Login
You're viewing the functional.cafe public feed.
  • Nov 14, 2026, 9:23 PM

    Ultimate truth, really?

    So here is something I know to be true in two independent ways:

    <omitted because it is not important for this discussion>

    One way is my proof. The other way is somebody else's proof, which was published before I was aware.

    If I had been aware, I would probably not have produced my proof.

    But, as a matter of fact, I am not happy with my proof or their proof.

    So I am trying to get a proof better than both.

    Why am I doing this if I already know it is true, in two independent ways?

    Because I don't think the two existing proofs capture the real reason why this is true.

    I think I have a better way to explain/understand why it is true.

    Mathematicians do this **all the time**.

    If anybody is telling you that mathematicians only care about what is true, they are utterly misguided.

    We always want the best proof. And if not the best proof, another way to understand the same thing from a different angle.

    💬 9🔄 33⭐ 71

Replies

  • Nov 14, 2026, 9:37 PM

    @MartinEscardo yes, and I find it funny that we have no general scientific way to explain why one proof would be better than another one. Some reverse mathematics arguments work, but they don't capture what mainstream mathematicians are interested in. It might be a matter of length, or of complexity of the theories that are used in the proof. (Amer math monthly is a treasure cave for elementary proofs of mathematical gems.) In the end, it's really about a point of view about the objects that appear in the statement, and many distinct point of views can coexist and be all equally valid.

    💬 1🔄 0⭐ 9
  • Nov 14, 2026, 9:50 PM

    @MartinEscardo In Hilbert's 24th problem, there is a comparison of proofs with rivers and the idea that one should investigate the space between two proofs in the same way that one looks at the space between two rivers. What formal language would be able to discuss this, I don't know ? (Many interesting mathematical theorems are propositions hence their proofs are equal for theories such as Lean's type theory, or Hott.)

    💬 1🔄 1⭐ 6
  • Nov 14, 2026, 10:15 PM

    @antoinechambertloir @MartinEscardo

    A few years ago there was a special issue of "Philosophical Transaction of the Royal Society" on Hilbert's 24th problem: royalsocietypublishing.org/rst (All papers are (now?) #OpenAccess )

    Some of the papers there address the issue of proof comparison either in terms of simplicity or generally. In particular, the one by my former colleagues Malheiro & Reis tries to explain what Hilbert's approach using syzygies may have been.

    (I contributed a paper, but it doesn't focus on comparison, but instead on visual thinking and simplicity in proof.)

    #simplicity #proof #DavidHilbert #VisualThinking

    💬 0🔄 3⭐ 3
  • Nov 14, 2026, 9:41 PM

    @MartinEscardo I realised some months ago that some of the proofs/constructions in my thesis were cheating; it has taken the development of new ideas to overcome this and produce more honest proofs.

    The former constructions misused and abused setoids; the latter constructions should work for QIRTs. At least that is my hope as I am no expert in them.

    💬 0🔄 0⭐ 3
  • 💬 0🔄 0⭐ 2
  • Sgstuartgale@types.pl
    Nov 14, 2026, 10:18 PM

    @MartinEscardo was discussing something to this effect on Friday: making matrix multiplication more efficient is still being worked on for more than 200 years.

    💬 0🔄 0⭐ 1
  • 💬 1🔄 0⭐ 0
  • 💬 0🔄 0⭐ 1
  • 💬 0🔄 0⭐ 0
  • Nov 14, 2026, 11:13 PM

    @MartinEscardo Obvious remark that I am sure Martin already appreciates at least as well as I do: there may not be "the real reason why this is true" or "the best proof", because sometimes (1) a thing may be true for multiple overdetermined reasons and relatedly (2) different proofs may illuminate different aspects of why something's true and there might not be a single proof that's optimal for illuminating all of them.

    (I suspect this is the point @lanagol was making by asking about Pythagoras or quadratic reciprocity, but no harm making it explicit.)

    💬 0🔄 1⭐ 2
  • 💬 0🔄 0⭐ 0
  • 💬 0🔄 0⭐ 0