Turn a location-bound 30-day weather oracle into a Google product

While out on a walk, you pick up a magic relic. Wherever you place it, it gives a highly accurate forecast for the next 30 days, but only for the area within a 30-mile radius of the spot where it physically sits. It cannot report on any place it is not at. Google asks you how you would turn this relic into a product. ```hint Start from the constraint The relic only learns about the ground it is standing on. Think about what has to physically happen before any user far from you can benefit, and how that limits who you serve first and how often their data must be refreshed. ``` ### Clarifying Questions - Is there exactly one relic, or can it be copied, or split into smaller pieces that each still work? - If the relic is lifted well above the ground, for example on an aircraft or higher, does the 30-mile coverage change, and does it still count as being "at" the location below? - How long does one reading take, and how long is a reading useful before the location must be read again? - Which audience should the first version serve: everyday users of Google's weather experience, or businesses that plan around the weather, such as farms, utilities and logistics firms? ### What a Strong Answer Covers - A clear goal for the product and a choice of the first users, tied to who values a 30-day outlook most. - A realistic plan for physically collecting readings, including a rough coverage estimate with labeled assumptions and a decision on where to prioritize. - A product shape across Google surfaces, with a free versus paid split and a launch sequence. - Success metrics with guardrails, including how to verify the claimed accuracy. - Risks specific to a single irreplaceable object and to selling advance weather knowledge. ### Follow-up Questions - Suppose it turns out the relic cannot be moved faster than a commercial aircraft and cannot work from the air. How does your launch scope change? - Would you ever restrict access to the 30-day data, for example for traders in weather-sensitive markets, and how would you decide? - A reading for a region is now 20 days old. How do you present that data to users, and when do you stop trusting it?