I think that to an extent, it is a good idea for long term protection of the protocol as it wouldn’t run into a problem such as not having enough bots running to secure the network, but on the same token it does remove some of the decentralization of the protocol as well.
Another thing to consider is the transfer of risk. Being a liquidator incurs its own risk, such as the collateral losing value faster than the incentive amount, alpha loss due to paying transaction fees, and maintaining your own sufficient collateralization so that your liquidator doesn’t get liquidated. One thing to keep in mind with this is these inherent risks would be shifted from liquidators to the protocol itself in certain ways depending on how it would be designed and who the participants in this pool would be (whether it would be SNX stakers as a whole or those who choose to participate within its own pool).
In the former example pool above, it would almost defeat the purpose of a liquidator in general, which is to burn bad synthetix network debt for the incentive of bonus collateral, effectively removing that debt from the network as a whole. In essence, the debt doesn’t disappear really so much as it is transferred to someone else who has the ability to pay it off and thereby remove it from the network. If the protocol itself was the liquidator, the debt doesn’t really leave the network and I think it would occur further risk for all SNX holders.
Now if the pool was part of the protocol, but it had its own individual reward/liquidity pool that supplied the funds to liquidate, only a certain subset of the network would appear to incur the added risk, which I think would be a little more feasable of any option. Would be happy to discuss further, discord is also sicarius