What the work actually is

You spend the day on two alternating tasks. First, writing prompts that a strong general model should plausibly get wrong — questions that hinge on rule technicalities, competition format changes, tie-breaker procedures, statistical definitions that differ between leagues, or events recent enough to sit near a model's knowledge cutoff. Second, grading responses: marking hallucinated stat lines, fabricated transfer or draft details, confident misstatements of a governing body's rulebook, and reasoning that reaches a correct answer by an invalid route. Every judgement is expected to carry a citation to a reliable source — a league rulebook, an official statistics database, a reputable outlet — not a recollection.

Expect volume and expect documentation. Turing runs this as annotation work feeding research pipelines, so your rationales and your prompt metadata matter as much as the labels. You will also flag ambiguity upward: prompts where the sports community itself disagrees, or where a statistic has multiple defensible definitions, need to be either sharpened or excluded rather than silently graded.

What the screen looks for

  • Verifiable depth in at least one sport, plus real breadth. The listing says global sports, leagues, athletes, tournaments, rules, statistics, analytics. A screener will push past headline knowledge into rules edge cases and how statistics are computed.
  • Sourcing instinct. Can you name where an authoritative answer lives, per sport?
  • Calibration. Willingness to say "I would need to check the current rulebook" is scored positively; bluffing a citation is fatal.
  • Credentials as stated: a Master's degree or higher in any field, and 3+ years of professional experience — sports journalism, sports media, sports content, research, analysis, or reporting are the preferred paths.

Logistics

Fully remote and contractor-basis, with no medical or paid leave. The commitment as posted is 40 hours per week for an eight-week contract, with a minimum four-hour daily overlap with PST. Pay is undisclosed on this listing; Turing typically sets rates per project and per region, so treat any figure you have seen elsewhere as unconfirmed. Application starts with a take-home assessment before any human conversation.