Story points were supposed to be humble. The idea was to admit that software estimation is guesswork, replace hours with a rough relative scale, and stop pretending to a precision nobody has. Then the number escaped the team. Velocity became a target, targets became capacity plans, capacity plans became commitments to stakeholders, and the humble hedge turned into a fiction everyone agrees to maintain. Planning poker is now often a ceremony performed to produce a number that will be used against the people who produced it.
The problem is not estimation. Teams need a sense of size to make choices. The problem is what happens the moment an estimate becomes a goal.
From guess to target
The failure follows a familiar logic: a measure that becomes a target ceases to be a good measure, because people optimize the number rather than the thing it was meant to describe. Story points are unusually easy to game because they are defined by the team itself. If velocity must go up, points per story can go up, and the chart improves while nothing real does. The estimate stops tracking the work and starts tracking the incentive.
This is why single-metric management of engineering keeps failing. Forsgren and colleagues, in the SPACE framework (2021), argued that developer productivity is multidimensional and cannot be captured by any one number, and that isolated activity metrics, points, commits, lines, induce perverse incentives while satisfaction and flow predict outcomes as well as raw output. Velocity is exactly the kind of activity metric they warn about: legible, easy to chart, and easy to corrupt once it carries consequences.
The corruption is often invisible because the number keeps looking healthy. A team under pressure to raise velocity does not usually cheat crudely; it recalibrates. Stories that used to be three points quietly become five. Work gets split so more cards close. Estimates drift upward across quarters while the actual delivery of value holds flat or falls. The chart that leadership reads as a productivity trend is measuring nothing more than the team's growing understanding of what leadership wants the chart to say. Everyone is acting reasonably, and the instrument has quietly stopped measuring anything real.
The measurement that changes the measured
There is a second, quieter cost. Attaching stakes to an intrinsically engaging task tends to erode the engagement itself. Deci's early experiments (1971) showed that paying people to do something they already found interesting reduced how much they did it voluntarily afterward, the overjustification effect, and the meta-analysis by Deci, Koestner and Ryan (1999), covering 128 experiments, found that tangible rewards contingent on performance significantly undermined intrinsic motivation. When hitting a point target becomes the reward and the review criterion, the work that was its own motivation becomes a means to a number. Estimation theater does not just distort the plan; it can drain the reason people cared about the plan.
The DORA research (2024) adds the systemic version of the same lesson: unstable priorities damage productivity and raise burnout regardless of how good the leadership is. A team whose estimates are constantly reopened, renegotiated, and weaponized experiences precisely that instability. The ritual meant to create predictability manufactures the opposite.
Semler cut through the equivalent problem by refusing to measure the proxy at all. As he has told it, after his collapse at twenty-five he stopped counting the hours people worked and started asking a single question: was the pump delivered? The proxy, hours logged, had been standing in for the outcome, and once he measured the outcome directly the proxy became noise. Story points are the software industry's time clock: a proxy that is useful only until someone starts managing it, at which point it begins to lie.
The honest use of estimation is small and internal. A team can size work to decide what to pull next, to notice when a story is secretly enormous, to have a conversation about risk. The dishonest use is external and quantitative: velocity as a KPI, points as a promise, the estimate as a stick. The first is a tool. The second is theater, and the audience it plays to, the dashboard, the capacity model, the quarterly commitment, cannot tell the difference between real progress and a well-inflated number. Which is exactly why the number gets inflated.
Sources
- Forsgren, Storey, Maddila, Zimmermann, Houck & Butler (2021). The SPACE of Developer Productivity. Productivity is multidimensional; isolated activity metrics create perverse incentives, while satisfaction and flow matter as much as output. https://www.microsoft.com/en-us/research/publication/the-space-of-developer-productivity-theres-more-to-it-than-you-think/
- Deci (1971). Effects of Externally Mediated Rewards on Intrinsic Motivation. Monetary rewards for an already-interesting task reduced later voluntary engagement (overjustification effect). https://doi.org/10.1037/h0030644
- Deci, Koestner & Ryan (1999). A meta-analytic review of experiments examining the effects of extrinsic rewards on intrinsic motivation (128 experiments). Tangible performance-contingent rewards significantly undermined intrinsic motivation. https://doi.org/10.1037/0033-2909.125.6.627
- DORA / Google Cloud (2024). Accelerate State of DevOps Report 2024. Unstable organizational priorities reduce productivity and raise burnout even under good leadership. https://dora.dev/research/2024/dora-report/