this post was submitted on 22 Jun 2023
20 points (91.7% liked)

Asklemmy

44149 readers
1453 users here now

A loosely moderated place to ask open-ended questions

Search asklemmy ๐Ÿ”

If your post meets the following criteria, it's welcome here!

  1. Open-ended question
  2. Not offensive: at this point, we do not have the bandwidth to moderate overtly political discussions. Assume best intent and be excellent to each other.
  3. Not regarding using or support for Lemmy: context, see the list of support communities and tools for finding communities below
  4. Not ad nauseam inducing: please make sure it is a question that would be new to most members
  5. An actual topic of discussion

Looking for support?

Looking for a community?

~Icon~ ~by~ ~@Double_A@discuss.tchncs.de~

founded 5 years ago
MODERATORS
20
Deleted (lemmy.dbzer0.com)
submitted 2 years ago* (last edited 1 year ago) by IsThisLemmyOpen@lemmy.dbzer0.com to c/asklemmy@lemmy.ml
 

Deleted

you are viewing a single comment's thread
view the rest of the comments
[โ€“] lvxferre@lemmy.ml 16 points 2 years ago* (last edited 2 years ago) (2 children)

I'll abstract the problem a tiny bit:

  • a = the prize in box A
  • ka = the potential prize in box B; i.e. "k times larger than a"
  • p = the odds of a false positive. That is, the odds that you pick box B only and it got nothing, because dumb machine assumed that you'd pick A too.
  • n = the odds of a false negative. That is, the odds that you pick A+B and you get the prize in B, because the machine thought that you wouldn't pick A.

So the output table for all your choices would be:

  1. pick nothing: 0
  2. pick A: a
  3. pick B: (1-p)ka
  4. pick A+B: a + nka

Alternative 4 supersedes 1 and 2, so the only real choice is between 3 (pick B) or 4 (pick A+B).

You should pick A+B if a + nka > (1-p)ka. This is a bit messy, so let's say that the odds of a false positive are the same as the odds of a false negative; that is, n=p. So we can simplify the inequation into

  • a + nka > (1-n)ka // subbing "p" with "n"
  • 1 + nk > (1-n)k // divided everything by a
  • 1 + nk - (1-n)k > 0 // changed sides of a term
  • 1 + 2nk -k > 0 // some cleaning
  • n > (k-1)/2k // isolating the junk constant

In OP's example, k=1000, so n > (1000-1)/(2*1000) โ†’ n > 999/2000 โ†’ n > 49.95%.

So you should always pick B. And additionally, pick A if the odds that the machine is wrong are higher than 49.95%; otherwise just B.

Note that 49.95% is really close to 50% (a coin toss), so we're actually dealing with a machine that can actually predict the future somewhat reliably, n should be way lower, so you're probably better off picking B and ignoring A.

[โ€“] IsThisLemmyOpen@lemmy.dbzer0.com 4 points 2 years ago (1 children)

I'll abtract the problem...

Proceeds to teach calculus

[โ€“] lvxferre@lemmy.ml 2 points 2 years ago

BRB, finding a way to insert derivation by parts into that. :^)