Dumb Philosophy

It’s supposed to be profound … but mostly it is trivially defused.

Make a mistake in mathematics and you start again. Make a mistake in software, it’s a bug. Make a mistake in philosophy and the world has to hear about it endlessly. This site compounds the problem.

Rebukes

Classic arguments that have been getting away with far too much, for far too long.

• The Chinese Room
  John Searle, 1980 · philosophy of mind

  A man with a rulebook shuffles Chinese symbols and understands nothing, so no program ever could. It's a trick: the argument trades on our intuitions about books, and any book that could actually do the job would be bigger than a great many universes. We have no intuitions about those.

• The Gödel Argument
  J. R. Lucas, 1961; Roger Penrose, 1989 and 1994 · minds and machines

  Humans can see that a machine’s Gödel sentence is true, so minds are not machines. Lucas dropped the theorem’s “if the machine is consistent”. Penrose replaced it with “we know we are sound”, which leads to a contradiction with or without machines.

• Gödel Proves Whatever You Like
  Debray, Serres, Chopra, Jaki and others · abuse of mathematics

  A theorem about formal systems of arithmetic, cited as proof that Lenin had to be embalmed, that God exists, that physics can never finish and that a bicycle cannot explain itself. One defendant read the correction and retracted. He was the mathematician.

• The Orangutan Argument
  Yann LeCun, 2019–present · AI risk

  Machines never evolved a drive to dominate, so they will stay subservient. But a machine doesn’t have to want power. It only has to want something for which power is useful, and a two-step coffee run is enough to show it.

• Mary’s Room
  Frank Jackson, 1982 · philosophy of mind

  A scientist who knows every physical fact about colour sees red for the first time, and “it seems just obvious” that she learns something physics left out. Nobody can imagine knowing every physical fact. Jackson later called his own argument a triumph of philosophical ingenuity over common sense.

• Just Predicting the Next Token
  Bender et al., Chomsky, Marcus, Chiang, LeCun, Kambhampati · AI

  It only predicts the next token, so it cannot understand or reason. Every distribution over text factorises into next-token conditionals, and for any algorithm there is a corpus whose best next-token predictor is that algorithm. The premise rules nothing out.

• The Tao of Physics
  Fritjof Capra, 1975 · abuse of physics

  Modern physics and Eastern mysticism turn out to agree, once you pair sentences by translation. The theory he staked it all on was being overtaken as he wrote, and two afterwords insist otherwise.

• Machines Can’t Do Abduction
  Erik Larson, 2021; Jerry Fodor, 2000 · AI

  Guessing the best explanation is said to be unprogrammable. There is a program on the page that does one piece of it, and recovers Kepler’s third law from six numbers. You can run it yourself.

Think a rebuke is itself a dumb argument? Entirely possible. Open an issue on the repo and make the case.