TED Conversations

Casey Christofaris

Owner, CS3 Inc


This conversation is closed.

How do we prove an answer

that it How do we prove an answer

I just want to clarify that I do love science and the understanding of the universe that it has brought us. As well as the tech


Showing single comment thread. View the full conversation.

  • thumb
    Feb 25 2013: I guess that depends on the question being answered...but since you said "prove"...

    A reductionist and scientific perspective on this:

    Statistically speaking, you come up with a hypothesis (what you guess the answer is), and then you forget that, and try and prove the opposite: you try and prove the null hypothesis (that nothing happens, or that there is no effect). If you fail to prove the null hypothesis, only then do you conclude that the alternate hypothesis (what you guessed) was right Keeping in mind that some uncertainty is inevitable: 95% certainty is usually considered good enough.

    The nice thing about science as a method for generating answers is that the field is so concerned with being unbiased, that statistical tests are designed so that you can prove yourself wrong if it all possible - before considering that you are right. Then add into this the need for replication of results and peer review and you have a pretty elegant system for answering questions (at least ones that are testable - and "provable").
    • thumb
      Feb 26 2013: In short, we prove something to be true by doing everything we can to prove that it's false. What survives these attempts is considered to be true (with a confidence level proportional to our effort). "Survival of the fittest" - evolution applied to ideas.

      Another paradox of life. To prove ourselves right, we need to do our best to prove ourselves wrong.
      • thumb
        Feb 28 2013: But even is you have created that confidence level up on proof of an answer given time it will always be proven wrong?
        • thumb
          Mar 1 2013: Not always. It *may* be proven wrong with probability equal to one minus confidence level.
        • thumb
          Mar 1 2013: all things must pass - a wise Beatle said that..

          it's like you can't gauge happiness without sorrow, light without darkness, male without female. the duality behind the singularity. full circle back to where we started. life is very neat and tidy in that way.
    • thumb
      Feb 28 2013: Why would you start with answer to prove an answer? Seems self-fulfilling to me

      see ted convo
      • thumb
        Mar 1 2013: What Letitia said is correct. You don't start with an answer to prove an answer. Just the opposite, you start by trying to contradict the answer - thinking of any possible alternative explanations.

        "Meaning is exclusion." http://logictutorial.com/

        When you say something, the more your statement excludes, the more you say. E.g. saying that "the sky is blue" excludes all other colors from consideration. To say this is more meaningful than to say that "the sky is not green". To prove that "the sky is blue", essentially, you need to prove that it does not have any other color. The more colors you exclude, the more confident you are in your answer.

        If you deal with known and limited amount of possibilities, you can get your confidence level up to 100%. But this is rarely the case. Confidence level is 1 - estimated probability of being wrong. This is uncertain world. Probability is all we've got.

        So, it's not "self-fulfilling", but self-refuting.
        • thumb
          Mar 1 2013: "So, it's not "self-fulfilling", but self-refuting."

          So could that be seen as equals but opposite?
      • thumb
        Mar 1 2013: Re: "So could that be seen as equals but opposite?"
        Yes. Coincidentia oppositorum. The concept is not new.
        • thumb
          Mar 1 2013: Right so why does it matter how you come up with the answer if the answer is always the same
      • thumb
        Mar 1 2013: Which answer are you talking about?
    • thumb
      Mar 1 2013: Selecting the null-hypothesis is somewhat tricky. The scientist must make sure that the null-hypothesis and the hypothesis are mutually-exclusive and, in combination encompass all possibilities. Typically, if hypothesis is "A", the null-hypothesis should be "not A".

Showing single comment thread. View the full conversation.