Hacker News Reader

Comment by rwmj | Hacker News Reader

Back in the 1990s I spent a long time playing with Tierra (https://en.wikipedia.org/wiki/Tierra_(computer_simulation) ) and actually wrote my own distributed version[1] so it could run over multiple machines (computers were really slow back in 1991), and a simplified instruction set. Now I know that Tierra has a number of serious shortcomings compared to real evolution. Nevertheless it does really surprising things if you let it run long enough and then examine the cells that evolve in the system. Some cells become parasites, some become really large. I also saw really surprising optimizations happen spontaneously that I wouldn't have thought about.

Anyway given that the real thing has been running for 3 or 4 billion years, I'm not surprised that it has generated incredible complexity.

[1] http://oirase.annexia.org/dlife/

View on HN

Would you expand a bit on the shortcomings? I'm under the impression artificial life was an attempt in the 90s and after some disappointments fizzled away. Why is simple copying with mutation not sufficient for real evolution? In my naivety I hoped it would lead to digital evolution finding exploits to proliferate across the internet.

The Wikipedia page links to some papers (if you can find them - they are online but paywalled sadly).

In practical terms I found Tierra and my system to be very interesting to observe.

(Since I can't edit the original comment above ...)

Here's the description I wrote of my artificial life system: http://oirase.annexia.org/dlife/machineref.html

And here's the "god cell", the initial hand-written cell, from which all other cells evolve: http://oirase.annexia.org/dlife/god.dla

One issue with 'evolving' digital life is the primary competition is human assisted digital applications as the primary competition, effectively intelligent design, which is far more powerful ran random chance.

Non-artificial life actually gets massive benefits from horizontal gene transfer which is natures way of letting someone else do all the homework for you.

>According to Thomas S. Ray and others, this may allow for more "open-ended" evolution, in which the dynamics of the feedback between evolutionary and ecological processes can itself change over time (see evolvability), although this claim has not been realized – like other digital evolution systems, it eventually reaches a point where novelty ceases to be created, and the system at large begins either looping or ceases to 'evolve'.

But of course, evolution merely reaches equilibrium, which takes finite time and then stagnates. In real world evolution continues due to change of environment that moves the goalpost of equilibrium, extincts overoptimized species, so evolution can go further. Just add slow change to environment, like ice age or warming.

it's sort of amazing how a little visualization helps:

https://rednuht.org/genetic_cars_2/

Reply on news.ycombinator.com