That assumes fungibility and substitution. The reason we have 1% of people making the food is because that is the specialisation we can't eliminate by mechanisation and automated processes.

The "Spreading the unemployment" argument is a bit like saying we don't need all those violin players in the orchestra, just ship a few unemployed people in and give those violin players Thursday and Friday off.

Sure, abstractly. But to continue the analogy - there is presently a glut of violin players and other musicians, and everyone currently making up the orchestras have no time to do anything else.

And yeah, the "best" orchestra is going to face an insatiable demand for their specific time, and likely enjoys performing as much as humanly possible. But these conditions are exceptional, rather than the situation faced by basically everybody else.

Why not 4% with robots handling the rest?

I was saying what the numbers would change to at the same level of technological progression, but with policy changing to make the work more evenly distributed.

Obviously the long-term goal is ever-more automation. But current economic policy is based around creating full full-time employment, which is fundamentally at-odds with ever-less needing to be done due to automation. Hence the proliferation of anti-productive "bullshit jobs" where people ultimately aren't ultimately helping to produce anything (eg the entire medical billing system), but still need to cling on and go through the motions so they can eat.