With PoW, the block time is around 17 seconds. What is the target block time for Casper's PoS?
Somewhere between 2s and 7.5s, depending on what tradeoffs we choose between maximizing speed and minimizing consensus overhead.
One of the important philosophical points imo is that very high consensus overhead is actually not that bad provided that (i) it's constant and not per-transaction, and (ii) light clients do not have to bear it; for example, even if the Casper consensus process itself imposes a load of ~5 TPS (this is what we would realistically see at a 7.5s block time and 256 validators in the current protocol), if the greater resistance of the protocol to economic centralization pressures means that we can safely survive a load of, say, 100 TPS instead of 20, then it's worth it, even if it "seems inefficient" during the period where there's only 0.2 TPS worth of actual use load on the system.
The proposed blocktime is 2-3 seconds.
At least that's what Vlad Zamfir (Ethereum Researcher) is aiming for. There are more details on the subject in this epicenter bitcoin podcast #105 .
The link doesn't work, can you give some detail about how many TPS can it handle and why the blocktime will be so high, internet bandwith is high these days and also latency, i think you can go around the world in 400 ms easy Jul 31, 2018 at 22:40