How to Control JavaScript Randomness: Seeded Random Numbers and V8 Implementation

Background

CodeWars kata "Don't rely on luck" requires guessing a number that matches Math.floor(Math.random() * 100 + 1). The test fixture compares the guess with the generated random number, encouraging a solution that does not depend on luck.

Common cheat solutions

Because the test uses Math.random, participants often override built‑ins:

Math.random = function(){ return 0; };
var guess = 1;

var guess = 10;
Math.floor = function(){ return guess; };

Seeded random numbers in other languages

C provides srand(), Ruby seeds from system time, process ID and a counter. JavaScript lacks a comparable low‑level seed function.

V8 implementation of Math.random

V8 follows ECMA‑262 §15.8.2.14 and maintains an internal Uint32Array state. A simplified version of the algorithm is:

var rngstate; // Uint32Array initialized at startup
function MathRandom(){
  var r0 = (MathImul(18030, rngstate[0] & 0xFFFF) + (rngstate[0] >>> 16)) | 0;
  rngstate[0] = r0;
  var r1 = (MathImul(36969, rngstate[1] & 0xFFFF) + (rngstate[1] >>> 16)) | 0;
  rngstate[1] = r1;
  var x = ((r0 << 16) + (r1 & 0xFFFF)) | 0;
  return (x < 0 ? x + 0x100000000 : x) * 2.3283064365386962890625e-10;
}
function MathImul(x, y){ return %NumberImul(TO_NUMBER_INLINE(x), TO_NUMBER_INLINE(y)); }

Simple custom seeded RNG

A quick deterministic replacement derives output from Math.sin:

Math.random = function(seed){
  return ('0.' + Math.sin(seed).toString().substr(6));
};
// Example usage
Math.random(1); // → 0.709848078965

Practical use case

In HTML5 mini‑games, storing the seed value allows the entire random scene (item drops, enemy spawns) to be regenerated after a pause without persisting every object.

seedrandom library

The widely adopted seedrandom library provides a seedable RNG for browsers and Node.js. Installation examples:

<script src="//cdnjs.cloudflare.com/ajax/libs/seedrandom/2.4.0/seedrandom.min.js"></script>

$ bower install seedrandom

$ npm install seedrandom

True randomness vs. pseudo‑randomness

All the methods above generate pseudo‑random numbers, which are deterministic given the same seed. Genuine entropy can be obtained from external services such as random.org (atmospheric noise), HotBits (radioactive decay), or EntropyPool (multiple sources). A JavaScript wrapper for these services is available at https://github.com/RealRand/RealRand.

Conclusion

JavaScript does not expose a native seedable RNG. Developers can mimic V8’s internal algorithm, use a simple deterministic function, or rely on third‑party libraries like seedrandom. Controlling the seed enables reproducible randomness, which is essential for debugging, testing, and creating consistent game experiences.