On Sun, Oct 2, 2016 at 6:46 PM, Russell O'Connor via bitcoin-dev <bitcoin-dev@lists.linuxfoundation.org> wrote:

But I would argue that in this scenario, the only way it
would become invalid is the equivalent of a double-spend... and therefore it
may be acceptable in relation to this argument.

The values returned by OP_COUNT_ACKS vary in their exact value depending on which block this transaction ends up in. While the proposed use of this operation is somewhat less objectionable (although still objectionable to me), nothing stops users from using OP_EQUALVERIFY and and causing their transaction fluctuate between acceptable and unacceptable, with no party doing anything like a double spend. This is a major problem with the proposal.

Transactions that redeem an output containing (or referencing by means of P2WSH) an OP_COUNT_ACKS are not broadcast by the network. That means that the network cannot be DoS attacked by flooding with a transaction that will not verify due to being too late.

The only parties that can include the redeem transaction are the miners themselves.

Therefore I see no problem that an OP_COUNT_ACKS scriptSig transaction is invalidated after the liveness times expires.

If there is no expiration, then polls can last forever and the system fails to provide DoS protection for block validation since active polls can accumulate forever.