Didn’t read the article, but this seems infinitely appropriate: https://xkcd.com/1205/
Something to not forget about though is that manual tasks sometimes have a bigger chance of errors, especially when they become muscle memory and conditions are different, you are in the wrong directory, the wrong server, etc.
Also often manual tasks need to be documented, sometimes having a simple, easy to understand script is shorter and quicker to understand than writing documentation.
And one more thing is that one has to be realistic about time and should include everything that really is needed (logging in somewhere, etc.).
Of course the same is true for automating things. Cost of maintenance, does it require anything else, etc.
In extreme cases automation can end up taking up more manual labor than the absence of (faulty) automation.
Something related can often be seen in consumer software that “tries to be smart”, doing something automatically for you, but then requires you to be more careful or makes you have an extra step, etc.
This was in fact one of my original inspirations for the calculation model I am proposing.
I am even referring to it directly in another article: https://medium.com/p/identifying-the-best-use-cases-for-automation-e3a6ef1cd0f3
We’re definitely on the same line here!