I'm having trouble imagining a failure mode that would cause total collapse under load rather than just reduced per-user speeds. Networks have to deal with spikes in demand pretty frequently anyway, usually during the evenings when lots of people are streaming video. A sudden increase in overall demand throughout the day doesn't seem like it should be that big of a deal; and so far it seems like it isn't.
How about the mode where, video streaming makes normal browsing impossible for some users? Their packets just don't survive the public net in the blizzard of video packets? That would cause apparent collapse for some at least.
Not sure how 'net neutrality' would factor into that - a free-for-all would mean it might happen more, but a pay-as-you-go would eliminate a whole class of users. A hard problem.
When it gets full (like it is some places), then the only strategy is to drop packets. If they are dropped systematically (by some rule) then some demographic loses some part of their internet entirely. That's also well understood.
Its a product of non-neutral nets, where by definition they do something by a rule (instead of randomly for instance).
And even randomly dropping (neutrality?) stresses certain subsets of network traffic more than other e.g. video can recover from dropped packets; TCP traffic not so much. Again stymying certain classes of activity more than others.
It seems almost funny now, but in the late 90s there was a lot of talk about how the internet could fail under its own load. The arguments never made that much sense to me, but they attracted attention...
