Thanks wmf - I do recall reading that. If you notice the stats he has, if the block header verification process was done in parallel, there's a theoretical speed up from that "original best case 12s" block chain copy time to a potential 2-3s (possibly), and an actual time of 5s.
But ultimately, and correct me if I'm wrong, the very nature of the "copy all blocks to everyone" nature of the blockchain will make it close to impossible to reduce it down to the order of millisecond transactions?
Ethereum is a scam. Its author received millions of USD in funding, and after a year the only things he has delivered are blog posts that are mostly just PoS propaganda. PoS is a flawed concept, but its proponents dismiss all criticism by saying "we added all these patches and unnecessary complexity and fixed it! Look, no one has hacked it yet!".
As someone using the ethereum alpha software regularly, I have to disagree- They have produced a lot of great code so far, and it seems likely now that ethereum will be live and in operation in early March (though admittedly it'll still have some shortcomings at that point)
As for PoS vs PoW, yes PoS gives you less objectivity on "truth" in block determination- If you define "flawed" as anything less objective than "PoW" then it would certainly meet your definition. The question is how many people feel the way you do on this matter, and what impact this will have on the success of ethereum as a product.
There were several functional alpha clients even before the crowdsale. PoS vs PoW is somewhat beside the point anyway, as the codebase is still using SHA3 PoW: http://github.com/ethereum
