I think the definition for human latency [in the post] is not quite right. Human latency is not the time it takes to change communication channels, it is the delay in a business process that results when human interaction or intervention is required. Thus, human latency includes things like notifying someone of an issue, finding the right person to talk to, bringing all of the necessary people into a communications or collaboration session, actually discussing the issue and making decisions, ...

UC can help with several of these human latency items because it makes notifying people easier, makes finding the right person or persons for resolving an issue easier, makes setting up the communications medium easier, enables instant access to multiple mediums (e.g. voice and web collaboration), and so forth. When I say easier, "easier" could also mean faster, depending upon the processes, people, objectives, and tools a particular enterprise has.

Excellent points, and it prompted me to look back at Zeus's earlier piece on presence, where I found this, about how presence-enablement might be incorporated into a factory machine:

When the machine fails, its presence status would turn to off, which in turn could trigger an event or a series of events. The event could be a search through a skills database to find all qualified individuals to fix the machine. Based on the presence status of the qualified individuals, automated messages could be sent to only the individuals that show "available". Additionally, the failed machine's "unavailable" status could trigger an event that raises the production capacity of other machines automatically. The automating of these processes based on presence information can remove almost all of the human latency and radically alter the way we work.

So at least in some scenarios, "human latency" isn't necessarily about human ergonomics in the communications process, it's about cutting humans out of the process altogether, until the final stage when they actually need to be talking to each other to solve the problem.