Looks like they mean the cutting torch isn't working.
If I count correctly he got 9 cybertronium with the torch and 5 without it. Or I might be reading this wrong. I don't feel like doing the math right now but my gut feeling says this is statistically significant even with the tiny sample he did (60 samples on each population).
It's not even close to being a good sample to determine the actual difference (see for example the cube spading, 10000 samples and still not within 1% on some drops) but it's enough to show there is a difference. I don't see why the difference couldn't be 50%, just as stated.