1. 18
  1.  

  2. 9

    So I guess I see why people are ok with this, but it kinda irks me that Guido flat-out ignored the community and domain experts to just go ahead and do it. I guess I should just give up on Python if I find GVR using his BDFL powers to be annoying, but this just feels capricious and inconsiderate of the scientific and mathematical users of python.

    1. 5

      IANAM (I am not a mathematician), but I find tau less problematic than things like division by zero, mathematics is as much an art as a science - and it suffers from poor educators as much as any other subject, which reminds me of A Mathematician’s Lament by Paul Lockhart

      1. 3

        mathematics is as much an art as a science

        IANAM either but I have to say, no it isn’t. Art is subjective. Math is concrete. Math requires proofs.

        If I write a song and tell you it’s great and you say it isn’t, neither of us is empirically right. That’s art.

        If I write 3 + 4 = 12 and you say no it isn’t, I am empirically wrong. That’s math.

        There is art in math, and there is math in art. But math is not “as much art as a science.”

        1. 2

          Mathematics is a notation for the study of number, quantity, shape and space - which for me is why it is an art - it is completely created by people, unlike concrete which is a mix of water, aggregate and cement (sorry for being facetious).

          If I write 3 + 4 = 12 and you say no it isn’t, I am empirically wrong. That’s math.

          That might be wrong in base 10 but actually it’s correct in base 5 (luckily you chose a sum that is correct in base 5 :~D) - who is wrong? We have failed to communicate our notation - hopefully without the catastrophic consequences of the Mars Climate Orbiter.

          I agree that mathematics shouldn’t be subjective, and that 1 + 1 = 2 - but that proof took three volumes by Bertrand Russell and Alfred Whitehead…

          1. 4

            I see that Russell & Whitehead meme a lot in “pop-sci” discussions of mathematics, and it is really misleading. A machine checkable proof of 1 + 1 = 2 from the first principles of constructive logic can fit on a single piece of paper.

            1. 3

              It’s still a good point, though. Mathematics is very dependent on your choice of first principles, and we actually change those pretty frequently.

              Math was not very well formalized until Peano’s Axioms in the 19th Century. But we don’t even use Peano Axioms as the foundation of math anymore! We currently use “ZFC”, Zermelo–Fraenkel set theory. And there is still debate on whether to include the axiom of choice, which doesn’t affect everyday arithmetic but has significant impact on math’s farther reaches.

              Mathematics is a tall and enduring tower, but it’s one we have built.

              1. 2

                It that due to developments in mathematics and logic?

                1. 4

                  Russell and Whitehead weren’t trying to prove that 1+1=2. They were attempting to establishing a rigorous foundation for all known mathematics, which starts you off with mostly logic and doesn’t get to arithmetic until relatively late.

          2. 2

            Interesting article! Gonna share it with some Math teachers.

            1. 2

              Marvin Minsky memo on What makes Mathematics hard to learn? for the one laptop per child project also makes interesting reading as well.

              1. 2

                Thank you again! :-)

          3. 4

            Capricious and inconsiderate? He didn’t remove pi, did he?

            1. 8

              No, but take a look at rhettinger’s commnet, this has real downsides: http://bugs.python.org/msg272373

              1. 5

                I guess. I don’t really see a lot of tau usage accidentally slipping into code, though.

            2. 2

              I personally have much of a bigger problem with how Gamma(n) = (n-1)!. Why Legendre, why? Pi(n) = n! was good enough for Gauss and Riemann.

              1. 2

                Giving up on a language because a founder decided to have fun? If that’s not capricious I don’t know what is. Or rather, I know: this whole argument is about as silly as complaints about significant whitespace. Always has been a litmus test for me when to stop taking people seriously if they’re ready to judge a well-proven popular language on such superficial matters.

                1. 3

                  We’re not talking about him going to play go karts. We’re talking about him making a change to the language against the advice and requests of experts that it will affect. That should not be trivialized as him “having fun”.

                  1. 2

                    This is what he said himself: http://bugs.python.org/issue12345#msg272457. And looking at Python for 10+ years, I’d say decisions like this one are very much in the spirit of the language. If you prefer all decisions to be always made on a logical basis, then Python is simply not your language. (And I’ll reiterate that I don’t think judging languages on such merits makes one look “more professional”.)

              2. 5

                It’s okay if Python occasionally shows its lighter side in unexpected places. Think of the delight of future (junior) high schoolers who discover that Python participates in the tau debate. :-)

                Well, I’ve technically been a senior for about two months, but I’ll confirm nevertheless. ;-)

                1. 3

                  This whole debate feels like bikeshedding of the first order; people know enough geometry / trig to have an opinion and it clouds the debate. This is a constant. Yes, like e it’ll probably not be used a ton, but I for one have a bunch of programs littered with _2pi = 2 * math.Pi that this would actually be helpful, if trivial.

                  1. 3

                    Actually, I’ve heard there are an infinity of them.

                    Can you prove that?

                    Don’t have to. Axiom of infinity. :)

                    This thread made my day :))

                    1. 3

                      What’s wrong with tau in Python?

                      • greek letter names might be convenient when writing equations with pen and paper, as a variable name they are mostly a bad choice. While pi might be tolerable since everyone is exposed to it. tau is a pretty bad idea. A better name for tau might be circumference_of_unit_circle, which kind of describes what the value actually is.
                      • it seems the decision for including it is based on a perceived level of “cuteness”, a tongue-in-cheek thing, similarly Python already allows you to import antigravity. The added benefit is kind of questionable.

                      Whats wrong with tau?

                      π is the established constant, τ is not less arbitrary as a choice than π. While you can argue that τ * radius=circumference requires only one constant, and 2 * π * radius = circumference requires two. But what if we look at the area of a circle first. A = π r^2 needs one constant less than A = τ /2 r^2. If you find that the diameter d is your primary quantity for describing the circle, then circumference = π * d seems even simpler. Yet you might also start campaigning for ϝ= π /4, because A = ϝ d^2… The only real benefit of tau would be, that it might be more intuitively understandable when talking about angles.

                      1. 1

                        The amount of bike shedding here is staggering.