Note to self to come back to this one as there are a few things that need explanation, the difference between data centres and transit and peering being one.
Yeah I know I messed up the definitions there as peering is to reach a target network, not for transit to other networks.
Honestly, its hard to keep the different terms straight in your head when you aren't living, breathing networking.
But the crux of the matter is, Zen aim to have direct links to each exchange rather than routing everything via London and Manchester, as they are aware the current system doesn't work very well and wastes bandwidth.
What I'm not sure about is if they plan to have transit carriers more local to exchanges, effectively spreading their core network across the country, rather than having to take all traffic to one location and transit from there. The fact they implied that Youtube, Netflix, caches would relieve bandwidth on the links back to Zen suggests some sort of peering to local data centres where these caches would reside, which would seem to blur the definition between transit and peering. Or, perhaps more likely, I completely misunderstood what they meant.
Ideally you'd want a system where any two Zen customers on the same exchange, never need to hit the Zen core network at all. This is how Digital Region seemed to work, although its hard to be sure as all transit was in Sheffield so its not like the latency would be high either way.
Either way, they seem to be much more open about the fact their current configuration is less than ideal and the direction they are planning to go to fix it.