# What would change if we had scalable quantum computers **URL:** \ **Category:** In My Humble Opinion\ **Created:** [June 12, 2018, 3:33pm UTC](https://boards.straightdope.com/t/what-would-change-if-we-had-scalable-quantum-computers/815969 "2018-06-12T15:33:30Z")\ **Posts on this page:** 2\ **Page:** 2
**Author:** ![SamuelA](https://avatars.discourse-cdn.com/v4/letter/s/c77e96/32.png) [@SamuelA](https://boards.straightdope.com/u/SamuelA)\ **Post date:** [June 14, 2018, 5:25pm UTC](https://boards.straightdope.com/t/what-would-change-if-we-had-scalable-quantum-computers/815969/21 "2018-06-14T17:25:02Z")
> [@Textual\_Innuendo](#): > > Ah, got it. I wasn’t getting that you were making an assymetric-key cryptography argument instead of a general solution argument. > > But on key-exchange, I think there are already quantum-resistant methods with perfect forward symmetry, so that even if you broke the private key, your session keys still protect your other communications / sessions. On looking it up, SIDH fits that bill. But maybe I’m misunderstanding your implementation. > > Very interesting - so it sounds like you’re proposing a 5 layer implementation here. > > 1. Simulation space modeling > 2. Meta-model variance / accuracy tracking model > 3. Physical implementation of most promising output > 4. Physical testing of what you just built for desired properties > 5. Outputs in 4 feed back to 2 to train the model for better scoring > > The hard part there seems to be automating 3 and 4 except for very specific and constrained solution spaces, but I get it, and that sounds awesome. I’d like much the same thing for the microbio application. > > And I’m assuming the quantum evaluation is happening in 2? Quantum evaluation would be 1 and 2. That machine is at a top level just f(design) = predicted expected reward weighted by variance. So the system would need to guess many designs, evaluate them in simulation, then robots would fabricate the best candidates. Quantum computing is in no way necessary, just existing methods could only explore a few promising regions of the vast space of possible designs. All solutions would be local minima. (And the obvious thing you do is have a generative neural network predict a design a human mechanical or electrical engineer might pick to solve a similar problem) The reason a system like this would scale to superhuman performance is scale - it could have tried millions of real designs given enough robots and trillions of simulated possibilities. As for robot assembly - that’s solved with a very similar solver. Instead of guess a candidate design, check on simulation, try best candidate, update simulation and solver system that chooses initial guesses, you guess ways to move the robot arm that will result in progress towards assembly. ---
**Author:** ![Wesley\_Clark](https://sea3.discourse-cdn.com/straightdope/user_avatar/boards.straightdope.com/wesley_clark/32/20581_2.png) [@Wesley\_Clark](https://boards.straightdope.com/u/Wesley_Clark)\ **Post date:** [June 15, 2018, 12:51am UTC](https://boards.straightdope.com/t/what-would-change-if-we-had-scalable-quantum-computers/815969/22 "2018-06-15T00:51:56Z")
> [@Jim\_Peebles](#): > > Can you provide a link about a device that actually works to predict roulette? I’m interested too. All I could find so far is websites that are selling the program, so not sure how valid they are. I did find this. The argument isn’t that you can guess where the ball will end up, but you can predict the general area. So you bet on that 1/2 to 1/4 of the board, greatly improving your odds of beating the house. > **[A Mathematician Has Built a Machine That Can Beat The Odds in Roulette](https://www.sciencealert.com/a-physicist-has-built-a-machine-that-can-beat-the-odds-at-roulette)** > > When it comes to casinos, it's no secret that the house always wins. > [@](#): > > The team was able to demonstrate that simply knowing the rate at which the wheel and ball are spinning - before the ball starts bouncing and everything gets random - is enough to skew the odds. > > In fact, by using a system similar to Farmer’s where they recorded each time the ball or wheel passed a certain point, they showed that they could win on average 18 percent of the time - well above the negative 2.7 percent currently expected from a random bet. [Previous page](https://boards.straightdope.com/t/what-would-change-if-we-had-scalable-quantum-computers/815969.md?page=1)