I took my idea to the mailing list and advanced it a little bit but I'm not familiar with posting to the mailing list, so what I'll post below, I posted twice and can't see in the group yet. The first was on the web interface itself, after which I couldn't find a trace of having posted at all and hours later it wasn't showing up (as approved by moderation) so I posted again using thunderbird. If there is any merit to the idea, I'd appreciate your reply on the ML

I think the poison pill approach could be implemented as a soft fork after all, with a cleaner mechanism:

After activation at block height X:

1. **Vulnerable UTXOs cannot be spent directly** - they require a prior announcement

2. **Weak announcement** with no private key needed: "I intend to spend UTXO A with transaction X after block B+144"

3. **Strong announcement** with a commitment proof: References a potentially old, pre-fork commitment and provides proof that this UTXO was included

4. **After 144 blocks**: The UTXO can be spent according to the strongest announcement (oldest commitment wins)

This is a soft fork because:

- We're not "undoing" transactions

- We're adding new rules about *when* certain UTXOs can be spent

- Old nodes still see valid transactions, just with different timing

The key insight is that the "weak announcement" doesn't require private keys - it just declares intent. This preserves the validity of pre-signed transactions (they can still be announced and executed, just with a delay).

Meanwhile, anyone who created commitments before the fork can use "strong announcements" to override potential quantum attackers during the window.

This gives us poison pill protection while maintaining backward compatibility. No transaction reversal needed - just a new spending process for vulnerable UTXOs.

Does this address your hard fork concern?