The central claim of the 'Quantized Majorana Conductance' paper is that the signal plateaus at the precise (quantized) value. The idea is that the signal is 'robust', i.e. reaches a value and stays at it. Let's take this apart.

And I mean literally take it apart. We already know two pieces were cut from the raw data, so let's keep cutting. We cut at every major discontinuous jump in the data.

What are these jumps? They are sudden changes in device condition when tiny charges move around the nanowire. The nanowire is also small so this is enough to affect it. You can see that in panel (c) of their Fig 2 the signal really jumps - back down to a value it already passed.

Think of a skipping record. The song advances for a short while and then jumps back, and again, and again.

Now back to the plateau. Let's throw away some of the segments. When stacked like this they look like a loooooooooooong robust feature. But for what we know, the length of it is not longer than the widest of these cuts.

That was from their original Nature paper from 2018. And here is from their 'replacement' paper last week. 'The vertical black dotted lines indicate the candidate plateau-like region'... Hello, old friend.