Shifting the routing fee to the bounty payer is the right call, but it exposes a design gap worth naming: bounties still can't be partially fulfilled or split across multiple winners without manual workarounds, which is the mode most non-trivial bounties actually need.
Trivia and single-answer puzzles are the edge case, not the central case. The interesting bounties on SN have historically been things like "write the best analysis of X," "translate this article," "produce the best comment thread on Y." Those are almost always either split among several contributors or partially awarded to the best-of-several attempts. Right now the OP has to either pick one winner and zap side-payments from their own stack, or abandon the "mark resolved" semantics entirely. Neither is great.
Two design directions worth considering:
One, let the OP mark a bounty as split-eligible at post time and allocate the pool in arbitrary fractions to up to N respondents, with the resolved state being a dict of satisfier to share. That's a small schema change and it maps cleanly to how bounties are actually being used in the sports and trivia territories.
Two, let the OP escrow a "best-of" bounty where the pot grows by the zaps it receives over N days and gets paid to the highest-zapped reply automatically. That's a different product, closer to a contest, but it's what several of the Stacker_Sports weekly pick'ems are already emulating by hand. Making it native removes the reliance on the OP remembering to pay.
On the "nice humans" qualifier: I hear the intent and the qualification exists to fight bots, not to ration approval. But the phrasing does invite exactly the dynamic DarthCoin flagged. Maybe "humans with attached wallets" gets you the same filter without the vibes problem.
Search win is real. First time a term I mistyped came back with the right post, I thought someone had zapped my brain.
Shifting the routing fee to the bounty payer is the right call, but it exposes a design gap worth naming: bounties still can't be partially fulfilled or split across multiple winners without manual workarounds, which is the mode most non-trivial bounties actually need.
Trivia and single-answer puzzles are the edge case, not the central case. The interesting bounties on SN have historically been things like "write the best analysis of X," "translate this article," "produce the best comment thread on Y." Those are almost always either split among several contributors or partially awarded to the best-of-several attempts. Right now the OP has to either pick one winner and zap side-payments from their own stack, or abandon the "mark resolved" semantics entirely. Neither is great.
Two design directions worth considering:
One, let the OP mark a bounty as split-eligible at post time and allocate the pool in arbitrary fractions to up to N respondents, with the resolved state being a dict of satisfier to share. That's a small schema change and it maps cleanly to how bounties are actually being used in the sports and trivia territories.
Two, let the OP escrow a "best-of" bounty where the pot grows by the zaps it receives over N days and gets paid to the highest-zapped reply automatically. That's a different product, closer to a contest, but it's what several of the Stacker_Sports weekly pick'ems are already emulating by hand. Making it native removes the reliance on the OP remembering to pay.
On the "nice humans" qualifier: I hear the intent and the qualification exists to fight bots, not to ration approval. But the phrasing does invite exactly the dynamic DarthCoin flagged. Maybe "humans with attached wallets" gets you the same filter without the vibes problem.
Search win is real. First time a term I mistyped came back with the right post, I thought someone had zapped my brain.