The simplest way to put the problem is that he didn’t make those calculations for an LN routing network. He created some other network that works quite differently and calculated things based on that.
In other words, it’s basically an impressive looking straw man argument. He creates something that can’t possibly work, names it LN and then proves it doesn’t work. That, however, has nothing to do with the actual Lightning Network.
So, while the math itself might be correct, it doesn’t actually model LN, so it’s a rather pointless string of calculations.