You can't divide by zero.
But you can divide by arbitrarily small numbers tending down to zero. So, if you look at card-cost-per-coin-produced on a scale running from silver to copper, it gets better and better as you approach copper. Taking the limit, you can then say that copper is better than silver on this basis, even though it can't be evaluated by itself to have a value on this scale.
Of course, this normally not the right way to assess the strength of treasures, and copper should not in general be considered to be better than silver. I'm aware that was the joke, but I'm just pointing it out in case any fairly new players take it seriously.
That said, there are situations where this is a reasonable way to assess treasure strength, and where the cheaper standard treasures are therefore the better ones. If you will only take one more turn, and you have more 100% (or very close to) reliability overdraw than you have buys that you will use to buy treasures, then, in fact, you do want to evaluate treasure in this way. Adding four silvers to a hugely overdrawing deck is generally better than adding two golds, if the game will end soon. You're unlikely to have got into that situation without having missed an option to build more intelligently earlier, but it might happen. Just the other day I was playing IRL with my brother on a board with Ranger and Wharf for draw, but Royal Carriage was the only splitter (Cartographer and Menagerie the only other +action for a pseudo-village when replayed) and there was absolutely no trashing, so this was a very dicey kind of engine situation that would fall apart quickly when greening. He ended up saying "I would have won right now if I'd bought just one copper with one of my spare buys". Sometimes, particularly for megaturn similar very abruptly greening engines with overdraw, the total economy in your deck is much more important than the density of that economy. Interestingly, that was a game where coppersmith could have been absolutely epic. RIP coppersmith.