Roboticists Discover Alternative Physics (phys.org) 95
An anonymous reader quotes a report from Phys.Org: Energy, mass, velocity. These three variables make up Einstein's iconic equation E=MC2. But how did Einstein know about these concepts in the first place? A precursor step to understanding physics is identifying relevant variables. Without the concept of energy, mass, and velocity, not even Einstein could discover relativity. But can such variables be discovered automatically? Doing so could greatly accelerate scientific discovery. This is the question that researchers at Columbia Engineering posed to a new AI program. The program was designed to observe physical phenomena through a video camera, then try to search for the minimal set of fundamental variables that fully describe the observed dynamics. The study was published on July 25 in Nature Computational Science.
The researchers began by feeding the system raw video footage of phenomena for which they already knew the answer. For example, they fed a video of a swinging double pendulum known to have exactly four "state variables" -- the angle and angular velocity of each of the two arms. After a few hours of analysis, the AI produced the answer: 4.7. The researchers then proceeded to visualize the actual variables that the program identified. Extracting the variables themselves was not easy, since the program cannot describe them in any intuitive way that would be understandable to humans. After some probing, it appeared that two of the variables the program chose loosely corresponded to the angles of the arms, but the other two remain a mystery. "We tried correlating the other variables with anything and everything we could think of: angular and linear velocities, kinetic and potential energy, and various combinations of known quantities," explained Boyuan Chen Ph.D., now an assistant professor at Duke University, who led the work. "But nothing seemed to match perfectly." The team was confident that the AI had found a valid set of four variables, since it was making good predictions, "but we don't yet understand the mathematical language it is speaking," he explained.
After validating a number of other physical systems with known solutions, the researchers fed videos of systems for which they did not know the explicit answer. The first videos featured an "air dancer" undulating in front of a local used car lot. After a few hours of analysis, the program returned eight variables. A video of a lava lamp also produced eight variables. They then fed a video clip of flames from a holiday fireplace loop, and the program returned 24 variables. A particularly interesting question was whether the set of variable was unique for every system, or whether a different set was produced each time the program was restarted. "I always wondered, if we ever met an intelligent alien race, would they have discovered the same physics laws as we have, or might they describe the universe in a different way?" said Hod Lipson, director of the Creative Machines Lab in the Department of Mechanical Engineering.
"Perhaps some phenomena seem enigmatically complex because we are trying to understand them using the wrong set of variables. In the experiments, the number of variables was the same each time the AI restarted, but the specific variables were different each time. So yes, there are alternative ways to describe the universe and it is quite possible that our choices aren't perfect."
The researchers began by feeding the system raw video footage of phenomena for which they already knew the answer. For example, they fed a video of a swinging double pendulum known to have exactly four "state variables" -- the angle and angular velocity of each of the two arms. After a few hours of analysis, the AI produced the answer: 4.7. The researchers then proceeded to visualize the actual variables that the program identified. Extracting the variables themselves was not easy, since the program cannot describe them in any intuitive way that would be understandable to humans. After some probing, it appeared that two of the variables the program chose loosely corresponded to the angles of the arms, but the other two remain a mystery. "We tried correlating the other variables with anything and everything we could think of: angular and linear velocities, kinetic and potential energy, and various combinations of known quantities," explained Boyuan Chen Ph.D., now an assistant professor at Duke University, who led the work. "But nothing seemed to match perfectly." The team was confident that the AI had found a valid set of four variables, since it was making good predictions, "but we don't yet understand the mathematical language it is speaking," he explained.
After validating a number of other physical systems with known solutions, the researchers fed videos of systems for which they did not know the explicit answer. The first videos featured an "air dancer" undulating in front of a local used car lot. After a few hours of analysis, the program returned eight variables. A video of a lava lamp also produced eight variables. They then fed a video clip of flames from a holiday fireplace loop, and the program returned 24 variables. A particularly interesting question was whether the set of variable was unique for every system, or whether a different set was produced each time the program was restarted. "I always wondered, if we ever met an intelligent alien race, would they have discovered the same physics laws as we have, or might they describe the universe in a different way?" said Hod Lipson, director of the Creative Machines Lab in the Department of Mechanical Engineering.
"Perhaps some phenomena seem enigmatically complex because we are trying to understand them using the wrong set of variables. In the experiments, the number of variables was the same each time the AI restarted, but the specific variables were different each time. So yes, there are alternative ways to describe the universe and it is quite possible that our choices aren't perfect."
Re: (Score:2)
With mass and velocity you can figure out an object's momentum [9gag.com]. That can come in handy at times.
Only a camera? (Score:5, Interesting)
This software ain't discovering any laws of physics. We all know how relying solely on visual input can be deceiving.
All it's doing is mapping out the motion into minimal parameters. Cool in itself. Probably useful for movie making.
Re: Only a camera? (Score:3)
Exactly, it could be quantifying the speculation reflection, or a shadow, or the hue of a particular pixel for all that it matters for predictive behavior.
Re: Only a camera? (Score:5, Insightful)
This. While we can assume those scientists did their homework, they need to reproduce those results under different lighting, POVs, FOVs, etc. to ensure that it's not something silly like "size of shadow" that only in these specific conditions was "discovered" as a shortcut to a different measurement.
Re: Only a camera? (Score:5, Interesting)
To my eye, this seems far more like how humans learn to interact with the world as infants. They notice patterns, develop mental models, and then test them. Over time refining hyper specific models into more generalizable ones. They may not be 100% accurate, but they are "good enough" to make reasonable predictions about things like where a ball in flight will end up, or how far away an object is. Just as a pro ball player may be completely unable to describe the trajectory of a baseball in mathematical terms you and I can follow, he still "Knows" where the ball will go with surprisingly high precision. Just as this AI may "know" the next motion of a pendulum without being able to describe it in mathematical terms.
Re: (Score:2)
They may not be 100% accurate, but they are "good enough" to make reasonable predictions about things like where a ball in flight will end up, or how far away an object is.
Indeed. A classic example is the approximation of a thrown ball being modeled by a parabola, when in fact the path actually travels along an elliptical path, not a parabolic one.
Re: (Score:3)
Indeed. A classic example is the approximation of a thrown ball being modeled by a parabola, when in fact the path actually travels along an elliptical path, not a parabolic one.
Not in this atmosphere it doesn’t.
Re: (Score:2)
Re: (Score:1)
This software ain't discovering any laws of physics. We all know how relying solely on visual input can be deceiving.
You may think so, but ALL physics works that way. Neils Bohr said,
"It is wrong to think that the task of physics is to find out how Nature is. Physics concerns what we can say about Nature.”
We can only ever talk about that which we can observe. This is a subtle but important distinction. See PBS Spacetime:
https://youtube.com/watch?v=v-... [youtube.com]
Re: (Score:2)
That's a somewhat silly observation. Science in general is concerned with why things happen the way they do. Being able to state it is akin to going to a restaurant, reading the menu, and then claiming you can taste their food.
Re:Only a camera? (Score:5, Insightful)
A more complete version of the quote is:
There is no quantum world. There is only an abstract quantum physical description. It is wrong to think that the task of physics is to find out how nature is. Physics concerns what we can say about Nature.
Bohr himself wasn't saying that science was only done on the foundation of what you can observe.
He was speaking to what him and Heisenberg considered a fundamental truth of quantum mechanics- that what we can empirically measure is fundamentally multiple steps away from whatever reality is.
After all, Bohr's own discoveries regarding the atom were not observable, or even differentiable from orbital models (at the time) except by deductive reasoning.
We can only ever talk about that which we can observe. This is a subtle but important distinction. See PBS Spacetime:
Ugh, no.
That is not what Matt is saying.
We can absolutely talk about what we can infer about the universe based on the observed success or non-success of models we design.
That is to say, we can most certainly talk about that which we have not, and can not observe.
What person you are replying to is saying, is that observation alone, without the inference can lead to flatly incorrect deductions of what you're looking at.
An example would be pseudoforces like gravity and the coriolis force, where what you observe may only be valid in your particular frame of reference.
If the implication is that whatever future aliens we may encounter are not intelligent enough to deduce that the bizarre rotation-ward force experienced by a free-traveling object is actually just a consequence of you using a flawed coordinate system, then I don't think we have to worry about communicating with them, because they're completely stumped by the relativistic effects experienced at orbital velocities and altitudes.
Re: (Score:1)
Well, that probably *is* what he was saying, but I think it plausible that he's wrong.
We don't *know* that we're operating with a minimal set of assumptions. We don't *know* that our assumptions are consistent. And we don't *know* that there aren't other assumptions that would be equally good (or better). These are things we believe, and have reasonably good evidence for, but that's several steps away from certainty.
The basic value of science, all science, is the ability to make valid predictions. Once
Re: (Score:2)
Well, that probably *is* what he was saying, but I think it plausible that he's wrong.
Na. He's faithfully presenting the argument he's educating about.
The basic value of science, all science, is the ability to make valid predictions. Once you can make valid predictions, then you can act on them to do things. This is why it's essential that scientific theories be testable. If it's not testable (at least in principle) then it's not science.
You are conflating the scientific method with science.
An inference that is not testable can still be scientifically applicable.
Were this not the case, the Heisenberg Uncertainty Principle would not be science, when indeed- science cannot even take place at certain levels without it.
The scientific method requires testability. Science includes inferences made from those tests.
Re: (Score:2)
Nothing can be perfectly tested, but the Heisenberg Uncertainty Principle passed LOTS of tests. It may still be wrong, or at least an oversimplification, but it's been tested. How Bose-Einstein condensates react are tests of it. So is much of electronics. Etc.
(Actually, the way it's normally stated in English, at least, is definitely wrong. But that's not what the prediction actually says. The closer you get to what it actually says, the more you need to say it in mathematics. Which means I no longer
Re: (Score:3)
Others have addressed that there's more nuance, but even at face value, we have employed far more than a singular vision in a small frame of reference. So even to the extent we are limited by working within the limits of our observations, our observations are *way* more comprehensive than those used to cough up some variables to describe how to predict frames of a video based on how the video was before it.
If someone familiar with video codecs studied the result, they'd likely have a better chance at under
Re: Only a camera? (Score:2)
some circumstances
Right.
Newton actually stated his second law correctly. Force equals the change in momentum. But for us slowpokes, change in velocity is close enough. But as one pushes an object faster, more of the momentum change goes into the mass term.
I'm not dure if Newton had relativity in mind. But he knew how not to make the simplifying assumptions that others have lived with. And Einstein had to backtrack through.
Re:Only a camera? (Score:5, Insightful)
Yep.
For all we know it actually considered the angle of the camera.
Basically, AI is not magic. It's not learning something we do not already know, it's just discovering efficient ways of discovering things, which given the chance, AI will *AWAYS* take a shortcut.
In NLP AI's, they do that, a lot. Have an audio corpus? All recorded with the same hardware? chances are it learned that background noise or echo's in the recording are variables, or are part of the sound. So if you have it read it out as TTS, it will replicate the noise, or even amplify it. So you need to have a corpus of the same subject, but with different conditions (eg a different microphone, a different room) for it to learn that those acoustic properties are not part of the voice.
So what the roboticists here should have been doing is running the same tests with different "unimportant" conditions, different lighting, different cameras, different lens, etc. That would likely give the expected 4.0 variables.
Re: (Score:2)
"mapping out the motion into minimal parameters"
Mapping observations into minimal parameters is the definition of a scientific law.
Re: Only a camera? (Score:3)
If you have made a sufficient number of observations.
One is a prime number. Three is a prime number. Five is a prime number. Seven is a prime number.
Therefore all odd numbers are prime.
Re: (Score:3)
If you have made a sufficient number of observations.
One is a prime number. Three is a prime number. Five is a prime number. Seven is a prime number.
Therefore all odd numbers are prime.
First of all, one is not a prime. Look it up.
Second, mathematical proofs are completely separate from generating physical laws based on observation. We never prove anything in science, just accept items based on overwhelming evidence, including multiple observations AND ability to predict events.
Re: (Score:1)
We never prove anything in science
Good to know for my next argument with an AGW crybaby.
terminology (Score:5, Insightful)
Re:terminology (Score:5, Interesting)
However doing the math is not always easy using those variables. So it not uncommon to transform the variables into an easier problem. A good example is solving electrical circuits with phasors. Or QM using probability. When a solution is realized the variables are transformed back into physical space. The challenge is proving these transformations are equivalent, as Feynman did with the two prevailing formulations of QM
Discoveries like this are potentially useful as they may point to more efficient algorithms for solutions using modern computational techniques. The focus up to now has been efficient hand calculations or iterative solutions. If the analysis has actually created an more efficient transformation this could be very interesting.
Re: terminology (Score:2)
Re: (Score:3)
Agreed.
In the example of the two pendulums, TFS states that typical physics would distill this to 4 parameters (two positions + two velocities). (There are other models in "conventional" physics that use different parameters, like potential and kinetic energy, and represent them with Lagragian or Hamiltonian equations. These would still re
Re: (Score:2)
"I'm not sure how you get a fraction of a countable item, but never mind."
That means that you have variables that are partially correlated.
Re: (Score:3)
This is mainly a problem with whoever wrote the title of the phys.org article. In the NatComputeSci paper they correctly say "physical laws".
Postulates and Lagrangians (Score:5, Interesting)
Without the concept of energy, mass, and velocity, not even Einstein could discover relativity.
All of special relativity is dependent on two core postulates: that the laws of physics are the same in all inertial frames and that the speed of light is a universal constant and independent of the motion of the source or observer. You need a concept of velocity for that but neither mass nor energy. It's true that relativity has implications for energy and mass but those are derived from those core postulates which is all Einstein needed to come up with the idea.
If they are interested in using different variables to describe a system they should first have had a look a Lagrangian or Hamiltonian mechanics first since these descriptions are agnostic to the choice of variables.
Re: (Score:2)
It's not a winning proposition to try to educate Dunning-Kruger sufferers.
Re: (Score:2)
I know little about physics beyond the high-school physics. But the above sounds to me "if speed of light is a universal constant and ... then ....". I.e. it is an assumption that "speed of light is an universal constant". We also know that "assumption is a mother of all fuckup
Re: (Score:2)
Lagrangian mechanics (Score:4, Interesting)
> "I always wondered, if we ever met an intelligent alien race, would they have discovered the same physics laws as we have, or might they describe the universe in a different way?"
There are a number of DIFFERENT known, useful sets of laws of physics we already use, systems separate and different from the Newtonian-Relativity-Quantum model. Not just looking at different scales like Newton vs quantum, but describing macro effects using a totally different set of laws.
One example is Lagrangian mechanics. It has one fundamental law - nature is lazy. Objects always take the easiest path. This is known as "the principle of least action". Using Lagrangian mechanics, you can predict actions just like you can with Newtonian physics, but using different principles from a different viewpoint.
I see no reason to think that IF there were other intelligent life out there, they would pick Newton's laws. There are plenty of other options.
I wouldn't be at all surprised of they understood the more fundamental principles that cause or underly both Newton and quantum mechanics, seeing each as a subset or special case of the general principles.
An interesting sidenote - Lagrangian mechanics may have given us a way to finally model / predict the performance of different wing designs. For as long as we've been flying and using wind tunnels and such, we have to find the equations that define a wing's performance. Lagrangian mechanics may have recently solved that.
Re:Lagrangian mechanics (Score:5, Insightful)
I am a bit concerened with all this talk about new physics. This is all (Lagrangian Mechanics, and the AI-based interpretation in the featured article) classical physics, being described by people in slightly different terms that one might encounter in a first-year physics class.
Re: Lagrangian mechanics (Score:5, Insightful)
You are right to be concerned. Slashdot is becoming riddled with people who clearly do not understand basic math or physics, are obsessed with bizarre conspiracy theories and extremist left and right politics, enamored with anything that sounds remotely like science fiction. They take positions with understanding, they respond to the bias and premises of the article without questioning the premises or biases.
Very worrying.
Re: (Score:3)
Re: (Score:2, Insightful)
They didn't find anything (Score:5, Interesting)
I've ran many algorithms in a process known as system identification which identifies States in physical systems. The process is much simpler than using AI, and you can use it to determine how many state variables are in a system and then model the physical system.
A simple physical system that you can model is a heater, and you measure the heater with a temperature sensor. A simple model would require only two states to model the physical system, you can model that same system with however many states you want and the math will still give you an answer, the answer won't be correct but it will still give you an answer because you know that there are only two dominant states to the system.
If you model the states with a higher order then the physical system has the math works but what you end up is the system either doesn't get modeled correctly or you end up with two states that do the same thing as one state essentially.
Mathematically the AI is a black box and a neural net is not too different from using a deterministic system identification as both of these mathematical processes are simply modeling the states of whatever data you feed into them.
What system identification you can use what's called single value position, to determine how a state variables you need to actually model your system. These crazy researchers should also use a similar method, instead of chalking it up to new physics.
The Question (Score:5, Funny)
After a few hours of analysis, the AI produced the answer: 4.7.
The roboticists are currently building an even bigger AI in order to determine the question.
(or is the question hidden in TFS?)
Re: (Score:2)
I see what you did there, apparently people with modpoints didn't though.
Re: (Score:2)
And any fule no that the answer is 42!
Re: (Score:2)
dang it, beat me to it
Re: (Score:2)
AI == Artificial Intuition (Score:4, Insightful)
An intelligence would be able to explain their process, how they arrived at their conclusion.
What we're working with today isn't artificial intelligence, but rather artificial intuition. Think about it; you can sometimes intuit the answer. You can't exactly explain how you got there, but you did.
So until these models can explain themselves, let's stop using the term artificial intelligence and instead use artificial intuition. Bonus; we don't have to change any of the marketing material!
Re: (Score:1)
The "variables" may merely model the specific test samples and are not necessarily generic.
Re: (Score:3)
What we're working with today isn't artificial intelligence....
So true. What we have is what we used to categorize as an expert system, before the world hopped on the AI bandwagon. We don't have AI, never have, and likely never will. We have systems that, when programmed to process large data sets, can produce very specific answers to very specific questions very quickly. Many of those answers are nonsensical, while some of them are meaningful. It still requires a human to interpret the results and to give them meaning.
Re: (Score:1)
Rather than eternally battle over the definition, let's just agree there are different kinds and levels of AI.
Re: (Score:2)
What we're working with today isn't artificial intelligence....
So true. What we have is what we used to categorize as an expert system, before the world hopped on the AI bandwagon. We don't have AI, never have, and likely never will. We have systems that, when programmed to process large data sets, can produce very specific answers to very specific questions very quickly. Many of those answers are nonsensical, while some of them are meaningful. It still requires a human to interpret the results and to give them meaning.
Precisely. What many people seem to fail to understand is that every living thing is an expert system.
That doesn't mean it's an Intelligence.
Spiders are biological drones running firmware that is an expert system at identifying good locations for webs and then constructing webs. 99.99999% of human beings emerge from the womb with intellectual functions already many orders of magnitude greater than a spider. And yet newborn humans can't construct webs or climb walls, much less beat the original Deep Blue at
Discovery makes a prediction, AI nope (Score:2)
The issue is; not understanding how the AI "thinks", and this is not unlike understanding how a human thinks. At least humans have developed theories with predictions of what data will be measured.
Explain yourself! (Score:2)
Sounds like a job for the SETI folks.
Article summary: (Score:2, Troll)
They need a better Al (Score:3)
> After a few hours of analysis, the AI produced the answer: 4.7.
Everybody knows that the answer is 42, although you have to wait a bit more to get it.
Re: (Score:1)
> After a few hours of analysis, the AI produced the answer: 4.7.
Everybody knows that the answer is 42, although you have to wait a bit more to get it.
Just reverse feed 42 into it and see what happens
Re: (Score:2)
That's wrong!!! (Score:2)
Everybody knows that the answer is 42!
Re: (Score:2)
Totally, they should have re-evaluated their hypothesis at that point.
Conformal mapping and multivariate analysis (Score:2)
Re:Conformal mapping and multivariate analysis (Score:5, Informative)
LOL, I read the paper and found:
1) The authors do not claim to have found alternative physics. The phys.org article does, and confirms that popular journalism of science often gets it wrong. The paper claims to have found alternative state variables, which just means they are using basis functions to reduce data.
2) What the authors actually did was PCA, as I mentioned above. It is a simple way to find orthogonal basis vectors in complex datasets and has been used for decades. For example, I have used it for data reduction for over 20 years. At the end of the day, I don't find this paper interesting because I would expect that PCA would be able to reduce such a problem. It is not surprising.
Re: (Score:2)
I don't find this paper interesting because I would expect that PCA would be able to reduce such a problem. It is not surprising.
It's the same hype as we had with software patents. Take an old thing, tack "with software" (then) or "with AI" (now) at the end - and boom, there you go, profit!
Re: (Score:2)
Re: (Score:2)
I skimmed the paper, and they are not doing PCA. It's just something to compare against. They are using an encoding generated by a neural network. It's unlikely PCA could take pixel features from a video of a complex system and generate something useful for prediction.
Re: Conformal mapping and multivariate analysis (Score:2)
What good is a science (Score:2)
If it canâ(TM)t tell how it arrived at its conclusions
Re: What good is a science (Score:2)
Re: What good is a science (Score:2)
Re: (Score:1)
Re: (Score:1)
Generalized coordinates (Score:2)
Can't be sure without access to the paper, but it sounds like their program has succeeded in finding some canonical coordinates for the underlying system. This is neither new physics nor a particularly new idea, having been introduced in the 1800s. Not to say that their program might not find some interesting / useful sets of coordinates, of course.
The paper makes no such claim. Phys.org is wrong. (Score:1)
Notably, the phrase "alternative physics" does not exist in the paper, anywhere.
Same reality (Score:1)
Because physics laws comes from the abstraction of the shared reality, I think that the law should be same or close.
Because most "physics laws" are just simplifications, there is room for another intelligence to finds different simplification paths than ours. And of course, the same law could be expressed on multiple ways, although in the end it express the same thing.
Also, most advanced minds could avoid using simplificated formulas and use the most precise expressions that could be beyond our comprehensio
AI could make the same mistakes as humans (Score:2)
Consider an alternative experiment where the camera is aimed at pictures of fires, and the question is, "How do things burn?"
There's a significant chance that, lacking the ability to set up experiments, the AI would invent Phlogiston - after all, human beans spent decades chasing this false explanation.
So, what's the betting the AI in the subject case has reinvented some old and erroneous physics, such as the idea that first law of motion is that in the absence of force, an object is stationary.
Translation (Score:2)
"Extracting the variables themselves was not easy, since the program cannot describe them in any intuitive way that would be understandable to humans"
They pulled it out of their ass, like many studies today.
Seen this twice now (Score:2)
The first time was the two AI that started talking to each other in a language developed by the AI, they ended up pulling the plug because there was was no way to see what was being conversed.
Now this.. itâ(TM)s seems a general rule of thumb for the programmers is not to let the AI do something without a way for it to be able to explain itself to us meatsacks as to how it got those results.
Re: Seen this twice now (Score:2)
Going to edit myself here⦠just saw the fact check on this, was not AI, just chatbots and they didnâ(TM)t pull the plugâ¦. It was a while ago and I had forgotten details except the scary news articles.
This is hardly surprising. (Score:2)
For every mathematical system, there is an infinity of other mathematical systems that will also evaluate to the same results. It's not obvious that a naive intelligence would settle on the same one we use. A great example would be the Quaternion coordinate system. Who would have thought that using coordinate systems that use numbers that contain the square root of negative one would be superior to the ones we are most accustomed to using? It may be the hardest one to visualize in our heads, but it's cl
4.7? wrong! (Score:2)
Everyone knows the answer is 42,
New physics? Please. (Score:1)
Dimensions and quaternions (Score:2)
3D positions and motion can be described using any number of additional dimensions. Many years ago I was working on a game engine for Pocket PC, and was looking into Half Life's model system that they use for animating the humanoid models. It uses a bone and mesh type structure, where the mesh vertices are each bound to a "bone" and are transformed by that bone, and the bone in turn has a transformation relative to its parent bone that is described in the various animation sequences.
Anyway, the point is th
Alternative facts? (Score:2)
Does this "alternative physics" include "alternative facts"?
What good is a system? (Score:3)
What good is a system to discover new variables if it can't share and explain what they are and how they applied?
In mathematics and science, you need to show a proof of the calculations you performed and be able to defend them against scrutiny. If you can't, well then your outcome, regardless how close it is to being accurate, is unreliable and unverified.
Using Einstein as the example, he didn't just propose E=MC2 and everyone just accepted it. He had to prove his theory, prove his calculations, withstand scrutiny of his work... And, not being a renown physicist, it was a hard-fought battle for him and his theory...
So what good is an AI that says "Yeah I found 4 variables" but cannot explain them, cannot show a proof, etc?
but the fire loop example (Score:2)
the looping video of the fire needs 24 variables? I find that very strange. I wonder if edge effects (where the video is looped) is causing spurious non-physics.
reminds me of one of my advisor's observations- regardless how complicated a system is, in the end it's only a handful of variables that are important.
Well duh (Score:2)
Obviously this alien race would have some crude understanding of gravitation or mass/energy conversion to abduct humans.
Or maybe they were fed up with playing and created this large billiard table for us to live in.
It's simple (Score:2)
obviously
What's alt physics? Is it like alternative facts? (Score:1)
This reminds me of The Road Not Taken (Score:2)
A short story by Harry Turtledove where an alien race invades Earth. Well worth a read if you haven't read it before: https://eyeofmidas.com/scifi/T... [eyeofmidas.com]