I love this new world of cryptocurrency research where adversaries have godlike technical powers, the ability to corrupt honest parties & the ability to immediately and perfectly coordinate thousands of malicious nodes...but cannot, in any way, selectively censor network traffic. 
Look, if I could assume that networks are really reliable and uncensorable, I could build some amazing sounding protocols too.
It's not that these protocols aren't cool or useful - they are - but when they get advertised as being "decentralized" and "scaling to billions of users" they are being fundamentally dishonest about how real world adversaries operate.
Also about the general practical limitations of broadcasting packets to everyone in a user set over the internet.
The reason blockchains work (through PoW) is that the difficulty of producing a block is proportional to the computational power of the network. Someone can't just trick you into accepting a really weak chain - they have to have a lot of computational power.
When you remove the need for resources to produce a block, you open yourself up to all manner of selective censorship and cheap sybil attacks.
Unless you can assume all messages get delivered.
Unless you can assume all messages get delivered.
Which is a bad assumption, unless you are willing to have super nodes in your network and have the majority of users trust those nodes.
Partially decentralized banks are a thing, I guess.
Partially decentralized banks are a thing, I guess.
And because it's probably worth stating: I have no problems with papers being written under these assumptions, exploring the domain and understanding limits is important.
I have a huge issue with people selling their solution as a viable cryptocurrency.
I have a huge issue with people selling their solution as a viable cryptocurrency.
