In block-auction ePBS, it is easy enough for a validator to set the conditions under which they decide to switch to local building, e.g., they could compare the current value of their locally-built payload with the best bid they have received from builders, and decide to build locally if the difference isn’t too high. Under slot-auction ePBS, the validator no longer has certainty over how much their locally-built block will be worth at the time of proposing it, seconds later after they committed to the bid of a builder. At the time of commitment, a locally-built block may have been worth 0.8 ETH while the best bid from builder was 1 ETH. At the time of proposing, perhaps the validator is not able to obtain more than 1.5 ETH, while the builder’s block is suddenly worth 10 ETH.
In the example above, it appears as if the validator “lost” 9 ETH, as the builder obtained 10 ETH after bidding 1 ETH initially. But as slot auctions are early, one does not possess this information ahead of time, and thus the benchmark of the local builder must be revisited. While it is clear that a local builder does not have as much information in slot auctions as they do in block auctions, they can still devise a reasonable heuristic to decide whether they should take on the duty of proposing a block themselves or delegating this duty to a builder instead.