Numbers
summary
A fast tour from ancient numerals to software bugs with concrete claims: binary floating point makes 0.1 + 0.2 wrong, causing surprising error accumulation; an overflow caused the Ariane 5 failure; recommendation to use a single decimal type dec64 (56‑bit coefficient, 8‑bit exponent, exact decimal fractions up to 16 digits) as the sole numeric type in application languages; insistence that 0n and 0/n should produce zero* (and n/0 be signaled) to avoid class of optimizer and runtime errors.
AI-written summary · about
chapters (16)
- 0:00Start
- 3:36Lumber Systems from History
- 9:01Why Are We Still Using this Number System
- 16:17Signed Magnitude
- 16:43Ones Complement
- 17:22Two's Complement
- 23:30Floating Point
- 23:55First Form of Floating-Point
- 24:58Binary Floating-Point Standards
- 28:17The Associative Law
- 30:01Text Conversion
- 31:12Aryan 5 Failure
- 42:17Thank Yous
- 42:29Leonardo Fibonacci of Pisa
- 43:35George Boole
- 46:04They Wanted To Show How Important the Position of Scribe Was So They Made the Head Scribe of God and Had all of the Other Gods Come to Him for the for the Good Things That He Can Do Had Magic-because Magic Is Great and the Other Reason They Did It Was They Want To Thank Him They Want To Thank Him for Giving Them the Best Jobs in Egypt but You Know They Weren't Doing that They'D Be Out Farming or Hauling Rocks or Other Hard Work instead They Got To Write and and Count and Do Magic so that's Great Stuff
