(Disclaimer: I'm one of the authors of this I-D.)

Because by the time you went through all the steps you're describing here, you lost so much time already that the CDN may as well just serve you their content from the other side of the planet. CDNs are meant to *improve* latency. If two or more HTTP requests are needed to reach the goal, you really just completely miss it.

Also, HTTP caching wouldn't work very well if content gets served from "random" hostnames (because how else would you get the web browser to connect to the right CDN node?).

Remember that this is an EDNS0 option, to be implemented only by parties that need it. It's not intended to become a standard DNS feature.

And now you're blind?

(Aren't those disks still way too thin and transparent to be used for this?)

27W. Wow. That's only about four times as much as my whole laptop (with Intel graphics). Definitely very low-power! :-P

I sure hope they have a mobile version of that chip...

That has very little to do with this technology and more with a mistake in a IP-to-geolocation database somewhere, most likely.

About two months ago I used Blind Search for a week to see how Bing's actually doing. There were two classes of results that I found:

A) Two columns show fairly similar results (Bing and Google) and the third shows somewhat worse ones (Yahoo).
B) One column with good results (Google), the others totally wrong or at least significantly worse.

Although I just tried a query from the second category again and they now all show the best result first (instead of somewhere on the second page). I guess it's time for another week of Blind Search for me.

Akamai probably has *many* more CDN nodes in the world than the number of gDNS and OpenDNS datacenters together, so just being distributed is not enough.

Yep, it's home-made ... but why do you think it'd be Python?

I think this is/was a common thing with CDMA (or whatever the non-GSM protocol is called exactly) telephones. They get the time from the network and don't/didn't bother storing it anywhere locally.

Another difference is that the stereo wouldn't be even 1% of the price of the car instead of sometimes more than 10%.

Although I'm tempted to say calling it "brute force" would be more accurate. :-)

And maybe that's why almost every e-ticket already states clearly which airplane will be used for every flight?

> Am I missing something? If not, these guys are tards and making a big deal out of nothing.
Even more because this publicity probably burnt all their remaining bridges. :-)

And you seriously think that Google (or any other bigger company) is still using 2-CPU servers?

> Article mentions Android is based on Debian

I have mine for only a week so I may have missed sometihng, but I definitely haven't noticed anything "Debianish" on the phone. Non-GNU libc, very odd userland (it's not GNU nor busybox), etc...

Also, don't forget to set the kernel flag that enables the leap-second code. It's quite likely that this was the cause, and the kernel won't know by itself that it should insert one. See adjtimex(2) for more information.

It's definitely possible that some program couldn't cope with time jumping back. Actually, it may be just as easy to just test this by manually stepping back in time.