Twelve Beans, Fewest Weighings

This classic balance-scale puzzle is about extracting maximum information from a limited number of tests. It turns a simple-sounding situation into a structured search over many hidden possibilities. In quant prep, it is often grouped with logic and game theory style questions because you must anticipate all outcomes in advance and design a robust plan that can never fail, no matter how the scale tips.

It trains discrete reasoning, information-theoretic thinking, and careful state-tracking under uncertainty. You practice encoding possibilities, eliminating scenarios systematically, and designing decision trees that remain consistent across all branches. It also develops the habit of planning experiments that are globally optimal, not just locally reasonable, a core skill in high-level quant interviews.

This matters for quant interviews because real trading and research problems mirror this structure: many hidden states, few observations, and high penalties for inconsistent logic. Interviewers use this puzzle to see how you balance creativity with rigor, structure a complex search, and communicate clearly under pressure. It is a compact test of the mindset required for quantitative finance interviews and serious quant prep.

What it tests

This problem class is governed by the principle of information theory: each weighing on a balance scale can yield three possible outcomes (left heavier, right heavier, or balanced), so the number of distinguishable cases after $k$ weighings is $3^k$. To uniquely identify both the odd object and its nature (heavier or lighter), you must ensure that the total number of possibilities (number of objects times two, for heavier or lighter) does not exceed $3^k$. The power of this approach is that it transforms a physical comparison into a combinatorial encoding problem, where each outcome sequence maps to a unique scenario. The reason this works is that the balance scale's trinary output (not just binary) allows for a much more efficient search than sequential or pairwise comparisons. The structure is always about maximizing the information gained per weighing by partitioning the set into three nearly equal groups and tracking the possible states.