Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Ask HN: What would alien code look like?
31 points by coryl on July 12, 2010 | hide | past | favorite | 33 comments
Hi fellers,

Disclaimer: Not a programmer, but semi-technical and know a little bit about engineering concepts.

That said, lets be imaginative for a bit and pretend we made contact with an advanced space travelling race of alien. We have the opportunity to look their technology and programming behind it.

What do you think would be significantly different from the way they write their code and ours? Would they have the same logic as us? Could they use logic gates that we don't have yet? Would their computers have to be based in binary? What could they use instead?

Fundamentally, could their programs be translated directly into human code, like C++, or would a difference in conceptual understanding be so different that human code could not express alien logic?

Feel free to let your mind run with this. Would love to hear some crazy thoughts. Thanks



"Few of us have been to Mars. But science fiction has filled in many of the gaps. We know that the Martian civilization is immensely old. Its software technology, therefore, must be similarly antique. Obviously nothing like it exists on earth - though those of us who remember VMS might be deceived...

...Since Earth code is fifty years old, and Martian code is fifty million years old, Martian code has been evolving into a big ball of mud for a million times longer than Earth software. (And two million times longer than Windows...)

...There is only one thing to do with it: obliterate the trunk, fire the developers, and hire a whole new fscking army of Martian code-monkeys to rewrite the entire fscking thing.

This is such an obvious and essential response to the big ball of mud pattern that, despite the fact that we know nothing about Mars, we can deduce that it must have happened on Mars. Probably several times. Probably several hundred. For each of these attempts but the last, of course, the result was either (a) abject failure, (b) another big ball of mud, or (c) both.

But the last, by definition, succeeded. This is the crucial inference we can draw about Mars: since the Martians had 50 million years to try, in the end they must have succeeded. The result: Martian code, as we know it today. Not enormous and horrible - tiny and diamond-perfect."

No comment on what follows, but you might find it entertaining.

http://moronlab.blogspot.com/2010/01/urbit-functional-progra...


Warning for others following the link: that's an overly verbose post with too much unfunny humour, describing a language whose implementation - on Google code - is written in a language (Watt) the author defines in the previous post. That language, in turn, doesn't appear to be much more than a more obscure version of the lambda calculus, and about as readable.


Counter-warning: I liked it :) but I tend to like more fanciful, meandering approaches to CS topics, like Why the Lucky Stiff and Clifford Pickover. Or Lewis Carroll, for that matter.

Notable concepts from the article: a "fixed-point" of language design, the idea of working backward from an aesthetic (Martian) to a language, and Kelvin versioning, in which "releases count down by integer degrees Kelvin. At absolute zero, the system can no longer be changed."

It's true, though; I found the actual language the author's made (or is still making?) to be pretty opaque.


Logic gates are fundamental, and would be exactly the same. [1]

They don't have to be binary, but it's a lot easier, so it's a near certainty. Plus true/false is also fundamental.

So I think at a low level it'll be pretty similar.

Math is usually considered to be universal, and computers are based on math, so that will all be the same.

Memory access however could be totally different. Our machines are Von Neumann, but aliens could use a totally different architecture.

[1] There is one gate for each combination of binary input and output. There are three inputs: 0 0, 1 1, and 0 1 (1 0 is the same thing). For each input there are two possible outputs, total 6. There are 6 gates: and, or, xor, nand, nor, nxor (xnor)

Or put another way:

                                 xnor
             nor  xor  nand and  nxor or
  -------------------------------------------
  1 1:  0    0    0    0    1    1    1    1
  0 1:  0    0    1    1    0    0    1    1
  0 0:  0    1    0    1    0    1    0    1
(A gate that always returns 0 or 1 is useless, so those two don't count.)

If you want logic gates that depend on position, then you have more options (16 in total, compared to the 8 I listed), and you can see them here: http://en.wikipedia.org/wiki/Logical_connective#Table_of_bin...

Our computers don't use them, I suppose alien ones could, but I doubt they would.


Maybe they also invented the memristor:

http://news.ycombinator.com/item?id=397335

The memristor is an imply gate:

  0 0 | 1
  0 1 | 1
  1 0 | 0
  1 1 | 1


I believe I read recently that some experimental logic circuits work best when creating an "implies" gate (fewer circuit elements vs a nand), so implies is used as the fundamental gate.

Edit: jonsen's post has better information. I was indeed thinking of a memristor.


I'm not even going to venture a guess on the lower-level similarities. It's certainly easy to say that, of course they'd use binary, it's the obvious, simple choice. But I'm not an alien and don't presume to know what they would find obvious. Their "computers" might be built by wiring a bunch of alien cow brains together until they get something that can navigate starships but can't be trusted to add two numbers together reliably. Things are made from what's handy and then grow from there, not necessarily from whatever is the most elegant or mathematically pure.

One thing I would bet on is that they wouldn't code in anything resembling C++. Even now, only a handful of decades into human computing, languages are becoming increasingly high level and abstracted from the machines themselves. Assuming that our alien friends have had a few centuries / millenia to improve on things, I'd expect that they'd have progressed to much more intuitive, more automated, and less detail-oriented methods. Especially if we assume that they've developed reasonably strong AI, the very idea of "programming" could be as strange to them as someone building their own computer from wire and logic gates would be to us.


actually I saw someone at the retro computer fair who'd done just that with a ZX spectrum: http://www.zxdesign.info/


They will use balanced ternary rather than binary at the lowest hardware levels:

http://en.wikipedia.org/wiki/Balanced_ternary

I first learned of balanced ternary from this article:

http://www.americanscientist.org/issues/id.3268,y.0,no.,cont...

(Beware; the balanced-ternary example values there don't seem to be rendering negative digits correctly.)


That's a great question. I think any advanced space-faring civilization would have no further use of programming languages as we know them, as any ship or computer system would be sentient or even sapient by their builder civilization's standards. Communicating with these artificial minds would be as difficult as communicating with the aliens themselves.

We would have to overcome whatever language barriers we find before being able to do the equivalent of "programming" - I don't think there would be any language below whatever their equivalent of microcode though, nor do I think any language could express everything necessary.

Well, maybe Lisp ;-)


So you think they could just describe what they need, and their awesome AI computers could just whip up something just like that? If so, even the AI bots would express alien logic birthed unto them (unless we're talking about a species of AI that has evolved into its own race, but lets not go there :D). So at some point, something or someone has to put code down into some hardware.


Well I think anything that was used to describe the AI programming would have long since been lost to the ages, and the AI outgrown any known language which can describe it.


well really it's mostly hacked together with perl ;)


If they are advanced enough to space travel, and they evolved from a survival mode, it would imply that their thinking organs are not designed to deal with infinite complexity. Just like us, their brains would have been evolved to deal with some primary physical stuff (like falling rocks, fire, etc). So programming will be as complex to them as it is to us.

The complexities would have to be abstracted away, and if they were space travelling, the abstractions have to be reliable. So probably, by the time we meet them, they have some type of black-box elementary code particulars for many different functions that everyone needs to join together to produce machines.

Probably, they would have developed some type of inform interface method, where you can specify what you want in and out and the appropriate atomic block gets delivered to you. They would have some type of global repository that harnesses the power of many different individuals working on problems to avoid any form of repetition.

Currently, we interface with code through typing. This is inefficient for us, and we don't know how they will coummicate. However, for them to be capable of abstract though, they will need to create pictures in their mind. It does not seem farfetched to assume that they have much faster interfacing methods that work directly on pictures they visualise. This allows them join the atomic blocks much quicker.

In effect, they will try to eliminate the manual work entirely, and only do the specifications part of things (tell it what you want), and let the computer build the rest of the things.

On the other hand, perhaps they lack our sense of moral, and will work with biological computers, and not metal-based.


Imagine humans in prehistoric times thinking about making contact with aliens: "Will their spearheads be made of stone or a more advanced material?" If aliens will come we'll see that the questions you ask here are irrelevant, as the terms programs and programming languages will not apply to their technology -- it will be something completely different.

Admittedly, programming has mathematical foundations (algorithms) so it has a good change to still be relevant.

If they do have programs, then the according to the Church-Turing thesis the programs could be translated to human code, but not necessarily in a clear way.


Aliens exist in the same physical universe that we exist in.

However we only engage directly with a small subset of the universe and in ways that have been evolutionarily useful.

Aliens might well have evolved to engage in a different subset.

We certainly have different subsets here on Earth -- consider colour vision for example and the effects on colour coding interpretations between humans, fish, birds, reptiles, bees, etc.

Communication requires being able to predict response to stimulation. The degree of similarity with which we perceive a given stimulation will determine how well we can communicate.

Computer programs can been seen as a kind of communication with a computer, and it only makes sense for programs to be written in a convenient communicative form.

While it might seem obvious that aliens would use digital computers it isn't necessarily so.

Analog computers might make more sense for them -- the main reason we don't use them much these days is that they're hard to reprogram -- but with nanotech or biotech it might be far more trivial.

There are also quantum computers, and non-electrical computational systems.

People also assume numeracy, but consider aliens without recursion which deal with numbers by subitization up to some large quantity rather than counting.


The hardware and software would not necessarily be as distinct as it is in our technology. Alien technology might more closely resemble that of organic life, where the decision making processes are built into the physical structure of the system. Conceivably, these systems could also be designing through an evolutionary-like process, instead of the human engineering process.


I agree.. I would think that a more highly advanced technology would start to resemble organic life. I'm no engineer, but I'd say we're still a long way off from this in most regards, am I correct?


I think if we encountered aliens, they will not be "biological" but mechanical and electronic. These have by far the more superior characteristics needed for space travel.

The same holds for us, we are sending robots already and it will only be more and more robots in the future.

Also, "artificial" AIs will be the ones doing the programming. They can be extremely more complex than humans, because their electronic brains can actually be scaled up, and new data can be much faster interpreted and shared.

The code they write will be machine code, because they can compile their own abstract ideas themselves. On a low level the code will be extremely optimized. Like how an expert human programmer would optimize his 100 lines of assembly code in the 70s.

Resources will not be wasted, in fact they will have way more computing power and use it very effectively at the same time.

So, how would the code look like? Sadly only ones and zeroes. We would have to translate it to assembly ourselves and then extract more abstract concepts of it, for us to understand it.


"The code they write will be machine code, because they can compile their own abstract ideas themselves."

This doesn't seem likely. Unless this alien species is isolationist to the extreme, they'll need to communicate with others of their species. Therefore they need a language that can communicate high-level concepts, something assembly is not known to be good at doing.

Or to put it another way, even though we communicate with our bodies in a very low-level way (sending neural impulses to contract or relax muscles), our language is very high level. We don't say, "Fire neuron 10212, fire neuron 39232...", we say, "Come over here, please." We don't think in terms of firing neuron either; that's a job for the low-level subconscious.

In a similar way, I'd expect any artificial life form to have a language to express high level concepts, and "compile" with their subconscious.


Yes, but we were talking about writing code, not communication.


What's the difference?


What would alien code look like? Perhaps like Piet: http://www.dangermouse.net/esoteric/piet.html http://www.dangermouse.net/esoteric/piet/samples.html


As programming have mathematical foundations, I think it makes more sense to ask what alien mathematic would look like. Would their mathetmatic be isomorphic with ours? Would they have just their own versions of caluculs, set theory, topology, differential geometry, etc. with just some minor differences? As most of mathematics is based on few simple axioms (from which more complex 'things' are derived[1]) this seems quite unlikely - but on the other hand 'basic and simple' may mean something completely different to them; for example classic geometry for creatures whose vision is not main sense would kind of awkward, also they wouldn't probably think of some integrals as an area.

[1] At least that is how I see mathematics right now as a math/cs student; please correct me if I am mistaken.


It's not exactly an answer to your question, but if you aren't familiar with the Church-Turing thesis (http://en.wikipedia.org/wiki/Church%E2%80%93Turing_thesis), you might find it interesting.


Lets step back a little bit and look at the question of 'sentience'. So far, just as we only have one example of a planet full of life, we also only have ourselves (I am being inclusive of all life on Earth that has neuronal structures we collectively call a 'brain') to draw our references from.

Whether we look at snails with 12 neurons for a brain, or a human or cetacean that have the most complex brains on earth, the biochemistry behind how they work is all rooted from the same evolutionary source.

We can hypothesize that on a another planet where life arises using the same kinds of naturally occurring elements (hydrogen, oxygen, carbon, nitrogen, etc etc), and that this life lives in a similar environment (a warm, wet, ball of rock) that we might see convergent evolution across many, many worlds that share these properties.

DNA is what life here on Earth eventually settled on, but remember that the latest theories of life's beginnings on our world is that RNA probably came first. So its not 100% proof positive that DNA or its chemical equivalents are the only way to make a living, complex cell. Different replicative mechanisms could create different biochemistry, even on a world that is nearly identical to Earth.

The point of this is that even on a world with a different kind of replicative biochemical system, complex life could evolve similar large-scale structures. Fins for the water, legs for land, wings for air, eyes for seeing, etc, etc.

So, does this hypothetical situation also extend to things like brains? If this alternate biosphere uses a chemical system very different from DNA, how does that affect the chemistry of what makes a brain? Perhaps arsenic is like seratonin or dopamine, for example.

Putting aside the chemistry for a moment, does a similar physical environment that encourages things like legs, eyes, and skin encourage the development of similar brain structures to control those systems? Or, are there vastly different kinds of brain structures that can control the stuff that makes up a body?

IBM has been working on simulating the neuronal columns that make up a cat brain. Mice, cats, and humans all share some variation of these columns. It makes sense because we all share common ancestors and we inherited these basic structures from them. But would we necessarily expect to see these structures in life coming from a different biochemical history?

So pulling this back to the original question, how much of our mathematics and computing models correspond to how our form of intelligence has evolved?

Based on the universe we can see and test, we have developed a very large and thorough mathematical knowledge-base of how the universe fits together. But it does beg the question, if another kind of sentient life discovers the same mathematics do they utilize that knowledge in the same ways that we have?

Binary logic is a very efficient way to do our kind of computing. But its not the only way. Quantum computation is an easy example of a non-binary way to do computation. We've also only been doing this for a little over a century, and if we can take anything from the lessons of history its that we should never assume what or how we do something now is going to be the best way for all time.

I can't and shouldn't offer a definitive answer to the question. No one on Earth can provide anything more than a strong hypothetical argument one way or another. Remember, black swans and all.

What it would take to answer this would be to find that ever so elusive alien tech and stick some probes in it and start studying it. Experimental evidence is the only way to ever really prove something.

I will offer a hypothetical answer though, here at the end. From everything I have learned in my life, its likely that math and physics work everywhere we can see. And we can see a long, long way away. Even if some sentients on another world with a vastly different arrangement of biochemistry and thinking-stuff (aka brain analogues) invent their types of 'computers' that work differently than ours, I think we can understand how they work.

As we are demonstrating that we can simulate the physics of the universe in our Turing-complete machines (biology to nuclear explosions to air flow over a race car body), when we figure out all the nitty-gritty of how our brains work we'll simulate those too. Given the proper amount of knowledge of how a system works, such as for alien brains and alien computers, we'll be able to load them up and run them as well.

So, as long as the aliens we encounter are based on the same physics we are, there shouldn't fundamentally be a problem to figuring out how their systems work. There might be a high likely-hood that we find many, many similarities at the root level of their computing technologies to our own.


Speaking as a sci-fi author, who thinks about this quite a bit: there will be more kinds of programming, but it'll all pretty much resemble what we already have. It will be faster and easier, and use (or "waste", depending on your perspective) lots more resources. It would be completely predicated on the technology, not in "alien thinking"; if given a tutorial, humans could easily program the alien computers the same way the aliens could, after struggling with concepts that are about equivalently difficult to recursion or pointer arithmetic.

The kind of programs we know today (encoding static text segments using a tower of abstractions and compilers) would still continue to exist. What we today call "programming" as a whole, will be what aliens call "embedded programming"—the kind you do under severe resource-constraints, where every gate must be secure, reliable, and could be hit with hard radiation at any time. We'd easily be able to understand these programs, given an appropriate disassembler. They would likely be written mostly by constraint-solving AIs, however, not tediously by aliens, who have better things to do. Aliens would, however, check them over out of paranoia/"defense-in-depth." They would be intensely parallel: imagine how NASA shuttle software works, with several independently-contracted codebases on independent hardware voting on each decision—but multiply the scales a millionfold.

Most programs, however, would not be "embedded"; they could expect to rely on a "silly" amount of resources. They would be inscrutable, mostly JIT-genetically-evolved from small "goal scripts" (which themselves would still, indeed, resemble human source code—clear and exact communication of rules and intent within a business domain survives no matter the language used.) They would rely on a "universal library" of "common sense", both in process knowledge (algorithms, protocols) and semantic knowledge (datatypes, fixtures, "facts" in the RDF sense) that would allow programmers to actually write "Do What I Mean" programs as well as could ever be achieved.

There would only be a single "programming language", for the same reason that we, today, have a Unicode character encoding. All programs would be internally stored in something resembling S-expressions (just a binary-encoded tree), and represented, textually or graphically (or audially or kinesthetically or experientially or somatically or telepathically...) through a bijective transformation. Every identifier in a program would be a universally-unique resource-locating ID (i.e. something that is both a UUID and a URL) allowing for importing of live code from external (network) libraries, cross-program optimization and symbolic unification, global API versioning, and removal of all "name clashes."

Also, a bonus sociological perspective: there would be living software—like with carbon-based life, we would classify software into "plants" (natural resources that we care about mostly as an enabler), "animals" (feeling but unthinking goal-directed objects that we think should have some, but not all, the rights of sentients) and "people" (which would have just the same rights bio-people do.) "Animal" software would have control systems that mimic (or simply are) emotions; an alien Roomba would indeed be considered a pet. But in the converse view, the aliens would consider genetically programming carbon-based plants and animals to have symbiotic functions, just as we currently live in symbiosis with robots and machines. They might all be ultra-vegans, for the simple reason that any combination of flavor, texture, and nutritional content could be turned into the fruit of a plant (a symbiotic arrangement), and so would abhor the non-symbiosis with which we consume our own environment.


after struggling with concepts that are about equivalently difficult to recursion or pointer arithmetic

I think that's a bit optimistic. Even within the realms of human programming languages, competence may require struggling with concepts that are as difficult as, say, monads or virtual destructors or Prolog cuts or abstract factory factory factory factories or call-with-current-continuation. Those are all at least one step up from recursion or pointers. I don't see any reason to expect that alien languages would be nearer in conceptual difficulty to C than to Haskell or C++ or Scheme or Prolog.


Interesting ideas. May I ask which books you have written?


Their programming languages will be frighteningly similar to Javascript: http://www.flickr.com/photos/mendhak/2643251456/


Their programs would be single real numbers: http://en.wikipedia.org/wiki/Real_computer .


There are those (myself included) that would say Perl looks like it was created by aliens.


I think it's unclear. We have some ability to do math, but even our specialists learn it with great difficulty. (We see something a bit similar when we try to teach other primates language or math. They can learn enough to fake some of it, with lots of difficult training.) If we want to see more examples of human cognitive limits, I hear we can look to games like chess. We play them because we're bad at them.

It's very possible that there's things we can't imagine how to begin answering (even things we really care about) that another species finds trivial. Or funnier, we could have roughly similar mental powers, but our thought processes are so different that each species' insights seem entirely mystical to the other.

I hear that psychologists (and following them, AI researchers) keep finding that intelligence is buried in our visual and lingusitic systems. What if these aliens have very different organs? http://www.youtube.com/watch?v=auXQtoAyCGA&feature=PlayL...

And off-topic, might this explain "rubber-duck debugging," where we explain our problems to a rubber duck? (Because using language clarifies our thoughts.) And why we often write pictures on paper to better visualize abstract problems? (To leverage our powerful visual apparatus.)




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: