Idea: Using unix time as sync mechanism for multi-imp projects

I was thinking about multi-imp projects such as lighting and how that brings up the need for precise need for synchronisation.

My idea is too use unix time as part of the request sent to the imp so in pseudocode it’d be “at 123456.78 do xyz” or for a long running command like a fade “starting at 123456.78 reach xyz state by 123456.99”

What issues are there with this as an idea, the only one that spring to mind is you would have to calculate latency but you could do that by polling each device in the group for its latency are using the largest value.