Exactly. How many people could this support? My guess it would be a small proportion of an urban population.

> no active refreshing/polling

Maybe not twitter specifically, but the paper mentions a centralised message board like twitter, so we can assume some sort of interactive service would be suggested.

My main point is this that for such a specific bearer (bandwidth, channel plan, duty cycle, range etc) it is important to model the capabilities. I’m not sure how well this would support a crisis (notably free of central management) when there is the potential for large numbers of nodes.