[BLOG POST] - Casper Review

[Pheonike]

You're not addressing the incentives of those who develop new full node tech, which is what really matters in the long run in my opinion. They will likely be incentivized to keep their competetive advantages proprietary and launch in competition with incumbents rather than collaborate. As far as I can tell, the result of the design is that ROI on node R&D is maximized, meaning TPS optimization, which is what I'd prefer from a common computation infrastructure.

Also, ignoring that, lets not forget that centralization of block production at worst only exposes people to what is essentially very complex and ridiculously expensive phishing attacks.

They're a theoretical problem, but easily solved since economic dapps will run their own secondary consensus algorithms in the virtual machine to validate in a way that is tailored to their own security preferences (DPOS is probably gonna be a first choice in many situations). With that in mind Ethereum block validation is ultimately the transaction processing and anti-spam layer, not explicitly a trust layer. For that function casper seems to be the best algorithm invented yet.

Could you explain what you mean by "full node tech" and how the pishing attacks would work you mentioned?

With full node tech I mean hardware and software for processing the largest amount of EVM computations at the lowest possible cost, increasing marginal income and reducing marginal cost. The EVM has some exotic properties, such as a 256 bit word size, that means that right now it is ridiculously slow compared to a real computer, but with dedicated hardware it could actually become very fast.... when someone invents that in the future.

Oh and the phishing attack was just a metaphor for double spends, since they're both very similar types of digital fraud.

So you are saying there will custom hardware built for it like asics which will centralise the raw achievable performance to a handful of manufactures.

[liondani]

Wow! ethereum market cap dropped from $160M to $80M after your post! A 50% decline in one day! 
I hope we witness the same increase next time you post something... for bitshares   

[bytemaster]

EVM = Ethereum Virtual Machine

[btswildpig]

You're not addressing the incentives of those who develop new full node tech, which is what really matters in the long run in my opinion. They will likely be incentivized to keep their competetive advantages proprietary and launch in competition with incumbents rather than collaborate. As far as I can tell, the result of the design is that ROI on node R&D is maximized, meaning TPS optimization, which is what I'd prefer from a common computation infrastructure.

Also, ignoring that, lets not forget that centralization of block production at worst only exposes people to what is essentially very complex and ridiculously expensive phishing attacks.

They're a theoretical problem, but easily solved since economic dapps will run their own secondary consensus algorithms in the virtual machine to validate in a way that is tailored to their own security preferences (DPOS is probably gonna be a first choice in many situations). With that in mind Ethereum block validation is ultimately the transaction processing and anti-spam layer, not explicitly a trust layer. For that function casper seems to be the best algorithm invented yet.

Could you explain what you mean by "full node tech" and how the pishing attacks would work you mentioned?

With full node tech I mean hardware and software for processing the largest amount of EVM computations at the lowest possible cost, increasing marginal income and reducing marginal cost. The EVM has some exotic properties, such as a 256 bit word size, that means that right now it is ridiculously slow compared to a real computer, but with dedicated hardware it could actually become very fast.... when someone invents that in the future.

Oh and the phishing attack was just a metaphor for double spends, since they're both very similar types of digital fraud.

What is EVM stands for ?

[btswildpig]

If we believe that 2.0 is all about innovation , then any "flaw" -----If there is any ----- can be solved by innovation .

 It's not like Bitcoin who can not develop a new version ...... It maybe totally different even 6 months later . It already happened with BitShares .

[Rune]

You're not addressing the incentives of those who develop new full node tech, which is what really matters in the long run in my opinion. They will likely be incentivized to keep their competetive advantages proprietary and launch in competition with incumbents rather than collaborate. As far as I can tell, the result of the design is that ROI on node R&D is maximized, meaning TPS optimization, which is what I'd prefer from a common computation infrastructure.

Also, ignoring that, lets not forget that centralization of block production at worst only exposes people to what is essentially very complex and ridiculously expensive phishing attacks.

They're a theoretical problem, but easily solved since economic dapps will run their own secondary consensus algorithms in the virtual machine to validate in a way that is tailored to their own security preferences (DPOS is probably gonna be a first choice in many situations). With that in mind Ethereum block validation is ultimately the transaction processing and anti-spam layer, not explicitly a trust layer. For that function casper seems to be the best algorithm invented yet.

Could you explain what you mean by "full node tech" and how the pishing attacks would work you mentioned?

With full node tech I mean hardware and software for processing the largest amount of EVM computations at the lowest possible cost, increasing marginal income and reducing marginal cost. The EVM has some exotic properties, such as a 256 bit word size, that means that right now it is ridiculously slow compared to a real computer, but with dedicated hardware it could actually become very fast.... when someone invents that in the future.

Oh and the phishing attack was just a metaphor for double spends, since they're both very similar types of digital fraud.

[santaclause102]

You're not addressing the incentives of those who develop new full node tech, which is what really matters in the long run in my opinion. They will likely be incentivized to keep their competetive advantages proprietary and launch in competition with incumbents rather than collaborate. As far as I can tell, the result of the design is that ROI on node R&D is maximized, meaning TPS optimization, which is what I'd prefer from a common computation infrastructure.

Also, ignoring that, lets not forget that centralization of block production at worst only exposes people to what is essentially very complex and ridiculously expensive phishing attacks.

They're a theoretical problem, but easily solved since economic dapps will run their own secondary consensus algorithms in the virtual machine to validate in a way that is tailored to their own security preferences (DPOS is probably gonna be a first choice in many situations). With that in mind Ethereum block validation is ultimately the transaction processing and anti-spam layer, not explicitly a trust layer. For that function casper seems to be the best algorithm invented yet.

Could you explain what you mean by "full node tech" and how the pishing attacks would work you mentioned?

[Rune]

economic dapps will run their own secondary consensus algorithms in the virtual machine to validate in a way that is tailored to their own security preferences (DPOS is probably gonna be a first choice in many situations).

Could you explain this a bit more. Do you mean a DPOS-like consensus amongst the issuers of various types of collateral for pegged assets?

I mean a consensus layer built into the EVM that takes a portion of the income of one or more dapps, and in return validates transactions related to the dapp in "validation transactions" (which I guess could be described as mini-blocks - still need to think of a better name for it). I.e. privatized, modular, customizable consensus validation systems that can be rapidly iterated on.

[CLains]

Interesting post! 

On latency: https://www.reddit.com/r/ethereum/comments/3ff8g5/introducing_casper_the_friendly_ghost/cto6pd5

*typo: caper -> casper,
leaching or leeching?

[julian1]

economic dapps will run their own secondary consensus algorithms in the virtual machine to validate in a way that is tailored to their own security preferences (DPOS is probably gonna be a first choice in many situations).

Could you explain this a bit more. Do you mean a DPOS-like consensus amongst the issuers of various types of collateral for pegged assets?

[Rune]

You're not addressing the incentives of those who develop new full node tech, which is what really matters in the long run in my opinion. They will likely be incentivized to keep their competetive advantages proprietary and launch in competition with incumbents rather than collaborate. As far as I can tell, the result of the design is that ROI on node R&D is maximized, meaning TPS optimization, which is what I'd prefer from a common computation infrastructure.

Also, ignoring that, lets not forget that centralization of block production at worst only exposes people to what is essentially very complex and ridiculously expensive phishing attacks.

They're a theoretical problem, but easily solved since economic dapps will run their own secondary consensus algorithms in the virtual machine to validate in a way that is tailored to their own security preferences (DPOS is probably gonna be a first choice in many situations). With that in mind Ethereum block validation is ultimately the transaction processing and anti-spam layer, not explicitly a trust layer. For that function casper seems to be the best algorithm invented yet.

[liondani]

"Once it is clear who the “deciders” are it becomes safe for everyone else on the network to simply follow the leader and place bets without actually performing any validation."

I think the "deciders" are on this list     
https://etherchain.org/accounts

[Ben Mason]

What worries me is that some of these other projects don't seem to view a systemic centralised outcome as a fatal flaw.....

Nice review

[bytemaster]

http://bytemaster.github.io/2015/08/08/Review-of-Casper-Ethereums-proposed-Proof-of-Stake-Algorithm/