Linked by Thom Holwerda on Sun 30th Jan 2011 22:10 UTC, submitted by fsmag
Internet & Networking "I've always been a great fan of the law of unintended consequences. It takes you places. Unexpected places. Sometimes good, sometimes bad but never a dull moment. The recent kerfuffle over Pirate Bay is too well known to require detailed recounting here. What is really interesting though is where it might just eventually take us in terms of internet freedom. This article describes the one fallout of the legal judgements against Pirate Bay and how its response may unintentionally help to protect and promote internet freedoms."
Thread beginning with comment 460243
To read all comments associated with this story, please click here.
Improbable at best
by Ripples on Mon 31st Jan 2011 02:09 UTC
Ripples
Member since:
2005-07-06

I am not sure how P2P-DNS will work, if it can work at all. The key thing that makes DNS work now is that no one entity can "steal" a domain name from another without some sort of lawsuit or motion, and all unique DNS names are registered and brought back to a single entity. How would a decentralized system make sure that I am going to the right place when I lookup google.com? Perhaps some sort of reputation system. More importantly however, who do I talk to and how do I setup a new domain name?

Reply Score: 1

RE: Improbable at best
by atsureki on Mon 31st Jan 2011 02:35 in reply to "Improbable at best"
atsureki Member since:
2006-03-12

I didn't get all the way through the article - it's poorly edited and drowning in metaphors - but it's talking about turning DNS into torrent, so instead of a root nameserver, we must have root trackers swarming semi-official hosts files. But can you propagate upstream changes in a swam? I'm almost certain you can't in any existing Torrent application.

Torrent is the one p2p protocol least suited to providing alternate DNS services, because it doesn't register itself with metaservers. You need to find and download torrent files, which point to a single tracker, which means no search capabilities, except through the existing Web, which solves exactly no vulnerabilities to authorities of any kind. This just looks like one of those articles where the author rubs two buzzwords together to make smoke.

Reply Parent Score: 2

RE[2]: Improbable at best
by zlynx on Mon 31st Jan 2011 04:08 in reply to "RE: Improbable at best"
zlynx Member since:
2005-07-20

The linked article talked about turning DNS into a torrent, but I think they've gotten confused somewhere.

The last idea I read about was using a DHT, which some versions of BitTorrent also use, but the similarity pretty much ended there.

Reply Parent Score: 2

RE[2]: Improbable at best
by grfgguvf on Mon 31st Jan 2011 14:32 in reply to "RE: Improbable at best"
grfgguvf Member since:
2006-09-25

I totally agree (Free Software Magazine always tends to underperform my expectations and it's a total shame to link to it IMHO), however, torrent does provide mechanisms for decentralized search/discovery: distributed hash-tables.

Reply Parent Score: 2

RE: Improbable at best
by zlynx on Mon 31st Jan 2011 04:06 in reply to "Improbable at best"
zlynx Member since:
2005-07-20

The idea that I read about had designed it so anyone could claim any domain name. There were no controls on it. There's no central authority to grant and revoke permissions on names.

However, the idea included the cryptographic cost idea from BitCoin so that domain names "cost" CPU time to maintain. This would prevent people from squatting on names because it would quickly become too expensive in terms of CPU time to maintain more than 10-20 names for a single person. A larger company could buy more CPU and maintain more names.

It seemed pretty smart to me. I don't have the link to the article I read so I can't provide it.

Reply Parent Score: 2

RE[2]: Improbable at best
by Ripples on Mon 31st Jan 2011 04:49 in reply to "RE: Improbable at best"
Ripples Member since:
2005-07-06

That doesn't seem like a good solution either though. Why should you have to spend valuable CPU time on keeping your domain name when you should be able to use it for something more important to yourself or your organization.

Reply Parent Score: 2

RE[2]: Improbable at best
by WereCatf on Mon 31st Jan 2011 07:22 in reply to "RE: Improbable at best"
WereCatf Member since:
2006-02-15

However, the idea included the cryptographic cost idea from BitCoin so that domain names "cost" CPU time to maintain. This would prevent people from squatting on names because it would quickly become too expensive in terms of CPU time to maintain more than 10-20 names for a single person. A larger company could buy more CPU and maintain more names.

That would be a downright HORRIBLE scheme: it couldn't simply be implemented in any reliable way and could be bypassed with a simple crack. Oh yes, you can bet your ass that people would just modify the binary so that it bypasses the code eating on CPU.

Oh, not to mention that it would be unfair: you can't ensure that the same code runs the same on all different CPUs, thus eating more CPU-time on some architechtures and less on others! That would the competition for domains unfair and geared towards certain chips.

No, I do NOT support this.

Reply Parent Score: 2