I'm wondering, do I know any IT professional among my friends who has not, at least once, fantasized about doing exactly what Julian Assange and the Wikileaks team did?

I know I sure as hell have had that fantasy, for at least 15 years now. Much as I'm dismayed by the damage the Wikileaks revalations will have to the ability of the US government to perform its legitimate functions over the next two years, I know I thought of diving right into that niche. And I'm pretty sure you have too. And... you, dear reader, or yours truly, dear reader, might have been able to do it without the collateral damage that Assange is causing.

Of all the LVM2 command line utilities, only one lets you specify what you want to do by UUID. So say a motherboard dies on you on a Fedora system and you need the data. Well, if you move the drives to another Fedora machine, you now have two volume groups named VolGroup00. The kernel will use the second one for everything. But, when you look at the situation, you'll find both have been fully activated. So when you do "vgrename $UUID $NewName" to clear this up, it will complain that the logical volumes in $UUID are active. But you can't dl "lvchange -an " on them because their name refers to the other volume group. That's where Knoppix comes in.

Still, all that needs to happen is for lvchange to accept UUIDs as arguments. Or, the Fedora install process can start naming the volume groups something like "/dev/$FQDN-VolGroup00".

Observing how my mother uses the Internet (AOL keywords and Google; the DNS is dead to her) leads me to be thankful for all the political wrangling and litigation that has gone on about domain names. All that energy from lawyers and politicians directed to where it can do the least harm. No, really. All that energy could have gone into trying to force the taking away of IP address space, or to cut people off by way of routing tables, but didn't. And that's important, because the Internet's topology is more vulnerable than most folks notice, especially as far as transcontinental fiber goes. So much energy went into the contents of one mapping of names to IP addresses, leaving a much freer environment for others to go to, e.g. Google.

That said, if ever a protocol asked to be converted into a peer to peer variant, DNS is it, and I am amazed it isn't already done. A p2p DNS system would be way more robust than relying on your ISP's small list of DNS servers. It would also help make it difficult to employ DNS based censorship, which many countries do have in place.

Moreover, a p2p DNS system, with proper onion routing and freenet-style magic crypto fairy dust would not allow your ISP to keep logs of whether you wanted to go to www.osamashomobortionpotandcommiejizzporium.com.

Best of all, with such a system there would be a ludicrously low barrier to entry by a competing "dot in dotcom." Want to run your own .com registrar? Suck down the zone files you care about, assemble your own database, sign it with your own GPG key, and cast it on the p2p-DNS. If people care for your domain policies more than they care about whoever is running them at the moment, they can switch. But you don't even have to exist in any tangible physical form. People aren't too likely to make the switch unless Network Solutions does something really obnoxious, but the ease at which a competing registrar can move in will keep be a good incentive for NSI not to do it.

I'm imagining in my mind something like Freenet as far as crypto goes, but which uses just an LDAP schema for the domain name queries and the like, so it would not be a practical protocol for disseminating goatse or child porn or Gor novels or anything else people don't want to serve out of their hard drives. It would also not hog bandwidth. (Call it BitTrickle?) And by using LDAP it could easily be readapted for more uses.