Nothing here has been built or tested. This is where that matters, in order of
how much it would change what you get.
The power figure is three guesses multiplied. Cell efficiency, light lost
in the paper, current lost in the wire — each is an assumption in the
package’s own notes, and the output is their product. It reads like a
measurement and is not one.
No cell was ever tested. The package says plainly that no IV sweep was
performed — that is the standard measurement that tells you what a solar
cell actually produces, by sweeping the current and reading the voltage. Without
it, 12% is a datasheet number for a class of cell, not for these.
The fold life is estimated. The claim is more than a thousand open-and-close
cycles. That was not run. Washi is unusually good at this, which is why it was
chosen, but nobody has folded one a thousand times.
The light pattern was never measured. The caustics — the bright
shapes light makes when it passes through a folded translucent surface, which is
much of the point of the object — were not measured in a lab.
The assembly time is borrowed. The 38 minutes comes from similar builds,
not from building this one.
The front of the card is a generated render. It shows a person’s
face. No such person exists, nobody was photographed, and the card itself prints
a line saying so. You cannot tell that by looking, which is exactly why it is
written here on the page a stranger reads.
The back of the card is a schematic, not a cutting file. At 85.6 mm
across it could not be one. The 1:1 SVG and DXF ship with the package.
None of this makes the design wrong. It makes it untested, which is a
different thing, and the difference is worth a whole section on the page a stranger
reads.