Mysterious Treasure

About the app

Mysterious Treasure was not a game, it was a treasure hunt contest with real prizes. Prizes were hidden in areas on the map — a phone, a pair of headphones, a gift card — and players solved the published clues trying to be the first to find the prize. Every contest had a deadline; a counter on screen ticked down to the second.

The real technical problem here was not finding a location, it was verifying one. The treasure's exact coordinates were never sent to the phone; the user only saw the approximate area where the treasure was hidden. Otherwise anyone inspecting the app's network traffic could claim the prize without solving a single clue. The "I found it" claim was evaluated on the server rather than taken from the device, and devices reporting a spoofed location were rejected.

"First to find it wins" sounds simple, but underneath it is a race condition problem. When two people find the same treasure almost simultaneously, the prize must not be awarded twice. I reduced the step that picks the winner to a single write inside a transaction: the prize is awarded once, and later requests see that the contest is closed. By the same logic the countdown read the server clock instead of the device clock — you could not buy yourself extra time by changing the phone's date.

For entering a contest I left three paths: an in-app purchased token, a premium membership, or watching a rewarded ad. Letting users who did not want to spend money play too was essential for keeping the product alive; the moment the number of players drops, the contest itself stops meaning anything.

The app is no longer live. There were two reasons: problems came up on the app store policy side and I did not have the time to run a location-based, prize-giving game on my own. An app that is played in the physical world and hands out prizes was the most demanding product I have built, both in terms of policy and day-to-day operations. What I learned about location verification and race conditions has stayed with me; but if I design something similar again, I will factor in store policies and the daily operational load from day one.