Algorithms Need Math Not Code | Generated by AI
I’ll search for information about Leslie Lamport’s talk with this title.Based on my search, while I didn’t find a specific talk with exactly that title, Leslie Lamport has made this argument consistently across several talks and writings. His core message is:
Lamport argues that the difficulty in program verification comes from the algorithm itself, not from the programming language used to express it. He believes programmers who are experts in languages like C++ often still can’t write programs that do what they’re supposed to do.
His main points include:
Mathematics over programming languages: Lamport argues that algorithms should be expressed in mathematics rather than programming languages, since the need to handle large programs and produce efficient code adds complexity to programming languages and limits their expressiveness.
Think before you code: Lamport believes the best way to get better programs is to teach programmers how to think better, rather than giving them new languages or better debugging tools. He laments how programmers often cobble together a system before writing a proper specification.
TLA+ as his solution: He created TLA+ to force engineers to think mathematically and write specifications before coding. He wants people to “think outside the box” - and the box is “thinking programmatically” in coding languages rather than thinking mathematically.
His critique is fundamentally that focusing on programming languages misses the point - the real challenge is clear algorithmic thinking, which is best expressed through mathematical specifications.