Players sometimes ask how the object pool is assembled. The current pool contains over a thousand items, ranging from kitchen staples to wildlife to architectural details. Each one passes through a multi-step curation process before it ever appears in a round.

First, candidates are sourced from broad real-world categories: food, animals, plants, household objects, tools, vehicles, sports equipment, musical instruments, materials and natural phenomena. We deliberately bias toward concrete nouns that any adult player would recognize on sight, and we exclude proper nouns, brand names, and culturally narrow references.

Second, every candidate is checked for image suitability. The game only works when the object has a recognizable extreme close-up — meaning rich texture, clear color, and a distinctive surface. A blank white sphere fails this test even if the noun is universally known. Items that fail get either an exclusion flag or an admin override pointing to a better reference image.

Third, items are tagged for difficulty and category density. We try to keep the answer wheel populated with semantically adjacent decoys at higher levels (for example pairing 'plum' with 'fig' rather than with 'wrench') so that late-game rounds reward genuine recognition rather than process of elimination.

We continue to add and refine items every month based on player telemetry. If certain objects are guessed correctly more than 95% of the time at maximum zoom, we either retire them or move them to lower difficulty buckets. If others are missed by everyone, we look at the source image and often find a better one.