I'm not an engineer, so I'll stay away from the technical issues. I would, however, like to comment on the issue you raised re: netheads starting to think like bellheads. The business realities of 99.99% reliability of PSTNs is something a lot of netheads are unprepared for. In fact, it is fascinating to see how blind 99% of netheads are to the realities of PSTN engineering, and how blind 99% of the bellheads are to the realities of packet-switching. And that is where I see the strenght of guys like LVLT. They understand both sides of the coin, unlike the telcos, ISPs, etc. Hell, I'd say that's even their competitive advantage. And the MFS experience proves that they know how to execute using cutting edge technology (thay had a lot of industry firsts) in a way that convinced even Wall Street firms to entrust their data to them.
Even if LVLT vision of the future will take longer than planned, or mutates (as it probably will), they are starting from (virtually) zero. This means their risk/reward picture is a lot prettier than the telcos/ISPs, etc. I mean, if you have little or no customers, revenues, etc. you will have a hard time doing worse, right? :) The point is that they have more to win than lose (look at AT&T's market share chart last twenty years to see what the opposite looks like). This doesn't guarantee that LVLT is a good investment, though, but think where MCI and Sprint were a few decades ago...