The Architecture of a Cryptocurrency
github.com
github.com
Proof-of-stake is an interesting idea, but until I see empirical evidence showing that it actually works, I remain skeptical.
Here's a good read on the subject: https://download.wpsoftware.net/bitcoin/pos.pdf
There's no any formal proof of consensus (im)-possibility around. But we have some cryptocurrencies successfully working for years already(despite promises to hack them easily), also there're some simulations around, e.g. ours https://github.com/ConsensusResearch/ForgingSimulation and https://github.com/ConsensusResearch/MultiBranch (corresponding articles https://github.com/ConsensusResearch/articles-papers ). I hope more formal framework will be established someday, and would be happy to help with that(I can contribute with Coq code etc).
On the other hand with PoS I still haven't seen any data whatsoever. I asked a simple question: since there's no cost for someone to simulate a chain, what's stopping peers from doing this? I know there are checkpoints and things like Nxt don't allow deep reorgs, but for new users joining the network it means you need to point them to trusted parties that can give them this information: they can't trust the chain independently.
In the second place a chance of drawing a long chain suffix with cumulative difficulty better than network's is negligible. Short-range attacks are possible, but they are not a huge problem. Please read the forum post with some results got with simulations https://bitcointalk.org/index.php?topic=897488
http://bitcoin.stackexchange.com/questions/936/how-does-a-cl...
http://bitcoin.stackexchange.com/questions/29742/strongest-v...
Pieter Wuille is a core developer afaik: "The actual rule always favors the chains which required the most "work"."
P.S. Joining a network is implemented in the same way usually for both PoW/PoS currencies.
The Bitcoin checkpoints are rather redundant and can be phased out with no loss of security.
Note: A side advantage is that the bank's investments and fees can fund the system. Another is possible acceptance by regulators if it's (a) easy to regulate and (b) tied to their currency to support rather than compete with it.
P.S. How's Tezos going?