Economics has been subverted since the 1860s with the introduction of what became known as neoclassical economics (where "neo" is more usefully translated as "not" rather than "new") by Jevons and co. Even though classical economics (called "political economy" by Smith, Ricardo etc) does have some flaws, neoclassical and later schools get things so egregiously wrong that you are forced to ask whether it was deliberate. Modern so-called economists will tell you that land is not a factor of production. That's how wrong they are. The whole thing is bullshit from beginning to end.

The reason Bernstein is confused about what money is is that he is a bullshit not-economist. Money is not complicated. It's a go-between so that people do not need to barter. "Money" as a concept lets them value cows, pigs, chickens, wheat, barley, wine, beer, timber, pottery, copperware, silver jewelry, gold nuggets, etc etc in "dollars" (say) and convert chickens to dollars to beer instead of arguing over how many chickens is a pig worth and all the infinite number of other conversions. [1] It's divisible into smaller amounts, it's easy to carry around, its value is accepted by everyone. Not difficult unless you are a not-economist, in which case you obfuscate the shit out of it.

Which is not say that fiat money doesn't have some problems of its own but that is a problem of implementation, not a problem of "money" as a concept.

[1] A programmer would recognise this as translation between computer languages or file formats or whatever, where you invent an intermediate language/format and write N translators from the thing to the intermediate and back rather than N times N translators from every thing to every other thing.