The 2019 study published on medium only lists machine learning buzzwords in their methods section. The 2022 twitter link doesn’t explain the methodology at all; on the other hand, Messari is usually legit (but paywalled).
Regardless, programmable transactions are in fact a feature of DeFi, they are launched by real people anyway and it doesn’t automatically imply “wash trading”; that’s just bad faith speculation.
It does imply, however, that citing figures like "$1.72 billion of on-chain transaction volume in the last 24h" in the context of web3's health is misleading or false.
False? How? It’s verifiable on-chain data. Misleading? Well, it’s pretty common sense that many users will automate their transactions; that’s a big feature of an open, programmable financial infrastructure. And at the end of the day, even if we adjust the number down by two-thirds, then $600m volume in the last 24h is still a questionable definition of “dead”.
[0] https://defillama.com/dexs