TITLE attributes — is
susceptible to keyword stuffing. Indeed, I would bet that each and every
one of these patterns is already being abused by black-hat SEOs,
effectually or not.)
NB: This email isn't intended as a general endorsement of TITLE on
non-linking A. I'm deeply sceptical about misusing the TITLE attribute
for human-unfriendly data, especially on anything other than an empty
SPAN. I'm just saying I don't buy Martin's SEO-based argument against
non-linking A in particular.
More information about the microformats-discuss