Ethereum allows for an extremely low block time by keeping track of and rewarding miners of orphan blocks (uncles in Ethereum) thus including their work in calculating the longest chain.
Is there any reason to not implement something similar in Monero if it allows for faster block times and deals with the problem of stale blocks?
Pretty much anything could be done.
I'd been thinking about a block tree, where you'd have geographically [1] close miners mining at a low difficulty and low block time, and making their own quick chain. Every, say, 10 blocks, they'd hash their blocks (or 10 blocks in the recent past, to guard against reorgs) and a higher level set of miners would have a slower blockchain embed these hashes. Any number of levels. This would allow (A) faster confirmations on a local level without much more orphans due to the geographical (2) proximity, and (B) the ability to scale the blockchain to very distant regions. Like, you know, the actual moon.
This might or might not be similar to what Ethereum is doing, but it's certainly fun to think about.
[1] Doesn't work well with I2P
[2] Well, closeness with a metric based on ping, rather than distances
The "Uncle mining" currently tested on Masari, a Monero code-fork seems to correspond to your description. The block time is planned to be reduced to 60s. This test will allow to see if there are any reasons not to implement this concept.
You can see the current state of the implementation in this pull request: https://github.com/masari-project/masari/pull/71.
