Open the problem

you are in this shape if

• The team has been stuck at the same performance for a long time
• Everyone working on it shares one framing and one training
• The data or the question could be shared without giving away the business

the moves

Publish the problem and pay for the answer

A prize pays only for a solution that works, and reaches people who would never join your payroll.

Release the data

Hand outsiders what your experts have been staring at; they bring instruments your field does not own.

Fund an enemy inside the house

Pay a team to attack your own position — a red team, a blue army, a devil's advocate.

where it was solved

1. 1761John HarrisonNavigation / precision instrumentsHarrison spent decades building successive marine timekeepers (H1 through H4), engineering around the specific mechanical problems — temperature-driven expansion, the rolling motion of a ship, friction — that made ordinary pendulum clocks useless at sea, rather than trying to improve the astronomical method the establishment was already committed to.H4 was tested at sea in 1761 on a voyage to Jamaica and found to be off by only 5.1 seconds after two months, and in a second 1764 trial to Barbados was off by 39 seconds over 47 days — both results roughly three times better than the accuracy the Longitude Act required to win the full prize. The Board of Longitude, dominated by astronomers with a professional stake in the lunar-distance method, repeatedly refused to pay in full, demanding additional trials and disclosure of Harrison's mechanisms; George III personally tested H4 in 1772 and intervened on Harrison's behalf, and Harrison, then in his 80s, finally received prize-equivalent payment from Parliament in 1773.
2. 1948A&P (with USDA)Retail / agricultureA&P partnered with the USDA to fund the 'Chicken of Tomorrow' contest: independent breeders across 44 states submitted hatching eggs to state and then regional rounds through 1946-47, with finalists' chicks raised under controlled, identical conditions and judged purely on objective criteria — weight, feed efficiency, edible meat yield over a 12-week growth period. The national final at the University of Delaware in 1948 judged roughly 2,000 birds from 40 finalist breeders on 18 measured criteria, with no role for A&P's own preferences in picking a winner.Charles Vantress's red-feathered New Hampshire-Cornish hybrid won, decisively outperforming purebred entries, and Vantress's genetics — later crossed with contest runner-up lines into the Arbor Acres breed — became the direct ancestor of nearly all broiler chickens raised commercially in the world today.
3. 1998PixarFilm production / technology operationsThe film was ultimately recoverable only because supervising technical director Galyn Susman, working partly from home after having a baby, had been keeping her own informal, unauthorized copy of the production files on her home computer — colleagues drove to retrieve the machine and used it to rebuild the lost work. In the aftermath, co-founder Ed Catmull explicitly ruled blame-finding out of the recovery process: by his own account, Pixar 'never sought out who was responsible,' reasoning that identifying and punishing the individual wouldn't help anyone learn from the failure, and instead prioritized restoring the film, then fixing the backup system, then adding safeguards against a repeat.Toy Story 2 was fully recovered and delivered on its original schedule; the studio rebuilt its backup infrastructure with automated testing, distributed storage, and access restrictions on destructive commands rather than around any single person's conduct.
4. 2000GoldcorpMiningIn March 2000, Goldcorp posted 400 megabytes of Red Lake's actual geological survey data on the open internet — core samples, drill logs, three-dimensional models, 50 years of proprietary exploration records — and launched the Goldcorp Challenge: $575,000 in prize money to anyone, anywhere, who could analyze the data and tell Goldcorp exactly where to drill next.More than 1,400 entrants from over 50 countries — including geophysicists, graduate students, military analysts and consultants who had never set foot in a mine — downloaded the data and submitted target coordinates. The winning entry came from an Australian graphics team that had never seen Red Lake in person but built a 3-D model exposing patterns the site's own geologists hadn't flagged. Contestants identified 110 targets, over 80% of which yielded significant gold, unlocking deposits worth well over $3 billion and taking the company from roughly $100 million in value to a multi-billion-dollar producer within years.
5. 2006NetflixStreaming media / softwareIn October 2006, Netflix published roughly 100 million anonymized movie ratings from its own users and offered $1,000,000 to the first team that could beat Cinematch's prediction accuracy by 10%, measured by a fixed error metric, with an interim $50,000 Progress Prize each year for the best improvement so far to keep momentum going during the multi-year contest.More than 40,000 teams from over 180 countries entered over three years, and Netflix folded incremental gains from the annual Progress Prize leaderboard into its production algorithm well before any final winner was declared. In 2009, a merged team called BellKor's Pragmatic Chaos crossed the 10% threshold and won the $1,000,000 prize — a fraction of what recruiting an equivalent concentration of machine-learning talent in-house for three years would have cost, and Netflix had already captured much of the accuracy improvement for free during the competition itself.
6. 2008HuaweiTelecommunications equipment / corporate strategyIn 2006 Huawei formalized a 'Blue Army' (蓝军) unit inside its strategy function, its entire mandate to argue against the 'Red Army' — the units running the company's actual current strategy — by role-playing rival companies and proposing genuine alternative strategies, not just raising objections. Ren Zhengfei reportedly told managers that promotion required time served fighting for the Blue Army first. In 2008, as Huawei negotiated to sell a 49 percent stake in its terminal (handset) business to private-equity bidders including Bain Capital, Goldman Sachs, KKR and Silver Lake at an average valuation of roughly $2 billion, the Blue Army produced a one-page report arguing that abandoning the terminal business — just as the iPhone's 2007 launch had made handsets strategically central — would mean abandoning Huawei's future, and proposed a 'device-pipe-cloud' strategy keeping it instead.Ren accepted the analysis and the sale did not go through; the September 2008 financial crisis also disrupted the private-equity bidders that same autumn, but the Blue Army's report had already shifted the internal case for keeping the unit before the crisis hit. Huawei's device revenue reached roughly $4 billion in 2008, up about 82 percent year over year; the company formally created its Consumer Business Group in 2011, and by the following decade the consumer/device business had grown into one of Huawei's largest revenue sources, at times accounting for roughly half of total group sales.

what breaks in transit

• Opening the problem opens it to competitors too; decide what the data reveals before releasing it.
• Prizes attract effort, not necessarily the right effort — the specification is the whole design.
• The winning outsider often cannot implement; budget for the translation.