One Algorithm Against both Vertical and Horizontal Attacks

General
1

Here, I propose a proof-of-stake algorithm against both vertical and horizontal attacks on the block chain:

[ol][li]Vertical attacks consist in privately chaining a block subchain (as in “selfish mining”).[/li]

[li]Horizontal attacks consist in chaining blocks at the same height of multiple chains (as if there was “nothing at stake”).[/li][/ol]

The algorithm is quite simple:

[ol][li]Each block has a list of all addresses holding the stake that allowed chaining it. For privacy, we can use additional hashes of those addresses, or even stealth addresses. The only purpose is uniquely identifying each stake address, not publishing it.[/li]
[li]A node must discard any block of which the stake addresses intersect with those of the previous block in the same chain. This prevents vertical attacks with the same or intersecting stakes. It also creates an implicit rule by which the same or intersecting stakes cannot chain two successive blocks.[/li]
[li]A node must discard any two blocks at the same height of different chains with an intersection between their stake addresses. This prevents horizontal attacks with the same or intersecting stakes.[/li][/ol]

2

Correct me if I am wrong, but isn’t horizontal attack protection already in place via duplicate stake detection? My understanding is that there is a protocol rule which requires a node to drop blocks that contain previously seen coinstakes. A discussion in another thread talked about the value of turning off duplicate-stake detection, and it seems this is a bad idea based the [factual] cost to destabilizing the network by allowing DoS (via flooding the network with blocks from the same kernel) vs. the [theoretical] advantage of being able to ensure your stake gets minted sooner.

Reference:
PPC Coin Whitepaper page 4 “Block Signatures and Duplicate Stake Protocol”

Peer Coin Talk: What stops me from POS minting several different chains at once?

3

It seems you are correct: the horizontal protection is already in place. Additionally, there is no need for listing all addresses as the stake for each new block is from a single transaction.