The Lean Researcher
summary
Argument links Lean Startup rules to research and gives concrete, checkable examples: GPUVerify starts as manual two‑thread translations into Boogie and Z3 to test feasibility (works on programs that model thousands of GPU threads after roughly 20 examples), then builds a clang front end; GraphicsFuzz uses GLSL sandbox shaders to generate semantics‑preserving variants, finds real Intel/NVIDIA driver bugs and becomes a startup acquired by Google in 2018. Advocates minimum viable product and build-measure-learn cycles.
AI-written summary · about
chapters (15)
- 0:00Start
- 1:28The Reachability Problem
- 3:09Axiomatic Reasoning: Logical Encoding
- 4:48What about continuous state, continuous time?
- 5:48Cyber-Physical Systems
- 6:25A Simple Reachability Question
- 8:09The Logical View
- 11:08Related Theories
- 25:23Real Algebraic Integers = Integers or Irrational
- 26:21Are all real numbers algebraic?
- 28:33Digression: Cantor's First Proof
- 32:28Transcendental Numbers
- 33:32Liouville's Approximations
- 33:48An explicit transcendental number
- 34:53Lambert 1760s it is irrational
related talks
37mLessons Learned
34mLean Requirements
21mNeed for speed - Moving fast as a software engineer
01mIdan Gazit, GitHub, ETLS Vegas (29)
01h 00mThe Influential Software Engineer
40mFailure for Fun and Profit
40mThe Best Programmer I Know
26mEvidence Over Advice: Learning Programming Faster Using the Scientific Method