Hacker News new | past | comments | ask | show | jobs | submit login

Yep, that's a condensed version of the universal edibility test.

It's on page 103 in the FM 21-76 US ARMY SURVIVAL MANUAL: http://www.preppers.info/uploads/FM21-76_SurvivalManual.pdf




Step 1 reminded me of an amusing problem: - 16 bottles of wine, one has been poisoned.

- poison takes 1 hour to kill you, party is in 1 hour.

- you have 4 prisoners you don't mind sacrificing to find out which bottle is poisonous but obviously you'll need to do all your sampling now because otherwise you won't know which bottle not to use in time for the party.

EDIT:16 not 17 bottles of wine.


I feel like this is solvable using binary to assign which bottles to which prisoners, but only if there were 16 bottles. For example, bottle 11 dec = 1011 bin. So bottle 11 would be tasted by prisoners 1, 3, and 4. Then if 1, 3, and 4 die, we know bottle 11 is poisoned. Every combination of prisoner deaths would point to a unique bottle.

Except 17. Fencepost error on my part maybe?


No, you're correct I typo'd.


Actually it seems you can have a 17th bottle.

Use binary search to find the poisoned bottle among bottle 1 to 16. If you fail to find a poisoned bottle, it is the 17th


No, that's already covered by 0000, nobody drinks that one. Number the bottles 0 to 15 and it's clearer.


That riddle is sometimes called "criminal cupbearers".

Common presentation uses 1000, gives it away to use a power of two.

A really fun extension to it is to suppose that you use the same criteria (a thousand bottles, consumption of the smallest amount will kill, but after a delay that means you must perform a one-pass test) but _two_ bottles are poisoned. What is the minimum number of prisoners you need for your test to find the two poisoned bottles precisely and what is the procedure?

(This is _substantially_ harder.)


Only works if none of your prisoners are named Reed or Solomon.


Whilst this is a permutations problem, what parties need wine that much that you'd risk fatally poisoning people?


Not people, prisoners. The point is to not poison the actual people attending the party, yet still give them wine.


"Not people, prisoners."

I get that it's a game, but can't let this slip into the memesphere unchecked.

People are a superset of prisoners. Some of our largest problems as a society are a result of our not acting as such.


I think it's obvious from the setup of the puzzle that the prisoners lives are completely expendable without repercussion. That kind of precludes any of the more liberal/modern methods of handling prisoners and leans more towards medieval-style prisons.


I don't disagree with you on the obviousness of the set-up. I'm drawing attention to the more subtle language structure which, by design, defines prisoners as non-persons.

I quite like the mathematics of the puzzle, I also like justice and am aware that justice is a product of language in many instances, as many people interact with the world through language without giving much thought to it.


I don't think language states people aren't prisoners.

In fact, my statement was meant to ridicule the assumption that prisoners are worthless to point out they are people.


... and as second goal, becoming a prisoner in jail some time later, for the cheap prize of some bottles of wine.

This is the kind of problems that AI machines would try to solve, wrongly. Humans still score better understanding that some theoretical problems must not be solved.


Apparently you failed the scum detector. Good luck in your future encounters.


Hamming code.


"Reprinted as NOT permitted by U.S. Department of the Army, but by we the citizenry who paid for it"

Ah, preppers.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: