It's quite smart how Apple has marketed their displays relative to the resolution of the human eye. By emphasizing the 'Retina' name, they claim the resolution title, and simultaneously nip any further competition in the bud by ingraining consumers that more pixels aren't going to be visible anyway, so they won't bother paying more for it. (Of course that assumes any other manufacturer can even match this display... someday perhaps :)

Since the dawn of the internet, the only positive innovation we have seen from ISPs is simply increasing bandwidth. All the other real innovation has come from the endpoints. I see no evidence that ISPs are going to provide any positive benefits from allowing them to censor and filter our communications with these endpoints. We wouldn't stand for the US gov't enacting a 'great firewall of china', I don't see why we should allow the ISPs to set up a private version either. Further, we have utilities for things like water, sewer, electricity, etc. because it becomes a mess if we try to run these basic infrastructures as competitive markets. It just doesn't work for these industries. Internet connectivity falls into this category as well, most people simply don't have a competitive ISP market in their area, so arguments that consumers have can vote with their wallet if they don't like corporate policies are moot.

I heartily agree... I have long though CSS should have variables. And math with those variables. I don't know what these language designers are thinking leaving these out. And is inheritance of some sort too much to ask? Unfortunately, these conversion layers you've listed don't address unit conversion: if I want something to align with two blocks, one of X em and another of Y px, then I want X em + Y px, but only the browser will know what that value is. Especially if one of these blocks is based on % instead! So then we wind up with pixelized fixed-size layouts instead, undoing so many of the goals of flowing web page design and font size/resolution independence :(

Hmm, as Ovidiu suggests having your local SMTP server do this stuff for you... I'm pretty sure most individual email clients don't have personalized DNS records with public/private keys. (But I'd love to see S/MIME get mainstream!) I'd be more curious about how to get through authentication mechanisms on the SMTP server, and then let it do the heavy lifting. For my organization, it looks like that's just a matter of piping everything through an SSL connection with a username and password.