On the surface, this looked like a really cool experiment. The possibility for a (semi)private network layered on top of the public internet is something that I've been thinking about for some time, but have never tried in any large way to do it. I'm fascinated by the possibilities for a network structured something like the pgp web of trust, or a model for opportunisitic end-to-end encryption such as the freeswan project is working toward.
And that's where I think this offering (and make no mistake, it will be an offering for a commercial sercive once the beta phase is over) falls down: there is a single point of failure, whether the failure be one of availability, or of privacy. As far as I can tell, the encrypted tunnels all terminate at a single router. This does simplify things; one must only decide whether or not to trust the service provider. One downside is that the provider is exactly that -- a provider -- and if this service (or something similar) ever became popular, there is no doubt in my mind that the provider would be required to provide wiretap capabilities to various organisations if requested to.
Their service is based on a PPP connection which is tunneled using VTun. Cool. Simple, easy, and portable. All good things. This will easily provide sniffing protection for the members of this VPN. What it will not provide is security for the members, and may even provide the illusion of security, which is far worse than no security at all.
Users of this service must be clear on what they are signing up for. They will get the capability to route encrypted traffic traffic originating from their machine or LAN to another subscriber of this service. This does not ensure the privacy of said traffic. Either of the 3 places that the data being transmitted exists in unencrypted form (the two endpoints, and the router in the middle) could be compromised. This problem does not exist for this product alone, it exists for any service like this, whether the network be virtually private, physically private, and no matter what protocol is used (eg. PPTP, IPSec).
I have not explored the (imho) more interesting part of this story, most notably the social implicatoins implied by having many private groups of peole forming on the internet, nor have I touched on the possible services that could be implemented on top of a scheme where ther is some trust involved. It would take me hours to write my thoughts on that, and I don't really want to write an essay here, and I'm sure you don't want to read one. :)
(sorry about the rambling, looslely organized post, but I don't have time to tidy it up)
(for the record, I designed something like this for my company recently, but based on IPSec instead)