Because being consistent is very deliberately not supposed to require being identical. So long as the visual identity is maintained, and I think it is, then small changes are welcomed.
If the mark needs to be altered to work in a particular format, I think that could show a lack of foresight on the part of the designers. I'm playing devil's advocate though.
I think a very simple HTML + black shield border + black 5, could provide a useful additional mark; which might prove less obtrusive.
--
"It's very common to have structural changes in letters at different sizes in order to maintain readability, yet no one complains about consistency. Why?"
I don't know of any common examples of logos which make similar structural changes - but I'd be very happy to be proven wrong.
I'm on a really poor connection, so I'll do research later, but the difference between display and body variants in fonts comes to mind immediately. Knowing the two distinct use cases will highlight very different properties of the font's design, the letterforms are varied. The differences can be fairly large, however it is always clear that the two fonts are from the same family.
I see your point .. setting a typeface involves decisions re. kerning and many fonts use 'hinting' to ensure they can be read well at varying sizes, and some display faces are altered to suit headline text or specific use cases - but I think most logos have aims which are more specific than a typeface.
If people are free to alter the colours of a logo to suit a design, and the rules for doing so aren't defining clearly - before long a logo isn't able to fulfil it's purpose, because the unifying elements of the design become unclear.
That's not to say that a logo has to be the same everywhere - a successful identity system can break the rules, but if it does so, it needs to be very consistent in the way those rules are broken. I think agencies like Wolf Olins provide a good example of this kind of thinking.
I don't think the the adaptation of the HTML5 logo for the smaller favicon format is a feature - it seems like an afterthought.
(1) Do not correct the 5 and cause people to misidentify the logo at small sizes since it looks like a 6 then
(2) Correct the 5 at the cost of an almost imperceptible change in small instantiations of the logo
(3) Redesign the whole thing to use a font that is distinctive at all sizes
Wolf Olins is a particularly good example of the fact that brand is far larger than logo. I'm not even a particular fan of this new HTML5 brand, I feel like it overemphasizes and romanticizes a tool, but I won't get caught arguing that it's not comprehensive, or strongly suggested by all of the proposed instantiations on the website, or to weak to survive the likely numerous changes it'll suffer should it be widely adopted.
For example - this blog (http://almaer.com/blog/) is listed as an example of the logo in action on the launch page.
I think this page is a good example of my point; at this size, the logo suffers from the '5-becoming-6' problem - it doesn't actually communicate its message successfully at the most basic level.
--
I've had a think about this in a bit more detail - and the more I think about it, the less impressed I am.
I think the whole concept should have been based around producing a mark that can identify whether a product supports HTML5 technologies.
I don't think the logo should have been produced to place on a site (or teeshirt) to show allegiance to the cause - I think it should have been created along the lines of the DVD, USB, HD or Blueray industry support logos.
An identity that helps promote HTML5 as a consumer technology - and shows a consumer at a glance, whether the product they are about to download or buy is capable enough to adhere to the upcoming spec.
But maybe that's what this is? Still, I'm a bit doubtful about whether it achieves these aims.
As a marketer, I agree.