He rebuilt the Vortex Core in eleven days.

  Not from scratch. The architecture was sound.

  What he changed was the intake mechanism: from a deterministic filter to a probabilistic one.

  Instead of distinguishing ambient from cultivator Qi with certainty, the modified vortex sampled incoming energy in rapid bursts.

  Assessed each burst's probability of being cultivator-sourced.

  Rejected samples above a threshold.

  Not perfect. Could not be perfect.

  But right approximately ninety-four percent of the time.

  Eighty-nine percent accuracy on the first build. Not good enough. A one-in-ten chance of draining someone was a one-in-ten chance too many.

  He refined it over three more days.

  Ninety-four percent. The last six percent resisted every approach he tried.

  He could feel the asymptote — the mathematical ceiling beyond which diminishing returns made further improvement nearly impossible without a conceptual breakthrough he didn't yet have.

  Ninety-four would have to do.

  The parasitic drain dropped from two percent per minute to 0.12 percent — within natural Qi fluctuation.

  The cost: efficiency from fifty-four to forty-seven percent.

  In the Silt, forty-seven would have been godlike. In the Torrent, where serious cultivators achieved fifteen to twenty percent, it was merely excellent.

  This story is posted elsewhere by the author. Help them out by reading the authentic version.

  The gap had narrowed. He was still the best.

  But the margin was no longer absurd.

  Su Yiran designed the test protocol. She thought about edge cases his optimism-bias blinded him to.

  Little Abacus served as the test subject, standing at decreasing distances.

  "Thirty metres: no drain. Twenty: no drain. Ten: within fluctuation. Five: marginal, 0.08 percent. Three: 0.12 percent, consistent with prediction."

  He paused.

  "One metre: 0.15 percent. Still within fluctuation."

  "You were already at one metre."

  "For science."

  Chen Xi looked at the boy. Little Abacus stood within arm's reach, grinning, his topknot freshly re-tied at twenty degrees (Su Yiran had insisted).

  Chen Xi reached out and put his hand on the boy's shoulder.

  First time in six weeks.

  The contact was brief. Five seconds.

  The Probability Core held. The drain was negligible.

  The vortex hummed its steady, imperfect, approximate hum.

  It did not steal from the boy who counted everything.

  He sat on the roof of the Falling Leaf that evening and did what he always did — catalogued himself.

  Foundation Gate 9. Forty-seven percent efficiency, down from fifty-four.

  The Probability Core's filter cost seven percent, and those seven percent bridged the gap between dominant and merely excellent.

  He was still two and a half times the Torrent average. Still the best in any room he entered.

  But the rooms here were bigger and the people in them were Core Formation, and at Core Formation the efficiency gap narrowed because their raw power compensated for waste.

  When the quantum core gave him Gate 12 — maybe a third of the time, unpredictably — his output rivalled early Core Formation.

  He could hurt people two tiers above his nominal classification.

  When it gave him Gate 10 — the rest of the time — he was a target.

  Foundation tier by classification. Core Formation threat by capability. A walking contradiction that no assessment system in the Torrent was built to categorise.

  The Clearwater Crossing average cultivator sat at Foundation Gate 7, running about fourteen percent efficiency.

  The city guards required Foundation Peak and sixteen percent minimum.

  The sect elders who ran the city operated at Core Formation with eighteen to twenty-two percent.

  Chen Xi could outperform all of them — when the coin landed right.

  When it landed wrong, the guards could take him.

  He was, he reflected, the most dangerous unreliable cultivator in the city.

  Which was either an advantage or a death sentence, depending on probability.