In response to a statement from Elliot Noss, Vint goes on [paraphrasing]: 

"The notion that DNS is a business was surely not in its original design.  Today it is.  As a business, it's very different from the engineering design.  But engineering still has to work.  I have different views because I'm worrying about it always working.  The architecture was planned to be very hierarchical." 

How many new TLDs would it take to break the net, Elliot asks? 

There have been reports, Vint says.  Kleinsin jumps in [paraphrasing]:  "There's a third dimension.  People say, "Gee, it would be great to experiment."  We don't want to turn the DNS into an experimental laboratory.  We can't justify that on a stability basis." 

Vint to Elliot:  all is not lost.  This is a discussion, not a decision. 

Elliot:  You're hesitating here, but you're not telling me how many will break the net.  Tell me. 

Kleinsin:  my personal opinion is that experimental allocations of domain names (ways of doing allocations) are frightening to me.  Even one is frightening to me.  There have been several pieces of advice to the community that indicate that in terms of preserving DNS stability it is safe to add chunks of 10 tlds without any effect on stability.  But, says Kleinsin, there are other people who worry about upper limits -- those are very large compared to existing numbers of domains.  We could do anything in the same order of magnitude of current domains without an effect on stability.  My distinction, says Kleinsin, is that I'd like to see more justification for doing this than Gee, there is an opportunity here. What we're being given isn't enough justification -- we need broad guarantees to make certain that an effect of failure is zero impact.  [so complete safety has to be assured] The more flighty an application seems in terms of its approach, the more I'd like to have to comfort me on the risk side of the equation. 

Vint again:  there's quite a difference between processing at top level and processing at second-third levels.  Reason:  amount of coordination and management of root zone.  Scaling is much harder there. But we could double the numbers of TLDs without much effect on stability. 

Mohamed Diop:  I'm not seeing destabilization of the DNS through expansion.  This is the only issue ICANN has full responsibility for.  No one will give us the chance to recuse ourselves.  We can't declare that we can't do this job.  Someone else will show up to take on this domain business.  A core value of ICANN is to enable competition and choice.

Ivan Campos:  We're getting applications to do things that the DNS was not designed to do.

Roberto Gaetano:  I joined this organization 8 years ago in the hope of creating new TLDs to compete with .com.  We've missed that train.  Com continues to be dominant.  Do we NEED new TLDs?  Of course not.  Nor did we need cars when we had horses.  But I haven't seen enough information to see market benefits of new TLDs.  Now, with the changed situation at the end of 2004, someone who is serious about launching new TLDs should come with reliable market figures.  I am still in favor of the introduction of new TLDs, but I'm more cautious now. One thing we need to make sure is that the introduction of new TLDs should do no harm -- I'm with Kleinsin on this.  How would new TLDs do harm?  Well, with excuse of innovation, they could bend the DNS to do systems that it wasn't designed for.  And they could put ICANN in a situation where not having well-defined rules for new TLD applications puts us at real risk.  Then we'll be forced to delegate thousands that are more or less of the same type, or we'll be at risk of litigation.  So that worries me.