Time Travel, Trump and Compression
This will be my first attempt to blend politics and physics into a single blog article, wish me luck. In case this goes horribly wrong I want to state my intended point at the very top of this article. The lesson I wish to convey is that contrary to popular belief, advances in information technology and computing do not increase our control over or our ability to anticipate and prepare for future events. To make the article interesting I will point to the election of Donald Trump as President of the United States as an example of an event that nobody could or would have predicted two years ago. In fact, even up to the very moment of his election, the media, the polls and the collective impression of the world was that Trump was about to lose the election in a landslide, the value and power associated with predicting who will preside over the worlds largest economy for four years, the vast amount of political analysis, the computer modeling, the polls and the huge amount of social analysis, how could everyone end up so surprised? Even the NASDAQ appears to have been surprised and confused by the election of Trump at the cost of billions of dollars in volatility. The NASDAQ’s behavior after the election suggests that NOBODY in a position to profit from the election predicted that Trump would win. If all of the worlds best and brightest computer analysts and pollsters can’t predict a US Presidential election even a few hours before it is resolved with billions of dollars to profit from correctly anticipating the outcome, what hope is there that any of the things we attempt to use computers to simulate can possibly be correct? How can we pretend that all of our advances in communications, IT and simulation are progress when it appears that the more analytic resources we apply to anticipating a future event the wronger we are about it.
Now the big leap… there is an important physics answer to this question. The answer is that the better we are at predicting the future the more chaotic the future will become. Bear with me on this one. The problem gets to the heart of some very interesting physics ideas such as time travel. Our best modern understanding of physics is that if time travel were possible it would require infinite energy to accomplish and most likely destroy anybody attempting it. When we talk about time travel it’s important to note that traveling forwards in time is easily accomplished. We can simply freeze something for a long time and thaw it in the future to effectively slow it’s passage through time relative to its environment. The science fiction idea of suspended animation is effectively a practical solution for traveling forwards in time. So what we really mean by “time travel” is traveling BACKWARDS in time. Going back in time to a past event takes infinite energy. People are generally enthralled by the science fiction version of time travel because the implications are so difficult and paradoxical to wrap our minds around, however it’s interesting to observe that we actually travel backwards in time on a per-event basis using LESS than infinite energy on a regular basis. We do this kind of time travel all the time. How does that work?
Consider this definition of intelligence:
Intelligence is the ability to use past experience to accurately predict future events. The more complex and distant the future event you can predict, the more intelligent you are.
Implicit in this statement is a fascinating observation about the nature of time travel. Why accurately predict future invents if not to manipulate them? Suppose you are a wildebeest grazing the African savanna and you want to get a drink of water, but every time you go to the watering hole a pride of lions is hanging about waiting to eat you. You watch your wildebeest buddies go for a drink and get eaten, but then you observe that between 3pm and 5pm during the hottest period in the day the lions all go for a nap, so instead of getting eaten you wait until the lions are napping to go for a drink.
This story is conventionally a great treatise on how Darwinian evolution works, but for the purpose of this blog it is also an example of event based time travel. The wildebeest predicted a future event with very high accuracy and altered that future outcome by acting in the present on that insight. Another way of putting this is that the wildebeest expended energy to compute a prediction about the future and used that information in the present to ALTER the future. The consequence of predicting and altering the future prevented the wildebeest from increasing its entropy by DYING leaving only life in the wake of its accurate prediction of a future event. To put it another way we can think of LIFE as a form of computational suspended animation in which we spend energy to predict the future, then use that information to avoid dying. Avoiding death prevents our living bodies from increasing their entropy which is the same thing that freezing something accomplishes. In this analogy FREEZING something accomplishes forward time travel by slowing the rate at which entropy is impacting the frozen object relative to its warmer surroundings. If freezing moves objects forward in time relative to their warmer surroundings then THINKING is an analogy for moving events backwards in time by increasing entropy in the past to anticipate a future event and avoid it.
*I’m going to digress for a moment by adressing another question I know you are all asking yourselves. How does this interpretation of time travel relate to Einsteins theory of relativity in which acceleration slows an objects passage through time relative to its surroundings? Observe that it requires energy to freeze objects relative to their warmer surroundings and it requires energy to accelerate objects relative to their “slower” surroundings. Energy has to be applied to the object being slowed in time. Acceleration and freezing have an interesting relationship to one another, they both involve increasing the entropy of their relative surroundings to slow their local passage through time. That’s not a complete answer but an interesting observation about the relationship between freezing something and accelerating it and how both effectively result in the objects passage though time slowing relative to its surroundings.
Life as we know it is the result of DNA being suspended in liquid water in a state just between freezing and boiling. If we freeze or boil we die… but death can also find us when we fail to predict a deadly future event and avoid it. When that happens we also effectively melt in as much as death results in our entropy increasing dramatically… “life” in this context is a kind of refrigeration technology.
What does all of this have to do with the election of Trump? The more energy we spend analyzing past events in order to predict and manipulate future events the more unpredictable the future becomes as we prevent those “predictable” outcomes from occurring. Let’s go back to our wildebeest analogy. Our smart wildebeest tells the others; “Don’t go to the water hole between 3pm and 5pm and you won’t get eaten!” The other wildebeests learn this pattern and cease to drink except between 3pm and 5pm. Consequently the lions begin to starve until one of the lions realizes what is going on and tells the others; “Hunt between 3pm and 5pm and nap later, that’s when ALL of the wildebeests are around the water hole!” Now that the nap-time secret is out, the survival strategy of only drinking between 3pm and 5pm is gone because the best way to avoid being eaten while getting a drink becomes going to the water hole at random times and the best strategy for lions to catch a wildebeest is to hunt at random times. The life preserving benefit of the energy that was spent calculating the original prediction has been destroyed as the future pattern is consumed to for immediate survival.
In other words any future event that is valuable to our survival is worth devoting some energy to compute and anticipate. The expenditure of that computing energy to identify a future pattern effectively destroys that pattern by moving it backwards in time to provide a survival benefit in the present. This observation about using energy to move “events” backwards in time in order to keep us alive (frozen relative to our surroundings) is reminiscent of an idea from quantum mechanics called the Many Worlds Theory in which all possible quantum events that can occur do occur. The reason the quantum world looks random to us is because we live in only one of a myriad number of possible Universes in which every possible quantum event occurs and we are only conscious of the one we happen to exist in. This theory, although difficult to refute, is unsatisfying in part because of its unsettling implications about the scope of “reality”. The observation that the act of thinking and predicting the future destroys future events leaving only unpredictable chaos may provide another way of explaining why we cannot predict or compute certain quantum properties and why the act of observing a quantum event seems to alter it. The act of observing a quantum event applied energy to destroying its future predictability. We no longer need all possible future events to actually occur to explain the apparent randomness of the world we observe, it can be said that the act of observation or analysis imposes apparent unpredictability on just the things we analyze.
What do all of these thoughts about thought predict about a future world in which we have vastly more powerful computers to analyse vastly more information? In this context we will be spending much more energy to move much more of what is predictable about the future into the present where we can immediately benefit from it at the expense of increasing the unpredictability of the future. By removing predictable patterns from the future and leaving only noise that we cannot anticipate we increase the number of surprising or improbable events that we actually experience later. This doesn’t mean that accurately predicting earthquakes will prevent earthquakes from happening, we’ll just lose the ability to predict the consequences of an earthquake. The over analysis of a presidential election, for example, results in completely WRONG predictions and a shocking implausible outcome. You could say that the media effectively converted all predictable (computable) presidential election outcomes into present advertising dollars for their benefit at the expense of ensuring that the reality of the election would become unpredictable. Another way of saying this is that everything that any large collective of people believe to be true about the future will become FALSE because they will consume all of the value of whatever insight they may have before the future they anticipate arrives. More computing power and faster communication among people accelerates the rate at which people are wrong about the future unless they keep their insights secret.
Where do these strange views on time travel come from? I’m a data compression expert. Successfully compressing data involves the same phenomena. The smarter or better a data codec is, the more information it removes from an input data source. We can measure how efficiently a given codec has compressed data by computing how much entropy the compressed file contains. The better the codec the more random the data in the compressed file will appear to be. For example most people will recognize the first few digits of pi:
A compression codec that recognizes digits of pi may simply reduce all of the data in this sample to the pattern “p” which the decoder would recognize as an indication to simply compute digits of pi from a formula included in the codec to enough precision to match the original input file size. If you did not know this information about the codec and simply opened a .txt file containing the single character “p” you would have no information about what it represented, it would just be a random letter. This is what applying thought (analysis) to predicting the future does to the future, it removes the recognized valuable pattern from actually happening and leaves only noise or unpredictable events to actually occur. In this analogy Trump is elected President of the United States against all apparent odds and against all analysis and prediction. Whose fault is it? Anyone who profited from analyzing the future and predicting it in the present would be guilty of effectively consuming all other more probable future outcomes the same way our wildebeest trained the lions to randomize their hunting patterns by sharing its survival insights with other wildebeests.
How might these observations apply to things we currently believe about the future?
Will AI destroy all jobs? No because better AI renders the future more chaotic to prediction. A noisier future means more menial tasks will appear that computers cannot economically be programmed to perform because of their non-repetitious nature. For example a robot that does a great job of keeping an elderly person company may never be cheaper than hiring a human to do it.
Will we run out of energy and destroy the climate? No because there is a logical limit to the amount of energy it makes economic sense for humans to consume manipulating the future before an equilibrium of diminishing returns is achieved. If you can’t improve your life by manipulating the future beyond a certain point, you won’t spend energy trying. That equilibrium does provide a strong incentive to utilize energy increasingly efficiently to achieve greater prediction gains at the same cost.
Will Donald Trump be the next Hitler? No because everything we have learned about Hitler will prevent that historical event from ever repeating the same way. There will never be another Hitler.
Will there be a singularity in which the rate at which we accumulate knowledge approaches infinity? No because doing so would require consuming infinite energy in exchange for diminishing returns on accurately predicting future events. There is an economic equilibrium that confines how much energy it is worth spending on predicting the future. We can only borrow so much information from the future before the interest rate we pay for it exceeds the insights value.