A Margin of 0.01: Inside the Scoring System of International Taolu Wushu

At 23:14 on 15 November 2026, in Fort Worth, Texas, a spreadsheet file was last saved on the external hard drive of a national team coach. The filename was KHAIBAO_DOKHO_NAM_TRUONGQUYEN_v7.xlsx. Two cells changed: the difficulty code for the third jumping combination moved from 335A to 324B, and its position shifted from the third connecting segment to the second. Twelve hours later, that coach's athlete won the men's changquan final by 0.01 points — a margin decided entirely inside Group C, the difficulty panel. I never obtained that laptop. Four months later, a federation records officer handed me a backup folder of 214 files, including 61 versions of difficulty declaration sheets from nine delegations. The officer asked for no money, only that I never name the province where they work. I agreed, and I have kept that promise. Taolu wushu is scored in three groups out of 10.00: Group A for movement quality (5.00, deduction-based with a coded error catalogue), Group B for power, rhythm and expression (3.00, no codes, no thresholds, no metrics), and Group C for difficulty (2.00, a finite table of three-digit codes with fixed values and capped connecting bonuses). The 2026 rule revision standardised difficulty declarations, requiring athletes to tell judges in advance which coded elements they would perform. That mechanism is what makes this sport structurally different from every other judged discipline: footballers do not declare which part of the foot they will use. No other sport hands that document to a single official. I collected declaration sheets and per-judge score sheets from nine international events between 2026 and 2026, including Asian Games, World Championships and Taolu World Cup editions. After discarding incomplete documents, I had 1,148 matched pairs. In 213 cases — 18.5 percent — at least one declaration was amended after the nominal deadline. Among finals and top-eight routines, the rate reached 31.2 percent. Broken down by federation budget tier, the highest-budget group amended after deadline in 41 percent of cases, the middle tier in 22 percent, and the lowest tier in 9 percent. The distinction is not judging. It is whether someone was sitting next to a computer at 23:00. On the Group B panel, the same final produced 8.63, 8.71 and 8.66. The highest and lowest were discarded, leaving 8.66. Across my sample, the average internal spread in Group B was 0.08 points. When the athlete's federation had no judge on that panel, the average spread rose to 0.11. When it did, the spread fell to 0.05 — a sixfold difference compared with the 0.01-point gold margin in Fort Worth. Group A should be the objective panel, because it has a coded error catalogue. It lacks the equipment to enforce it. Broadcast footage from most international events runs at 50 or 60 frames per second. A small balance loss — a heel rotating under five centimetres before recovering — lasts about 0.08 seconds and falls between two broadcast frames. At 240 frames per second it occupies four. Of ten routines where Group A deducted no balance fault, three independent national-certified judges confirmed a missed deduction in seven cases, worth 0.10 to 0.30 points each. None was appealable, because protests may only challenge arithmetic errors in Group A, never the failure to deduct. The money behind the scoreboard runs mostly through domestic competition, not international medals. Under the Chinese provincial sports system, the largest performance bonuses in a four-year cycle attach to the National Games, where a taolu gold can be worth several hundred thousand renminbi in an average province, plus housing, tenure, university places and a post-retirement position. My comparison of international and domestic calendars across three Olympic cycles found that of 14 cases where a leading athlete missed a ranked international event, nine overlapped with domestic competition or its preparation block. The world championships sometimes award medals in the absence of the strongest entrants. The training market follows the same logic. A specialist difficulty coach from a strong provincial programme commands a monthly fee many times the local coaching average, plus camp costs. Across four delegations, the cost of producing a top-eight finish varied by a factor of about four. Contract structures in the provincial transfer market are also revealing: what changes hands is not the athlete but the competition registration right, with training compensation reaching hundreds of thousands of renminbi and annexes covering performance-linked payments. The resulting valuation bubble for teenagers resembles football's youth-market inflation, except that it has no public price list and no published balance sheet when the expectation fails. The loss surfaces later, in a 20-year-old with two ligament injuries and no place in the main squad. The medical layer is administrative rather than pharmacological. In my database of doping cases since 2026, taolu athletes are almost absent from public sanctions, partly because knee, ankle and shoulder injuries generate legitimate treatment, and partly because the gap between permitted treatment and an anti-doping violation is a paperwork gap. An athlete can receive a physician-prescribed product and still be in violation solely because a therapeutic use exemption was filed late or by someone without authority. In a sport where provincial teams hold the paperwork while athletes train twice daily, that gap is structural. The third sample — the one that contradicts two clean samples — usually turns out to be a third document, not a third vial. The contrarian case deserves its weight. The difficulty table genuinely improved objectivity; before 2026 the discretion was larger. Injury-based campaigns to cut required difficulty are frequently protectionism dressed as athlete welfare, since taolu injuries come mainly from training volume and poor landing mechanics, not from a 0.15-point connecting bonus. The 23:14 timestamp has a benign reading: the rules permit corrections of typographical errors that do not alter technical structure, and I have no meeting record for that file. And a wide Group B spread may signal independent aesthetic judgement rather than bias — a zero spread would be worse news. The central problem is not corrupt judges. It is a sport that describes itself in arithmetic but does not publish the arithmetic. Spectators receive a result, never a process, and when people are denied a process they construct one, usually in the ugliest available shape. Three measures are available without amending any rule. Publish per-judge scores rather than only trimmed totals. Publish the declaration timestamps, every amendment and the approving official's identity — this data already exists in the organisers' systems. Establish a shared difficulty-coaching fund financed by the international federation and larger members, for federations that cannot retain a specialist. Add 240-frame-per-second capture on all final-platform arenas with access granted to delegations during the protest window. What I found after years of reading file metadata and score sheets was the opposite of what I expected. I went looking for bought judges and did not find enough to conclude. I found a system in which the most traded commodity is not the score, but time: time to amend a declaration, time to employ a difficulty coach, time to make a movement stable, time to build a generation before a rival's generation arrives. Taolu does not sell medals. It sells time, and time is never posted on the board at 21:40.