Again, disclaimer: I'm not an expert.
I would dare to say no, it's not an improvement over Bitcoin. With bitcoin's poW, mining pools don't have any special powers that solo miners don't also have. So while the hashing power is "pooled", consensus isn't really "delegated" beyond the same protocol rules that affect everyone.
If you are talking about supernodes solving the nothing at stake problem because they can check signatures in real time, that is definitely taking a step toward trusted authorities. In contrast, proof of work acts as it's own timestamp.
How does a mining pool not have power that solo miners do not? A mining pool can produce more blocks than a single miner...
I'm absolutely fascinated by your insistence that bitcoin is trustless network. There is no such thing a trustless transaction. You can create an environment that incentivizes a desired outcome. But at the end of the day all of these systems rely on people. Bitcoin is less of mathematical experiment as much as sociological one. We know that public key cryptography works, but what we do not know is whether or not individuals can trust no one while simultaneously trusting everyone. But this is the nature of economics and think it has been empirically proven that I do not have to trust an individual I transact with but rather I can trust society as a whole.
I'm not an expert either but I feel that your analysis of bitcoin and related systems lacks understanding of organizational theory. All organization whether virtual or not face the same fundamental constraints and the delegation of power is necessary for scalability.