Here's a very interesting article about Ethereum Casper, probably the highest profile attempt to supersede Proof of Work using collateral based consensus. The end product is still a form of Proof of Stake though.
https://blog.ethereum.org/2015/08/01/introducing-casper-friendly-ghost/Some common criticisms that I have gathered from t'internet (Thanks to Herman Schoenfeld)
No longer P2P - Casper fundamentally changes the network topology from a Peer-2-Peer (P2P) network where all peers participate in the protocol equally to a "distributed Client-Server" network where one class of peers (validators) have special privileges over another class of peers (nodes) within the protocol itself. This is a fundamental difference from PoW and impacts security and redundancy. First of all, it's not "nuke proof" like the Internet was supposed to be (whether real or legislative nukes). Similarly, like Client-Server the Casper-driven Ethereum network suffers from a single point of failure problem (it's 250 points in casper). Suppose that during a staking period that all the validators happen to be geo-located within the same jursidiction (e.g. the EU). Suppose that this jurisdiction then passes laws to outright (or effectively) ban Ethereum due to it's challenge to banking/economic status quo. Those validators will shutdown overnight and the network will be permanently down. This is a realistic scenario. Note with PoW, this can't happen since new miners can easily spring up outside that jurisdiction and the network can resume where it left off.
Violates protocol neutrality - In a P2P network, all peers participate equally in terms of the protocol. For example, in Bitcoin distinctions between nodes, miners, lightweight clients are independent of the protocol. At any one time any node/miner/lightweight peer can freely convert to any other without impacting any other peer. Not so with Casper. Validators are effectively bound to their status, ordinary nodes cannot independently elevate their status, and worst of all, ordinary peers are forced to accept the status of validator peers. There might be a tonne of attack vectors lurking inside here.
Introduces stakeholder risk-coupling - In PoW, a miner who mints a bad block does not affect the profitability of a miner who mints good blocks. Not so in Casper. The bad actions of one validator can influence the profitability of a good validator. Casper adds economic disincentives to mitigate these scenarios and stop bad validators from minting bad blocks, but this approach assumes that bad validators are rational economic actors. What if a validator is compromised by a hacker or shutdown by a Government? What if a validator is infected with malware? In these real-world scenarios, validators may not act rationally. As a result, other validators (and the network itself) will not be protected.
Stakeholders batton-passing problem - Another coupling issue with Casper is that new validators must be approved by a suitable set of old validators. This is fundamentally different to PoW where miners can begin/cease mining independently of each other (0-coupling). Have the implications of this been researched. What attack vectors can exploit this?
BitTorrent analogy - Just think, if all the entertainment industry had to do was capture 250 peers to take down the entire BitTorrent network permanently, do you think BitTorrent would be around today? If Ethereum really does challenge the status quo, how hard will it be to take down the 250 validators? UPDATE A sustained DDOS attack on the validators could realistically achieve this.
If you want to read the full discussion between Schoenfeld and Ethereum staff, here's the link.
https://www.reddit.com/r/ethereum/comments/4euiut/fundamental_problems_with_casper/