CODEWITHSUNDEEP
random
user2461391
user2461391

Reputation: 1433

Is it possible to generate human-random numbers?

Say I told a human to pick 10,000 random numbers between 1-100. I believe studies have shown that humans are bad at picking random numbers, so the 10,000 numbers shouldn't be evenly distributed.

How would I have a computer generate 10,000 similar numbers - not a true random number, but "human-random"? Would it be as simple as just implementing a crappy random number generator or would there be more to it?

Upvotes: 2

Views: 494

Answers (3)

Farmer Joe
Farmer Joe

Reputation: 6070

Well the first step in writing any algorithm is to thoroughly analyze what you are trying to accomplish. So in this case you would need to sit down and figure out how a human being picks random numbers.

Do they follow any consistent pattern(s)? What are the qualities of the distribution you would like to end up with?

For one thing, human beings can change their strategy for picking numbers while still picking the numbers, so that kind of phenomenon would be tough to program.

I think if you could figure this out it would be some serious insight into human nature, psychology, and artificial intelligence.

Upvotes: 0

Sergii Dymchenko
Sergii Dymchenko

Reputation: 7209

You should look and the studies and understand more formally what human-random is: shape and parameters of the distribution.

Than you model that distribution using uniform distribution or more advanced distributions your language has. (C++11, for example, has a lot of good stuff for random numbers)

Upvotes: 2

Phillip Schmidt
Phillip Schmidt

Reputation: 8818

Depends on how scientifically accurate (if there is such a thing in this case) you want to be. Do you have statistics on how humans generally pick random numbers? Is it closer to the start number? Closer to the end? Around the middle? Generating weighted randoms is easy. You just need to know ahead of time how you want to weight them.

Upvotes: 0

Related Questions