In this article, Georgetown Law professor Adam J. Levitin states that since bitcoin outputs are unique, each output could be subject to a different tax liability depending on its cost basis. He argues that this decreases fungibility,
I think Levitin's argument is false. In my mind, the fungibility property of money is only relevant in exchange between parties. I think it only matters to the receiving party. To the receiving party, it is immaterial how long the giving party has owned it, nor how much capital gain the giver needs to pay.
While I have no specific education in the matter, it would seem to me that money's fungibility property is useful only in that it trivializes the receiver's assessment of the money's quality aspects. If a received bitcoin is indistinguishable from any other received bitcoin, the receiver need only think about overall quantity rather than quality of each bitcoin.
The giver needs to evaluate the quality of each bitcoin, but that determination is not coupled in time to the transaction. Indeed, this decision process is something that could plausibly be automated according a policy decision by the giving party - perhaps via some mechanism such as the ZGL wallet.