That's an interesting article. I'm aware of this sort of radiative cooling, but usually convection and conduction, along with evaporative cooling will be much, much more important than radiative cooling. I'd like to see some kind of report or article showing the cooling potential in W/m^2. I remember someone made a metamaterial optimized for emitting heat in the wavelength of the atmospheric window, but the value was quite low.

You could build a few test configurations.

One as stated, insulated fridge, open on top.

One, insulared fridge, open on top, under an open air shelter, no walls, just a roof, where the roof is a significant distance away from the top of the fridge, so airflow is not impacted, but there's no view or the sky. Possibly variations on this with the roof being ir transparent or not.

Another test configuration could have the fridge closed, but with forced air ventilation.

He did try a basic control with his first version: “i got ice consistently over several nights, even though there was no frost on the ground, and pots of water sitting next to the box cooker did not freeze.”