Thanks Sergio!

On Thu, Apr 24, 2014 at 5:13 PM, Sergio Lerner <sergiolerner@certimix.com> wrote:
For more information you can check my post: http://bitslog.wordpress.com/2014/02/17/5-sec-block-interval/
Also NimbleCoin is a new alt-coin that uses 5-sec block intervals, allows 100 tps and .... it's based on BitcoinJ (NimbleCoinJ now). So not only it is possible, but it was coded by Mike itself.

Fascinating! I think that's the first time I heard of an alt coin entirely based on bitcoinj as its core implementation. Looking forward to your release.

My understanding is that dogecoin suffers somewhat from having so many headers. SPV clients have to download them all in sequence so the more blocks you have, the more data they must download and thus the slower they sync. Sync times for SPV wallets today are fast enough that unless you spend six months in the jungle with your phone switched off, you probably won't notice. With 5 second block times unless there's some other solution you'd have much worse UX.

BTW, Pieter experimented with relaying blocks as hash lists (actually merkleblocks) and I believe he found that it could often fail and be slower if the mempools were not quite synced. At any rate, it was apparently more complicated than it looked. That may be a side effect of trying to reuse the Bloom filtering code however.
 
Another solution to achieve <5 secs block intervals is this: http://bitslog.wordpress.com/2014/03/20/mincen-a-new-protocol-to-achieve-instant-payments/

MinCen looks like a rather interesting idea. I will read the paper.