Cost Externality
A burden created by one party and absorbed by another who had no ability to refuse it, which is a separate question from how large the burden is.
Reviewed July 16, 2026Stable
When not to use it
- Where the payer is compensated and consented, which makes the burden a cost of doing business rather than an externality.
- As an argument against any system that imposes work, since all systems do and the question is whether the payer had a say.
- Where the burden is bounded and small, in which case the distributional point is real and not decisive.
Reach for something else instead
- Return the cost — friction that puts verification back on whoever makes the claim, which addresses origin rather than magnitude.
- Compensate the payer — paying reviewers and triagers makes the burden sustainable without reducing it, and makes consent real.
- Cap the load — bound the burden regardless of demand, so the payer's exposure does not scale with someone else's volume.
- -
Read more on the blog
- The sepsis model caught 7% of what clinicians missedA sepsis warning system ran at hundreds of US hospitals before anyone outside the vendor validated it. The external check found the number that matters is not the one being reported.
- Slop scores premium 70% of the timeThe first rigorous measurement of machine-generated content in ad buying found it passes every quality check the industry uses, and passes them better than real inventory does.
- Generation takes seconds. Debunking takes hours.Four open source projects closed their doors in one month. The cause is not bad contributions but a cost ratio that inverted, and the same maintainer who shut his bounty credits the technology with finding 100 real bugs.
- 61% flagged for non-native writers, 3% for nativeThe tool that would answer how much text is machine-written does not work, and its errors concentrate on a specific group for a reason that is structural rather than fixable by tuning.
Further reading
- Wong et al. (2021), External Validation of a Widely Implemented Proprietary Sepsis Prediction Model — alert volume absorbed by clinicians who did not procure the system.
- Stenberg (2026), The end of the curl bug-bounty — evaluation burden externalised onto an unpaid maintainer until the programme closed.
Primary sources, listed so you can check the claims on this page rather than take them on trust.
Where people go wrong
- Reporting an aggregate error or alert rate without disaggregating who absorbs it.
- Reading a system as sustainable because the party measuring it is not the party paying.
- Treating exit as evidence the burden had just become severe, when it is usually evidence it had been severe for a while.
At a glance
FieldSafety & Ethics
Questionnot how large the cost is but who pays it
Three signspayer is downstream, unpriced, and cannot decline
Distinct from error asymmetry, which compares error types rather than payers
Usual first measurementexit
DifficultyBeginner → Intermediate
Where this sits
A starting point. Nothing needs to come before it.
0Levelstarting point
0Needs firstnothing
0Opens upnothing further
1Areastays here
Computed from the prerequisite graph, not assigned. How this works