Daniel Lemire's blog

2 min read

My rules

  1. Lemire’s creativity rule: Originality is overrated. It is often believed that an idea is only worth pursuing if it is original. Why do something that others have already attempted? Though it is true that you should not enter a crowded market, Google was not the first nor the last search engine. Apple did not produce the first personal computer, nor even the first smart phone. Amazon was not the first online store. Originality of its own sake is probably overrated.
  2. Lemire’ definition of science: Science is neither a collection of facts nor a specific method. Rather, science is the concept that we should constantly challenge received ideas with trials and errors and seek the truth for ourselves.
  3. Lemire’s benchmarking rule: Never benchmark on a laptop. Most programmers use laptops. Laptops are designed with a limited thermal envelope which means that they may frequently switch to lower-power modes, making it harder to run accurate benchmarks. It is also a broader concept: accurate measurements are difficult to achieve. You need much effort to get good measures. Experiments are difficult (but important).
  4. Lemire’s political rule: it is not going to end the way you think it will. Naive players often assume that they have won when their opponents are getting ready for an offensive. In 1941, Hitler thought that Europe was his to command. He was about to fail. In 1991, the Soviet Union occupied one sixth of Earth’s surface, it was a mighty empire. The next year, it would be gone.
  5. Lemire’s Freedom rule: Freedom is something you earn by working constantly at it. Freedom requires considerable psychological strength. Extrinsic motivations (prestige and glory) and fear are the strongest chains. The leader of your country may have less freedom than the kid playing in the courtyard. You must learn to stand alone: it is one of the most difficult step a human being can take, but also the most important.
  6. Lemire’s computing rule: resolute engineering trumps mathematics. It is often not the algorithms and the human cleverness that matters most, but rather the raw speed and efficiency of our computers.