First self-replicating mathematical creature

The Game of Life is a cellular automation first devised in 1970 by mathematician John Conway. It's played by setting simple rules and then watching how the cells live, die, interact, and form complex patterns that evolve over time. Last month, Canadian computer programmer Andrew Wade managed to spur the emergence of the game's very first self-replicating mathematical creature. It's named Gemini. From New Scientist:

Gemini's implications extend to the real world. "There's a fascination with the complexity that is coming out of these incredibly simple rules," says Susan Stepney, a computer scientist at the University of York, UK, who ran Gemini inside Life, at New Scientist's behest. "Eventually that leads on to biology, putting simple atoms together to make complex life."
Because Wade's replicator copies itself piece by piece, it is analogous to a photocopier rather than a living cell, she says. But it still has implications for understanding life. "The fact that it's doing it differently from biology is in itself interesting, because it shows there are multiple ways of solving the same problem. It's a very impressive technical achievement."
It's doing it differently from biology, showing there are multiple ways of solving the same problem Stephen Wolfram, famous for championing cellular automata as a replacement for scientific equations, disputes Gemini's relevance to living cells. He says that feeding a program to a universal constructor merely to create a self-replicating creature - Wade's approach, and Von Neumann's original suggestion - is overkill. He points to a much simpler example, a one-dimensional cellular automaton known as "rule 90" that will duplicate any starting line of cells after a certain number of steps.
Rather than contributing to our understanding of life, Wolfram says Wade's discovery could help devise ways to build a molecular-scale computer, starting from tiny components like the cells in Life. "This discovery is helping us understand the world of constructing things from dumb components," he says.

"First replicating creature spawned in life simulator"

UPDATE: Lots of debate in the comments here, and also at New Scientist, about whether this is really as much of a novelty in the Game of Life as the article suggests. It'll be interesting to see how the discussion evolves. Get it? EVOLVES!!! Hahahahah....

David Pescovitz is Boing Boing's co-editor/managing partner. He's also a research director at Institute for the Future. On Instagram, he's @pesco.

MORE: Technology

More at Boing Boing

The technology that links taxonomy and Star Trek

Hackers prepare for first "national holiday" in their honor

Cormac

The New Scientist article is massively overstating the pattern from what I’ve read. It is just a spaceship that happens to be using a constructor assembly. Specifically the LifeWiki says that Gemini is an oblique spaceship (not sure what the oblique means given the wiki didn’t explain it).
- jere7my
  
  An oblique spaceship is one that doesn’t travel orthogonally or (bishop-wise) diagonally.
William R. Buckley

For all the references to self-replication, what is missing is references to something better than self-replication: ontogeny. See my paper, Computational Ontogeny, for a real eyeopener.

Gemini is a programmed glider, and nothing more.

GoL has yet to produce a general constructor, much less a universal constructor.

I however have constructed a machine which not only makes copies of itself, it actually builds itself.
- friendpuppy
  
  How may I read your paper without paying $$ ?
  - Anonymous
    
    You can find a free version of this and another of Buckley’s papers here:
    http://calevinst.academia.edu/WilliamRBuckley/Papers
    They are a bit short, although that can be a good thing. His work definitely seems much more significant in the goal of simulating true self replication in a cellular automata than the Gemini spaceship.
- Anonymous
  
  I would love to read your paper. How may I find and access it?
jere7my

In William Poundstone’s 1985 book The Recursive Universe, he describes a self-replicating life pattern. Did that not work out, or was he addressing a different challenge?
- Anonymous
  
  Well, life has been known to be Turing complete for some time, so that implies that this kind of machine exists, because you can write a quine. I guess this one is interesting because it is small enough to be simulated in practice.
  
  But I have to agree with comment #1 that this pattern does not seem to replicate itself more than the glider does…
- Mark Gritter
  
  “The Recursive Universe” describes the elements of a self-replicating pattern; basically a proof of existence that such a pattern could be built in Life. But it was only a theoretical exercise, nobody had actually specified how to put all the parts together.
- Anonymous
  
  There are lots of self-replicating patterns known in cellular automata, for instance in HighLife, which is similar to Conway’s rules. However, although Conway’s life is Turing complete so a replicator should exist, none is yet known. As pointed out, if the original doesn’t continue to work, you have a puffer or spaceship not replicator.
Felton

Great! As if math weren’t intimidating enough already, now it comes in the form of a self-replicating creature? ;^)
kslays

While this is NOT a self-replicating pattern and New Scientist is sensationalizing (as usual) to the point of falsehood, the innovation of using a simple glider tape between two spaceships may eventually yield a self-replicating pattern. I define a self-replicating pattern as one that creates an identical copy without destroying the original. If the original is destroyed (as in this example), then it is a spaceship. This pattern is also notable for being the first oblique (ie not one of the eight cardinal directions) glider.
- David Pescovitz
  
  Key phrase: “I define…”
  
  Maybe others define it differently?
  - Matt Deckard
    
    You’re missing the point. Regardless of how you want to define a self-replicator in the context of the Game of Life, there’s no disputing the fact that a pattern that replicates itself while also destroying the original is defined as a spaceship within that context. If that’s what this is doing, then it’s nothing more than a spaceship and it’s nothing new.
    - Andrew
      
      It is a spaceship, but the way it replicates is (I believe) unique, creating a new copy of itself some distance away before destroying the original pattern, rather than ‘shuffling along’ as other spaceships do. Gemini looks like a deeply impressive pattern.
      
      Though it’s worth noting that this isn’t the first self-replicating mathematical creature. That honour belongs to von Neumann’s Universal Constructor.
      - ikoino
        
        Wow! I was going to refer to Langton’s loops, never seen an illustration of von Neumann’s Universal Constructor before. Wikipedia made my day.
        
        Gemini is interesting in that, unlike rule 90, the rules of how a cell reacts to its neighbors is local and comes from beyond its immediate neighbors over time. Reminds me of Palmiter’s protozoa, but synchronous: http://bit.ly/ayreBe
      - Keenan Pepper
        
        “Gemini is interesting in that, unlike rule 90, the rules of how a cell reacts to its neighbors is local and comes from beyond its immediate neighbors over time.”
        
        This doesn’t make sense. Rule 90, like all the other elementary 1D CAs, is local. The state of a cell at time t depends only on the states of itself and its immediate neighbors at time t-1. It doesn’t directly depend on any cells farther away.
      - ikoino
        
        Here is an implementation Dewdney’s original sim that inspired Palmiter: http://bit.ly/9HnUWU
        
        This is close to what appeared Scientific American – it is what inspired me to explore cellular automata, which has consumed literally thousands of hours of my time.
tregeagle

I misread the headline as, First self-depreciating mathematical creature. I think I prefer my version.
gwax

How is this any different from any other quine in programming?
Ito Kagehisa

Heh, back in the day I used to go out drinking with Conway’s lab techs… intelligent and fun people.
endymion

I thought the article, and especially the video, were astoundingly confusing. I’m a mathematician who has spent some time playing with Conway’s Life, but I couldn’t figure out what they were really talking about.
SamSam

The links that Cormac #1 was trying to link to, 1 and 2, show that the LifeWiki at conwaylife.com also considers this a spaceship, and nothing more. NewScientist is sensationalizing,

Also, what a bizarre video. Was it designed to make people think they were looking at a slide through a microscope or something? Why was it all fuzzy?
KnoxHarrington

There are already dozens (hundreds? more?) of self-replicating cellular automata. (See, for instance, here. I’m partial to Reggia’s work on the subject.) What’s the innovation here? This reads like another New Scientist article written by someone who doesn’t understand Science.
codesuidae

Is cellular automata creation still in the days of ‘real programmers’, ala “The story of Mel” or have we moved on into patterns created automatically from rules specified by the programmer (or even farther? Can I compile C code to a CA environment now?).

I’m thinking of the evolution of CPU design starting from the early days of painstaking manual placement of transistors up through today when a good deal of the low-level work is hands-off.
Gilgongo

Seems like a more sophisticated version of the Gosper Glider Gun:

http://conwaylife.com/wiki/index.php?title=Gosper_glider_gun