world problems with naive shuffling
I'm writing a number of articles meant to teach beginning programming concepts through the use of poker-related topics. Currently, I'm working on the subject of shuffling.
As Jeff Atwood points out on CodingHorror.com, one simple shuffling method (iterating through an array and swapping each card with a random card elsewhere in the array) creates an uneven distribution of permutations. In an actual application, I would just use the Knuth Fisher-Yates shuffle for more uniform randomness. But, I don't want to bog down an explanation of programming concepts with the much less coder-friendly algorithm.
This leads to the question: Just how much of an advantage would a black-hat have if they knew you were using a naive shuffle of a 52-card deck? It seems like it would be infinitesimally small.
It's not like you're writing a poker program that will be used for an actual online gambling site. An ability for someone to cheat at the program isn't a big deal when you're teaching people how to program.
Leave a note saying that this is a poor model of the real world (with a reference to it as a possible security flaw), and just keep going with the teaching.
The knuth shuffle is an insignificant change compared to the naive shuffle: Just swap with any card in the remaining (unshuffled) section of the deck instead of anywhere in the entire deck. If you think of it as repeatedly choosing the next card in order from the remaining unchosen cards, it's pretty intuitive, too.
Personally, I think teaching students a poor algorithm when the proper one is no more complicated (and easier to visualise!) is a bad approach.
It turns out the advantage is quite significant. Check out this article
Part of the problem is the flawed algorithm, but another part is the assumption that you can get "random" numbers from a computer.
链接地址: http://www.djcxy.com/p/85142.html上一篇: 迭代时从NSMutableArray中移除的最佳方法是什么？
下一篇: 世界问题与幼稚洗牌