Estimation question

Quantify the number of linear road miles in Google Maps.

Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.

What this question tests

Estimation via geographic decomposition, breaking a global quantity down using land area and road density assumptions.

Define scope: global road miles mapped by Google Maps, including all road types, not just highways.
Break the estimate into a funnel: total land area of countries with strong Google Maps coverage, times an assumed average road density per square mile.
State assumptions explicitly, for example roughly 60 to 70 million square miles of land area well covered by Maps, and an average road density that varies sharply between dense urban areas and rural or unmapped regions.
Segment the estimate by region type: dense urban areas have far higher road density per square mile than rural or wilderness areas, so a single global average understates the range.
Calculate a rough total and sanity check against known real world figures, such as the commonly cited total road mileage in the United States alone (around 4 million miles), scaled up globally.
Note the exclusion of roads Google Maps has not yet mapped, particularly in developing regions, and reduce the estimate accordingly.

Anchors the estimate against a known real world reference point, US total road mileage, to sanity check the final number.
Explicitly segments urban versus rural road density instead of applying one flat density assumption globally.
Accounts for Google Maps coverage gaps in less-mapped regions rather than assuming full global road coverage.

Applying a single flat road density assumption across wildly different geography types.
Forgetting to adjust for regions where Google Maps has incomplete data.

How would you validate this estimate using data Google actually has internally?
How does this number change if you only count roads currently used for live traffic data?