Tensions in Practice: Easy counts and useful contributions¶
An illustrative bug-report reward
A software project wants useful defect reports. Paying for every submitted report makes the reward simple to calculate, but a reporter can increase the count without finding more useful defects. Paying only for reports judged useful changes what earns the reward, while adding review work and disputed judgments. In both arrangements the reporter can anticipate the rule and choose what to submit next.
Make rewards clear
Give contributors a predictable, observable basis for a reward.
Reward useful work
Make the payoff track the defect-finding contribution the project actually needs.
Why these aims pull against each other
A cheap observable count can diverge from useful contribution. A closer assessment takes effort, can be contested, and can itself become a new target for gaming.
Choose an arrangement to see what changes and what remains difficult.
Arrows show the stated work, authority, or access paths. Position, length, and color do not measure time, risk, cost, or performance.
What this choice protects
What it costs
When it fits
Compare the arrangements
Pay per report
Pay on the count of submitted reports, while useful defect discovery is assessed separately.
- What it protects
- The payment rule is easy to communicate and tally, with little value assessment before payment.
- What it costs
- Additional trivial or duplicate submissions can earn rewards while consuming project attention without comparable value.
- When it fits
- Plausible only where report counts are already a good enough indicator of useful contribution and the gap is not cheaply exploitable.
Illustration note: The source supplies the per-report gaming example; the separated value branch is an editorial explanation. No behavior change or gaming rate is predicted.
Assess contribution
Review reports against declared usefulness criteria before calculating the reward. Submitting other reports remains possible, but does not automatically earn payment.
- What it protects
- The payoff can move closer to the contribution the project wants, making raw volume alone less rewarding.
- What it costs
- Review consumes attention and delays payment. Judgments can be wrong or inconsistent; contributors may optimize the review criteria instead.
- When it fits
- Fits when useful contribution can be assessed credibly at a cost justified by the proxy problem.
Illustration note: The reviewed-payoff arrangement is an editorial response, not a claim that a new criterion perfectly measures value or eliminates gaming.
What this illustration does—and does not—establish
Incentive: Target Behaviour versus Measurable Proxy supplies the gap between rewarded proxy and intended behavior. The alternatives make the payoff feedback explicit; the reviewed-value response and its residual costs are editorial synthesis.
- Both rules are hypothetical. No reward amount, productivity gain, or optimal bounty policy is supplied.
- The active feedback through a decider distinguishes this case from a thin specification that misses intent even without strategic response.
- The second arrangement changes consequences attached to available actions. It does not prohibit reports or guarantee aligned motives.
Source entries
Incentive
Incentive: Target Behaviour versus Measurable Proxy supplies the proxy problem, including payment per bug report. The example keeps the payoff connected to the reporter’s next choice rather than treating it as passive measurement.
T1
T1 — Target Behaviour versus Measurable Proxy. The payoff must attach to an *observable* proxy, but the designer wants the *target* behaviour, and the two are never identical. The tension is measurement: the population optimizes whatever the payoff is pinned to, not what the designer meant. The failure mode is Goodhart drift — taxing measured emissions drives firms to unmeasured pathways, paying per bug reported floods the queue with trivia. Diagnostic: ask how far the proxy can be pushed while the true target stagnates or worsens; if that gap is wide and cheap to exploit, the scheme will be gamed before it works.
A consequence feeds back into choice
Their characteristic failure modes — perverse incentives, gaming, crowding-out of intrinsic motivation, Goodhart drift — are symptoms of the same structural commitment: a payoff is attached to a proxy, and the population optimises for the proxy.