The author looks at one of the drives, but they bought 2. What size was the other one?

It may just be that there's some distribution of "usable flash blocks" in every flash chip, and they happened to get one which fell at the bottom end of the acceptable range. You'd have to sample quite a few before you could be certain this isn't just an outlier.

It's clearly stated. One is 16 billion bytes, the other is 15.4 billion bytes.

They got two new ones.

But it would be very weird if different drives in the same batch had different capacities. That's a lot of wasted effort.

Not strange at all; first they make the batch, then they test them to see how many of the bits are actually writable, and sort them according to how well they turned out.

This is also how phone screens are made (a huge panel is produced, then lit up; some pixels are stuck, but phone-screen-sized pieces can be cut out of it where there's enough good pixels).

This is also how most CPUs and GPU models are made - test each sub-component, disable the parts that fail testing, then price the result according to how much of it works.

Yep. This is very common and is called binning:

https://en.m.wikipedia.org/wiki/Product_binning

But they were in the same bin.

Right, but all that means is they met the minimum standards for that bin and didn’t meet the minimum standards for the next one.

As far as I can see the author now owns 4 memory sticks, but only looked at the flash on 2 of them.

I bought two of the same model from the same retailer and the other one reported the exact same capacity.

Good Take.

I already though it would be really weird if the flash ship is really smaller.

Which is great. And if they would have followed their own logic (1GB == 1 billion bytes) and labeled it as a 15GB or 15.5GB stick everything would be fine.

I don't think memory mapping a defective stick really need additional memory, regular memory sticks already have to internally remap flash memory because of how flash memory can only delete multi-MB blocks at once.