Have I invented perfect random number generation computer chips?

So, last night I slept for a very long time. While I slept, I dreamed that I was explaining to a scientist how to do real random number generation with a computer chip. And, you know what, I think it would actually work.

Computers are not good at randomness

The only thing that computers are good at is giving you the same output for any given input each and every time. It is their predictability that makes them so darn useful.

This is a problem when what you need is randomness.

The closest we have to computer-generated randomness is algorithms that give outputs that are hard for humans to guess. The thing is, for any given seed value, you still get the same output. That’s why Minecraft seeds are a thing.

One of the better algorithms is the Mersenne Twister. This is what powers mt_rand() in PHP.

Nevertheless, this sort of randomness is only pseudorandom. Accidentally use the same input, get the same output.

How can we do better?

One of the best ways to generate cryptographically secure randomness is to use some large external source of chaos. For example, Lavarand. Lavarand is powered by a wall of lava lamps, which the system takes photos of at intervals. The system then extracts some amount of the binary of the image, interprets it however the designers want (all things in computers are just binary, and we only know what sort of thing they are by leaving hints). This data is then used as the seed for pseudorandom generation.

The inclusion of a pseudorandom step means that although the randomness is pretty good in terms of being unguessable, it is still not truly random.

My answer to truly random numbers from a computer

We start with a 64-bit quantum chip. Quantum computing is different from the computing we know now. QBits sit between 1 and 0 until they are read. This allows for some very specific types of computation at speeds that we cannot currently reach with classical computing.

Qbits – quantum bits – suffer from a significant weakness. They are so tiny that a single photon can flip them. Rather than fight this data damage, we can harness it.

We choose 64 because qbits are a bit tricky to make at scale, and 64 bits is what our regular computers use. You could go for a bigger size, say 128, 256, 512, 1024, etc., for cryptographic uses.

One way would be to give the top of the chip a convex glass “lid” that encuredges light into the chip. Another would be to add a layer of tritium and a phosphor, which would glow and send off photons. You could also use some other radioactive matter as long as it gives off something that can flip a bit.

All of these flips are entirely unpredictable and non-repeatable. For the same seed, you will not get the same data.

With this set-up, you could start with a sudorandom seed and prime the quantum chip with it. Wait a reasonable length of time and recover the now changed and truly randomised data.

There are other quantum properties we can use to further enhance our randomness. One such property that creates headaches for classic for modern computing is quantum tunnelling. We are pretty much at the point of the nm scale that if we go much smaller, electrons may just elect to jump to another bit of the chip.

We can use that too. The chip that reads the randomised qbits could itself have a fine enough size that electrons will at times put themselves somewhere unintended. Use that for the data in and the data out, and we now have an array of bits that had three episodes of randomisation. This step is not necessary, but it would speed up the time to true randomness for each set of 64 bits.

The technology exists for true randomness

None of what I have suggested is beyond our current technology level. Other than the very small nm chip design, which might need a little R&D to perfect. We can make this if we want to.

I see a few possible uses for this idea:

1. Cryptography for robust and secure internet connections and the like
2. Gaming, especially for procedurally generated and rough-like games
3. Computer-generated graphics where a certain kind of noise is needed

How to steal this idea

If you are someone with the means to take this idea and turn it into a product, I ask only for accreditation and some company shares as a thank you.

Over to you

I’d love to hear your thoughts and comments. Agree, disagree, or point out a flaw in my thinking – it’s all good. Talk to me.

#flippedBits #MersenneTwister #noise #PHP #quantumComputing #randomness #RNG #Technology