You have a Time Displacement Teleportation Device (or two). Your computer is making split-second decisions, & it knows an outcome or event 50µs from now and can correct decisions. You now control something; 50/50 decisions are now 100% success for you. With two time pumps 1:3 odds become 100% assuming both computers are given the same three choices with identical outcomes. And so on. In fact your odds have at least increased for every unknown decision. This easily turns into profit, but not at a casino or lottery - not enough time for a human response compared to this question. It’s a split-second decision.

These refrigerator-sized time pumps are commercially available and are changing the world with inexhaustible energy. But, you just “repurposed” a time pump to teleport your computer information back to itself in the very near past, so it can make a precognitive decision. This decision will greatly change your near-future, middle class average life, and you can do it repeatedly but not iteratively (you can’t send it back to the same decision moment more than once). Also, your modified calibration of the pump increased the jump time, but unfortunately only allows you to use it 3 times per day within a 30 second window, at a time determined by the position of the earth that day.

You start thinking about what important information you can know in 50µs:

  • can it guess a password or PIN? No.

  • Can it make a go/no-go decision based on an immediate outcome?

  • Can it hack a multiple choice computer test? Maybe but not online. Network lag is a big problem.

  • Can it place a winning stock trade? Unlikely.

  • Can it bet on an online game? Network again!

  • Can it improve Deep Learning algorithms with minute precognition?

Q: What split-second outcomes did the computer predict to change your life?


It won't help you for anything outside of local events.

Light travels at 1 foot per nanosecond so converting to microseconds that limits you to events that occur within 50,000 feet (15 km). Anything further away and you may know about it early, but your response to it won't be faster than actually being there. Besides this, you will have to deal with network lag and so on, so unless your computer is in a wall street colocation center with other servers, the lag is simply too much.

Suggestion: Change the question to 50 milliseconds and it may be much more useful to you.

  • $\begingroup$ I mentioned the network problem in the question. $\endgroup$
    – Vogon Poet
    Oct 23 '19 at 21:21
  • $\begingroup$ It can’t work at 50 ms, the information would already be over a mile away by that time. As the information travels back in time it also travels through space at 85,000 mph (124 feet per millisecond). Because the information is briefly frozen in spacetime but earth is not. $\endgroup$
    – Vogon Poet
    Oct 23 '19 at 21:32

I'm not a quantum physicist, but I wonder if it would cause a superposition of states to prematurely collapse.

Maybe that would let you predict or change the output of a quantum computer that was being used to crack conventional cryptography keys. Maybe that would give you enough time to transfer money out of the bank account the key was protecting.

Or maybe you can save Schrodinger's Cat and claim the reward money (although I doubt there's a reward as Schrodinger put the cat in the box himself).

  • $\begingroup$ Lol I did think of the poor cat when I conceived it. $\endgroup$
    – Vogon Poet
    Oct 23 '19 at 22:51

Rather than using it to directly trade a market, say Gold Futures (see what I did there?), you could instead use it to train an artificial intelligence to trade Gold Futures. Deep Immediate Future Prediction Reinforcement Learning or something.

To put another way, you have 3 options in futures. Buy, sell, hold. The benifit of your machine is that it can explore the future of these three options. Constantly. These become data points. So a trade you made 1 second ago has that many microsends of data points. RL can be designed to look at these data points and realize the place in the time line that made subsequent actions valuable, so when combined with other market data, your 50 microsecond advantage is leveraged into several hours or days of advantage

For further explanation, let's say a single second passed and you have a ton of observations that ended up 50 percent buy, 50 percent sell. Okay not much there. But if 98% of your immediate observations are buy, your reinforcement learning agent knows that's a good idea. Now it monitors the market from the future to make a decision on when to end the trade. And the cool part is that you don't have to understand the relationship of the predictions for it to work. It could be 1percent sell, and it will sell, because it learned a counterintuitive quirk of the market by literally predicting the future

  • $\begingroup$ Oooooooh! A new shiny toy! $\endgroup$
    – Vogon Poet
    Oct 23 '19 at 22:24
  • 1
    $\begingroup$ How is this any different than mining past historical data? $\endgroup$ Oct 23 '19 at 22:25
  • $\begingroup$ Now... I agree. How? Do futures respond in microseconds? But DEEP LEARNING is actually an application here. $\endgroup$
    – Vogon Poet
    Oct 23 '19 at 22:25
  • $\begingroup$ Training the ai, It doesn't make a difference, however with the future prediction as an input, among other data (volume, spread etc) it can learn how to make longer term decisions from milisecond data. In otherwords, if a 'sell' decision is right 50microseconds from now, it is also likely to be correct 50ms from now. Then it just learns the combination of other inputs that make it profitable. Also, when to close a trade is just as important, so it can predict that as well $\endgroup$
    – krflol
    Oct 23 '19 at 22:40
  • $\begingroup$ OK my head is hurting. If you can actually detail this out in your answer I am intrigued. It will be able to respond to trades only 50us in the future, how does this improve learning? (I really want to know) $\endgroup$
    – Vogon Poet
    Oct 24 '19 at 0:05

Not the answer you're looking for? Browse other questions tagged or ask your own question.