>To clarify, in my example of the veteran car mechanic assessing someone's proficiency, there's no set-up required, they can verify it 'in the moment'.
>
>No special or unusual tools or environmental conditions are needed to conduct verification.
Well at least some environmental conditions are required: if you ask someone to change an oil filter, there has to be a car in the environment to do it on. To get any value of out of it, you also have to be able to oversee the process - it's no use if you're instructing someone over the phone, for example.
More generally, I think the point of the original post, and the original comment, are that verifying that something was communicated successfully requires out-of-band actions. Physically demonstrating something, or watching someone demonstrate something, are both out-of-band: neither work over all communication channels, like user manuals, or phone calls.
Even in-person communication is often restricted to make verification difficult: imagine if you're in a water-cooler meeting with another developer, and you mention that you think they should take a different approach to a certain problem. Are you going to follow them back to their desk to verify that they really choose to do so? Probably not: it's both incredibly rude, and a bad use of your own time. But there is nothing in the water-cooler conversation that you can really do to check that they'll take your advice.