Removing the Element of Surprise by Aaron M. Feeney

Thanks for explaining so patiently, Aaron. Such a machine is infeasible, but I was wrong to imply that it would always be so (sorry); I can't prove that. I still want to re-read your essay from the vantage of reason as an imperfect past/future viewer (not a predictor, but a fuzzy viewer). I'll do that shortly (on a clear thinking day!) and see what comes out of it. - Warmly, Mike

Aaron,

Your abstract does draw in the reader, dangling a technology before us that has the future steering humanity. Building a foreknowledge machine is perhaps more reasonable than trying to get humankind to work together on a solution that provides survival and a future.

I would wonder how long it will take to build a foreknowledge machine. Will we not destroy ourselves before that happens? Most of us speak of the right approaches and means of steering the future. Yours is ready made -- at least upon discovery.

Jim

Your paper is an interesting topic. Time travel and obtaining information from the future is an interesting subject. I will be honest that I doubt it can become a practical technology, well not in the way most people think. I am bumping the score of your paper back up.

On the plus side with time travel, it can only work in a quantum universe. A closed timelike loop exists only if the quantum wave in the closed path constructively interferes with itself. The grandfather paradox is then "solved" because there is a probability amplitude that I might kill my grandfather and the closed timelike loop can exist according to the proper constructive interference of the closed loop with itself and with commensurate outcomes in the future. This does permit one to perform NP problems in P time. One can prime factorize large numbers in P time. It is done essentially for the same reason one can build a time machine after getting the instructions, where upon you travel into the past and give yourself the instructions.

The big problem with time travel is that the sort of spacetime which permits closed timelike curves (time travel) violates the Hawking-Penrose energy condition. The energy part of the stress-energy condition is such that T^{00} < 0. This results in some funny problems. In particular the quantum field theory for the mass-energy source of this spacetime has no minimum level. This means there is an infinite number of states the field can drop down and in doing so emit an infinite number of radiation quanta. That is really bad, and an argument for why this sort of thing does not exist.

In spite of this closed timelike loops exist in the interior of black holes. A Kerr-Newman black hole has a timelike interior bounded by the interior horizon that is a Cauchy horizon and a singularity with closed timelike curves that orbit it. This permits a range of hypercomputations, or second order λ-calculus. Problems that are undecidable by Turing-Godel theorems are solvable! Again the wave function constructively interferes with itself to give the output. Quantum fields interior to the BH can be entangled with fields outside. Consequently it is possible to solve hard problems with black holes.

You might be interested in my paper . I address this issue from a different perspective.

Cheers LC

Copied from my essay blog site

The (Closed Timelike Curve) CTC quantum paths I discussed yesterday are not something I entirely understand. It is I think likely that these CTCs are entirely contained inside event horizons of black holes. I doubt these exist in spacetime we causally interact with. Of course I might be wrong, but time travel seems to pose troubles that I doubt exist.

Though CTCs exist inside black holes quantum states in the exterior that are entangled with interior states will serve as a quantum computer. The closed paths interior to black holes will permit second order λ-calculus and computation of NP hard problems in P time. States exterior to the black hole will compute them as well. Of course there is a bit of a problem, for while this entanglement exists to properly read the output a classical key must be transmitted from the interior. That is a difficult question. The black hole quantum mechanically decays and this interior key does in effect escape. Whether or not the quantum information that escapes can be "used" as the key is unknown.

Of course from a practical perspective I doubt that black holes will be used as quantum computers any time soon. Quark-gluon plasmas have quantum gravity amplitudes in the quantum gravity - quantum chromodynamics correspondence, and this physics may influence the types of scattering processes involved. So it is not likely this will be a practical technology.

If you want to send material to me my address is lcrowell@swcp.com

Cheers LC

Hello Aaron - As promised, I read your essay and here is my comment. Score will follow. I agree with the comments above - your essay is fun to read and easy enough to follow. While there is the feasibility issue, and also the technical one of non-predictability, I'm also intrigued by the sociological issue. This is not necessarily a new point of discussion, since physical determinism and divine omniscience have similar attributes to the foreknowledge machine and they have been around for a long time. (See my blogpost on this topic: Time and Free Will. So here is the question - would a truly rational human, having in hand one of your foreknowledge machines, be inclined to give up the human struggle to make his or her own choices? Would they still be human?

Much thanks and best of luck! - George Gantz (author: The Tip of the Spear)

Hi Aaron,

The concept of time-viewing is interesting! I like how it interacts with images from distant places/times. I went into the essay skeptical, since I don't think travelling back in time is likely to be possible, but I like your conceptual approach to "foreknowledge machines", which could just as well be implemented as very powerful computers.

It is interesting to think that, assuming that humans wise enough to discriminate between desirable and undesirable outcomes are in charge, desirable outcomes would more often be fixed-points under foreknowledge, and that undesirable outcomes would more often not be fixed points, so whenever a foreknowledge machine did return a value, it would be more likely that we would see (and end up with) a desirable outcome. Again, I wasn't initially sure that this would really work, but it does seem right on closer inspection.

My only complaint is my first one: my limited understanding of physics is that sending information backward in time isn't a reasonable possibility. We could build prediction machines of some kind, but making them powerful enough to serve as foreknowledge machines with long-range abilities doesn't sound like a likely possibility (or, if we could, then we could also build other machines that would make a much better difference).

Overall, though, interesting essay!

Let me tell you a story.

A Foreknowledge machine was built in a secret location and its existence announced to a stunned world. Suitably white-haired and bespectacled wise men were trotted out in front of TV cameras to explain how powerful security protocols and AI censors had been put in place to only allow the prediction (or non-confirmation) of positive outcomes. Politicians backed up their technical assurances with organization charts detailing an intricate oversight bureaucracy. Top military brass explained that the technology was classified to prevent would-be terrorists from using it to predict successful attacks. As further assurance against the possibility of abuse, the wise men pointed out that only a superpower could afford the astronomical amounts of energy needed to create significant CTCs. No would-be terrorist would be able to run one of these things in his cellar, even if he somehow managed to get his hands on the blueprints.

In a gesture of international good will, several very positive predictions pertaining to other countries, including historically antagonistic ones, were presented, ranging from six months to five years into the future. A sceptical and, initially, even frightened world resolved to check the outcome of the first batch of predictions half a year down the road.

They all came true.

Then the one-year predictions were fully confirmed. Then the two-year ones. By then, nobody was even holding out for the five-year check any longer; they were all too busy dreaming up wonderful predictions to submit for consideration by the rapidly swelling oversight bureaucracy. The power to select predictions for confirmation by the Machine quickly became the most sought-after perk among politicians of all stripes. And for good reason: swelling Machine-related government "business" and ensuing scandals aside, all was not only well, but better than ever. Time and again, wonderful visions were submitted to the Machine and confirmed to be part of the future, setting the plans in motion which, inevitably, led to their realization. Humanity, so it seemed, had entered a true Golden Age.

Until one day, the predictions stopped.

Not because the Machine stopped. Not because of technical problems. Everything was in order, the giant staff dedicated to operating the Machine made sure of that. They checked and rechecked absolutely everything over and over again. It was just that every single scenario submitted to the Machine was now being returned as not confirmed.

Initially they shrugged it off. A fluke, they said. Given the amount of scenarios submitted, a bad run was bound to happen at some point. Soon everything would be back to normal.

But it didn't. As time dragged on, old predictions already in the pipeline kept coming true like clockwork. Six months, one year, two years. But all new submissions came back as not confirmed.

After three years there was no denying it any longer. Something was terribly wrong, not with the Machine, but with the future. A probe was initiated to find out exactly where. A sequence of simple predictions so trivial they would never have made it through the usual selection process was submitted, spaced in time according to a simple golden mean algorithm. The result was incontrovertible: all predictions, even the most trivial, failed after The Date. Although they certainly tried, this was something not even the mighty security apparatus surrounding the Machine could keep under wraps. Soon the whole world was talking of nothing but The Date.

The Date. The day beyond which nothing good would happen. For all practical purposes, the end of the world.

Religion bounced back from an all-time low to become more popular than ever. Preachers were the new rock stars. Scientists appeared in heated televised debates on how the world would end, waving their new popular tomes, fresh off the printing presses, at each other. Asteroid impact, one said. Black hole, countered another. Alien invasion, opined a third. Antibiotic-resistant, airborne pandemic, suggested a particularly pedestrian one. Financial markets crashed.

When paparazzi caught a giggling President lying on his back on the White House lawn, dressed in an old "No Future" T-shirt and smoking a haphazardly hand-rolled cigarette, it was all over. People stopped going to work, then stopped leaving their houses altogether. Store shelves gaped empty as the intricate supply chain which had supported industrial civilization collapsed. Gangs formed, looted what was left to loot, but mostly concentrated on killing each other off. Suicide rates shot through the roof.

When The Date finally rolled along, there was nobody left to notice.

Ten thousand years later, give or take a few millennia, an alien probe flying by the solar system snapped a sequence of photos showing the ruins of Earth's giant cities and the still perfectly preserved Machine on the far side of the moon. The probe's AI was only sophisticated enough to recognize and document the remnants of a planetary-scale technological society, but the electronic archivist at the receiving end of its data stream had no problem recognizing the signs. It had seen the same pattern many times before.

Earth was filed under "Killed by self-fulfilling prophecy".

Dear Aron,

I like your essay, it is thought-provoking and also fun reading. I never had this point of view but why not. I like also the 3 types of machines.

My own work (exotic smoothness and cosmology) uncovered the mysteries of CTC's, I was unable to avoid them in my theory, but maybe you point to an interesting solution where I have to think about.

In particular you carefully discuss the problems of time travel (also possible avoidance of paradoxa).

For more comments I have to dig deeper.

Best Torsten

PS: I like it so much to give score 8!

Hi Aaron,

I have read your essay for a second time.

The kind of foreknowledge machine that you have said can be made just shows all of the possible outcomes. Isn't that like a chess player can look ahead and compare the outcomes of different moves by visualization. Perhaps our decision making/steering, is reliant upon us not weighing up myriad possibilities but only considering a limited range of possibilities. Important if urgent decisions are needed or the consequences are of little importance whatever choice is made.

Is the machine's accuracy outweighed by the information overload as all possibilities are given? Or is its value in allowing us not to think about consequences of actions but to allow a machine to do that for us?Isn't thinking and even making mistakes part of what makes us human? For usefulness there would need to be probabilities assigned to each foreseen outcome. That may be useful for accurate risk assessments. (Just me thinking aloud.)

There is a good example of the 'Cassandra type' machine in the film 'Paycheck', writer Philip K Dick, directed John Woo, 2003 with dire consequences of self fulfilling prophesy writ large. I don't believe in time travel, it is incompatible with my way of thinking about the universe and physics.Interesting to hear you views and the references to work overcoming the temporal paradoxes.

Regards Georgina. I will be voting when I am ready to do so. I'd like time to digest all that I have read.

Hi. No, not really a great deal of thought, I typed it out immediately after reading your essay.

Regarding your objections, the predictions in this little story are not produced by the machine. They are produced by submitters who want to have them verified by the Machine, typically as part of a scenario which may require the verification of multiple predictions. A simple example given in your essay is the prediction that an airplane will arrive at destination according to schedule.

So no, I do not agree that there are "clear differences".

The serious point here is that your proposal for taming the potential risks of the hypothetical Foreknowledge machine is incomplete. If the machine consistently fails to provide foreknowledge about an exhaustive range of desired outcomes, you will know that something undesired will happen, even if you do not try to directly obtain foreknowledge of the undesired outcome.

My little story is about a particularly simple case in which all foreknowledge proves unobtainable beyond a certain date, thus identifying that date as The End, ironically due to a sequence of events caused by the very existence of the Foreknowledge machine. It's written in jest, but it exploits a real hole in your argument.

Very interesting essay, Aaron. I think your ideas are valuable, although I have my doubts about the feasibility and value of a foreknowledge machine. A few thoughts:

1) I don't think a Cassandra machine is possible either, but I don't think you demonstrate this logically. Your thought experiment seems to me to assume that the machine that executes the program on the basis of the Cassandra machines input cannot go wrong. But I don't think you can assume that any physical machine is perfect.

2) I didn't understand your explanation of why we see/remember the past but not the future. I don't think the speed of light is the issue, since the laws of physics seem to be CPT-invariant (essentially, should run the same forward as backward).

3) I would add, however, that we probably can't have foreknowledge of events--like a wave of radiation from a gamma-ray burst--traveling toward us at light speed.

4) Even if foreknowledge machines are logically possible, they may not be physically possible. The obstacles may not be philosophical, but physical. And even if there is some way to observe the future from the present, we may lose a lot of resolution at any distance through time.

5) I found the idea that foreknowledge machines would end war intriguing. You are right that countries that go to war generally (but not always) do so because they disagree about what the outcome will be. But it seems to me--perhaps I have misunderstood what you are trying to say here--that outcomes that depend on the human use of information about the future are precisely the kind that a foreknowledge machine couldn't foresee.

6) Even if a foreknowledge machine were possible, I would be very surprised if this was something we could develop in 20 or 30 years.

7) I agree that better predictive technology would be valuable. But machines that are occasionally wrong are still extremely useful if their error rates are low. We don't necessarily need theoretically perfect foresight.

In any case, I enjoyed reading your essay. I'm rating it now. Good luck in the contest!

Best,

Robert

Aaron, I think Tommy's saying we could discover the time of a future catastrophy, or one that's merely apparent. Suppose the machine shows all flights landing safely (as you say), but only till 9:13 p.m. tomorrow. Thereafter it returns no view for any flight in the world, nor any other class of future outcome. Naturally we infer a global catastrophy tomorrow night. That inference might be wrong, but it frightens people and causes them to act in dangerous ways, perhaps with catastrophic consequences. Hence the diagnosis, "Killed by self-fulfilling prophecy". - Mike