[dns-operations] gTLD controlled interruption for fun and profit

Jim Reid jim at rfc1035.com
Sun Apr 30 08:44:24 UTC 2017

> On 29 Apr 2017, at 19:27, Warren Kumari <warren at kumari.net> wrote:
> I personally believe that name collisions have to potential to cause
> real issues,

I thought so too Warren. But so far there’s been no sign of those real issues actually happening. That could well change of course if .home, .corp, .mail, etc ever got delegated. Has anyone documented an actual instance of a name collision event or problem with any of the recently created gTLDs? 

I alo wonder if anyone has found a collision and reported it to ICANN after they'd stumbled across this controlled interruption tripwire.

> but that the way that ICANN chose to "mitigate" them (the
> above) is entirely inadequate / basically a no-op.

It’s almost certainly a no-op. That doesn’t necessarily mean the controlled interruption strategy is inadequate.

I think it’s providing a reasonably appropriate level of defence/assurance against something which has a vanishingly low and possibly zero probability. Controlled interruption is a prudent safety net IMO, albeit one that will probably never meet a genuine collision event.

> The fact that this has been sitting here, unnoticed for ~a year
> demonstrates that -- if "controlled interruption" was actually
> working, this would have been noticed before...

Maybe. However this is flawed logic. Just like the tinfoil hats some of us wear to protect ourselves from killer unicorns. Controlled interruption might not be working (for some definition of working) because nothing is triggering a noticeable collision event.

More information about the dns-operations mailing list