The thesis is sound but I think this underplays just how hard and rare "genuine creativity" is. Especially in the more technical fields.
If you're painting an oil canvas it's by definition creative, but say programming even designed from scratch apps follow established patterns.
Actual flashes of genuine creativity in a consequential manner are rare there. Could you produce a flash like fast inverse root on demand? I sure can't
The most important part of FINVSQRT is the right shift by one bit. This halves the exponent, which gives the square root to within a factor of 2. Subtraction inverts the exponent since we're looking for an inverse square root. The second most important part is the Newton-Raphson iterations which refine the initial approximation. The most magical seeming part is also the least important: what you do with the mantissa in the initial approximation.
Comments
The thesis is sound but I think this underplays just how hard and rare "genuine creativity" is. Especially in the more technical fields.
If you're painting an oil canvas it's by definition creative, but say programming even designed from scratch apps follow established patterns.
Actual flashes of genuine creativity in a consequential manner are rare there. Could you produce a flash like fast inverse root on demand? I sure can't
https://en.wikipedia.org/wiki/Fast_inverse_square_root
The most important part of FINVSQRT is the right shift by one bit. This halves the exponent, which gives the square root to within a factor of 2. Subtraction inverts the exponent since we're looking for an inverse square root. The second most important part is the Newton-Raphson iterations which refine the initial approximation. The most magical seeming part is also the least important: what you do with the mantissa in the initial approximation.