We aren't forgetting it. We just know the database sucks at being a database. * Transaction volumes are abysmal. Which means I have to build some kind of layer

We aren't forgetting it. We just know the database sucks at being a database. * Transaction volumes are abysmal. Which means I have to build some kind of layer 2 transactional layer over it. (i.e. I have to build a second database to work around the shortcomings of the blockchain database) * The mechanism of BFT that a blockchain uses is expensive. Proof of work, Proof of Stake, the others. This expense has to escalate in order to retain Fault Tolerance. It is among the least efficient way to achieve fault tolerance that you can imagine. It also exerts a push toward centralization. * The storage capacity of the database is also terrible See my point above about needing to build a Layer 2 storage mechanism to be at all viable. All of this is playing out in practice right now in web3 space. Merkle trees make a fantastic version of a public audit log. But blockchain consensus mechanisms all seem to suck. You'd frankly probably be better off going with Raft or something instead. But that would be too centralized for the adherents so they don't. Meanwhile for the exact same reasons that Raft leads to centralized nodes blockchains lead to mining pools at the cost of massive inefficiency. [EDIT: mis-worded]