# Explore vs Exploit ## The Idea in Brief Every decision involves a tradeoff: do you try something new (explore) or stick with what works (exploit)? Explore too much and you never benefit from what you learn. Exploit too much and you miss better options. The optimal balance depends on how much time you have left—explore early, exploit late. --- ## Key Concepts ### The Fundamental Tradeoff Exploration gathers information but foregoes known rewards. Exploitation uses existing knowledge but foregoes learning. You can't maximise both simultaneously. Every choice to try a new restaurant is a choice not to return to your favourite. ### Time Horizon Matters The optimal strategy depends on how much time remains. Early in life, career, or a project: explore heavily. The information has time to pay off. Late in the game: exploit what you've learned. There's no time to benefit from new discoveries. The maths: if you have one meal left in a city, go to your favourite restaurant. If you have a hundred meals, try new places early. ### Diminishing Returns on Exploration Early exploration is high-value—you're likely to discover something better than your current best. Late exploration has lower expected value—you've probably already found most of the good options. This is why the optimal strategy shifts toward exploitation over time. ### The Explore/Exploit Dilemma in Algorithms Computer science formalises this as the multi-armed bandit problem: you have multiple slot machines with unknown payoffs. How do you maximise total winnings? The answer involves algorithms like Upper Confidence Bound (UCB): explore options with high uncertainty, exploit options with high known value. --- ## Implications **In career:** Take risks and try different paths early. As you age, specialise in what you've discovered works for you. A 25-year-old should explore; a 55-year-old should mostly exploit. **In business:** Startups should explore product-market fit aggressively. Mature companies should exploit proven models. The mistake is exploring when you should exploit (endless pivoting) or exploiting when you should explore (missed disruption). **In relationships:** Early dating is exploration. Once you've found a good match, exploit—invest in that relationship rather than continuing to search. **In learning:** Sample broadly first (explore many subjects), then go deep (exploit the ones that resonate). T-shaped knowledge: broad exploration, deep exploitation. --- ## Sources - [[Algorithms to Live By]] — The mathematical framework; optimal stopping and multi-armed bandits - [[Dead Companies Walking]] — Failing companies often exploit dying business models too long, failing to explore alternatives - [[Antifragile]] — Taleb's "tinkering" is structured exploration; barbell strategy balances exploitation (safe assets) with exploration (risky bets) - [[Worlds Hidden in Plain Sight]] — The ratio of exploration to exploitation shapes the life history of any adaptive system