The thing is that, as you noted yourself, the analogy doesn't hold, for people are not AWS servers waiting for charge or failovers. If they chose to work 20h/wk, it's presumably to spend their time for other pursuits, eg. family, and they often won't be able to or won't agree to work more when their employer need it, perhaps unless it is planned well in advance (eg. Black Friday period).
> Basically, the company is renting two brains for the price of one.
Or, as I questioned, you might got two partly focused and perhaps less skilled people thinking about your problem, instead of someone more skilled totally focusing on it. As was brought in another comment, if you got two freelancers, then yeah, you could pick two brains, but if you have two family guys, not so much.
I also don't think it's necessarily true that two part-time people would be less skilled or less focussed on work. It's the very point of the original article that working fewer hours can increase the quality and focus of the work done during that shorter period.
I've worked with many talented and exceptional "family guys" (and family women) who were skilled and totally focussed during working hours. I haven't noticed a correlation between people's priorities outside working hours and ability to do good work.
To me, it depends on how you cut the hours, and to what extent. I'm all for working shorter days, eg. 6-8h, 4-5 days a week; however, the more you get close to half-time or below, the more difficult it becomes to stay fully focused and as motivated as a (more) full-time employee. This is admittedly less of a concern if the person uses their free-time on related projects (eg. writing open-source stuff).
It would seem to me that 2 people working 20h/wk offers some significant advantages over one person working 40h/wk.
Firstly, it allows for fail-over; when one person is unable to work (due to holiday or illness, for example) the other might be able to handle 40h/wk for a short period to cover. If one person leaves the company suddenly, it allows for continuity of knowledge while a third person is hired and trained.
Secondly, it allows for variable output; two people can "burst" up to 80h/wk in response to workload, whereas one person would find that very stressful.
Thirdly, multiple people bring multiple points of view, increasing the chances of serendipitous ideas and solutions, and increasing the potential benefit of using people's special skills or experience. Basically, the company is renting two brains for the price of one.
There are some downsides too: The management overhead is higher (equipment, legal, payroll, etc...) and things that were simple when done by a single person are now more complex since there needs to be coordination between multiple people.
To use an incredibly bad analogy: It's like moving from a monolithic "big iron" server to a pair or farm of smaller ones. There are many advantages, but it means it's now a distributed system with all the problems that can bring.
The problems aren't insoluble and don't limit 20h/wk employees to low-importance tasks: Many open source and volunteer projects function perfectly well with many, small, variable inputs from their contributors, many of which will be high-importance.