Simulated Annealing

An optimization algorithm inspired by metallurgical annealing—heating metal then slowly cooling it to reduce defects. The algorithm starts by accepting both improvements AND random worse solutions (exploration at 'high temperature'), then gradually reduces willingness to accept worse solutions as it 'cools' (exploitation). This avoids getting trapped in local optima—good-but-not-best solutions that pure hill-climbing would never escape. The key insight: sometimes you need to get worse before you can get better, and the willingness to accept temporary degradation determines whether you find the global optimum.

When to use it

When optimization efforts have plateaued at a good-but-not-best solution; when you suspect you're trapped in a local maximum; when early-stage exploration needs to be broad before narrowing; when the willingness to get temporarily worse might enable getting permanently better.

How it can help

Apply annealing thinking: early in any exploration (career, product, strategy), maintain high 'temperature'—be willing to try approaches that seem suboptimal because they might bypass local maxima. As you gather information, gradually reduce exploration and increase exploitation. The critical insight for careers and businesses: if you've optimized to a local maximum and feel stuck, you may need to temporarily accept worse conditions (get 'hotter') to escape and find something better.