There are tradeoffs all around. I think of Bitcoin as running one of the more pure experiments out there by keeping a more narrow focus. Other projects subscribe to the "fat protocol" thesis and pack more features into the base layer. The trouble happens when they need to continually tweak these extra features -- it often breaks applications and businesses that built on top of that feature.

It's important to run all the experiments.

I agree most people want something that's usable. They also want something that's stable ... and predictable ... and that no one can steal. All that's really keeping someone from stealing my STEEM is my password -- no 2FA needed.

My biggest complaint with fat protocols though is they get too clever, have too many bloated options, and ultimately could be fragile. I think Steem is a good example -- tons of useless features and complexity -- just look at the "wallet" page as a newcomer and your head will spin. Why are there four options for currency: Steem, SP, SBD, Savings?

So while you can scale up transaction speed, you might be creating a technical mess that can't scale up in terms of features and usability.

My thought is if we choose to sacrifice decentralization, it should be at a higher layer than the base protocol. With Bitcoin, you could imagine this in the distant future if new banks use the Lightning Network for transfers and we, the users, choose to keep our bitcoin in the bank -- not so much on the main chain. The banks would form a centralizing function, perhaps by tracking our bitcoin balance in a centralized database.

So we would have the choice as a user of centralizing or not. Perhaps there could be a variety of banks with different emphases (security vs interest rates vs fractional reserve). But the base protocol doesn't require or dictate any of this.