Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Excuse the cliche question, but what is Life?

If we make an AI whose thoughts and communication are practically indistinguishable from humans, and it controls a bunch of things that affect the physical world, is it alive?

If there's a disembodied mind somewhere that can think and imagine but never communicate or interact with anything, is it dead?

What about a human in a coma or permanent dreaming state? What about when they later wake up?

Okay nevermind thoughts and communication; What if there's an armada of "dumb" robots, spawned from a single factory, that can't think, only acts according to preset instructions, but still goes on to affect many things for hundreds of years, what should they be classified as? Would they be considered "artificial" life or automatons even if they existed for thousands of years and their creators were no longer around?

I don't think there is any single absolute criteria to classify all possible entities that must be out there or could theoretically exist, but the answer to "What is life?" is probably this:

Whatever other life thinks is life.



Life is that which is subject to Darwinian evolution.

That definition catches every single thing you'd call life, excludes all the things that clearly aren't, disambiguates some grey cases, and illuminates one or two new phenomena to be life which you might not have realized.

To be clear, Darwinian evolution means:

- It makes copies of itself (reproduction)

- The copies take traits from the original (inheritance)

- The traits can vary among the copies and between generations (mutation)

- The copies' ability to make more copies depends on the traits they inherit (selection)


> It makes copies of itself (reproduction) ...

Again, what about species that don't reproduce themselves, but spawn from a single "factory"?

That factory may be nothing like them. It could be a mechanical factory producing automatons (with human-level thought if you need that criteria), or a universally-unique "mother" creature, or even a non-sentient spawning pool where they spontaneously form every now and then.

If there are entire planets populated by such entities, and they have governed over many things for as much time as we have occupied our Earth, would you still classify them as not-life in your dealings with them?

That would be like not considering Taiwan a country. (Sorry, couldn't resist injecting contemporary political commentary.)


> or a single "mother" creature

How did the "mother" creature come to be? By what mechanism did its complexity emerge?

There is no known or plausible mechanism by which complex life can arise without evolution, and technology which does not change its design is (very appropriately) by the definition above, not life.


> How did the "mother" creature come to be?

It could be a human "mad scientist" creating hordes of Pikachus.

Suppose the scientist has also attained biological immortality for himself, and the Pikachus have spread to multiple planets over thousands of years.

You come upon such a colony planet, but you have no knowledge of their creator. They display no clues that hint at the scientist's role in their civilization (for a civilization it has become by now.)

You cannot discern how the Pikachus are coming into existence. Would you classify them as life? Why not?

How would your interactions with them and their interactions with the universe be fundamentally different than any other life you encounter?

What if they were robotic instead of flesh-and-fur?


I assume by "spreading" you mean that the Pikachus can reproduce?

Can they mutate? If so, then they satisfy all four boxes and they count as life.

If the Pikachus are made of cells, they'll certainly mutate/vary, since their heritable material is DNA, which is quite fragile.

Really, it's hard to imagine having "reproduction" without "mutation" in the imperfect physical world where errors are possible— I can't imagine any copying process, even a technological one, which succeeds with perfect fidelity and 100% probability. So if you have #1, #3 is basically going to be a given, because the universe is imperfect.

I don't think "what material they're made of" matters much, except that by definition flesh has to be alive, since it's made of things (cells) that tick all four boxes.

If the Pikachus don't reproduce themselves, and/or are created with perfect fidelity every time, then yes, I would call them technology and not life. We certainly wouldn't say that iPads (Pikachus) coming from a factory ("mother creature") are alive.


> I assume by "spreading" you mean that the Pikachus can reproduce?

No, they're all created by the mad scientist but can think, reason, communicate, teach, interact, invent and construct perfectly fine on their own.

> I don't think "what material they're made of" matters much ... If the Pikachus don't reproduce themselves, and/or are created with perfect fidelity every time, then yes, I would call them technology and not life.

So, to distill your criteria for life, it comes down to:

• Species must consist of "individuals" who join in pairs to produce another individual.

• New individuals must be slightly different from the individuals who reproduced them.

In other comment [0] you say that you could consider software to be alive:

> Note that "alive" by the given definition doesn't imply "conscious" or "intelligent"

> We certainly wouldn't say that iPads (Pikachus) coming from a factory ("mother creature") are alive.

But if an iPad had a software process that connected to another iPad, and they both commanded the factory to create a new, slightly different iPad, would they be considered alive? :)

[0] https://news.ycombinator.com/item?id=21231043


> Species must consist of "individuals" who join in pairs

Where did you get pairs from? If the Pikachus reproduced by binary fission (splitting in half and then growing a new half, like cells do), or had 18 genders, they would still evolve. They only have to make copies. Doesn't matter how.

> No, they're all created by the mad scientist but can think, reason, communicate, teach, interact, invent

If they can't reproduce (and/or aren't made of stuff that can reproduce), I would say complexity of behavior is irrelevant.

For example, the identical iPads coming off the factory line are still technology, whether they are loaded with simple software or complex software. The complexity could go all the way up to AI, but they would still be technology.

Note that we are used to "intelligent" meaning "alive", because the only intelligent thing we know of at this point in history (us) is alive. But I think it is perfectly valid to suppose that non-living things could be intelligent or even conscious.

> So if an iPad [...] created a new, slightly different iPad, would they be considered alive?

Yes, under those conditions they would certainly start evolving, and I think that's a perfectly reasonable line to draw between technology and life. In this new example, the "phenotype" of the iPads would change in a self-sustaining way, emergently, independent of any design or intent.


Alright: An individual of an species has to initiate the production of another individual of that species, and the new individual must have some differences from the preceding individual.

Do I have it correct?


Yes, but it's not everything. You've got two and a half of the original four criteria there:

> An individual of an species has to initiate the production of another individual

️^ reproduction

> of the same species

It depends on what you're defining to be "the same species", but if we take that to mean "shares traits with the original", then that's "inheritance".

> new individual must have some differences

^️ mutation

The one you're missing is "selection"— that the future success at copying depends on the traits you inherited.

If you have variable traits, but they don't at least indirectly result in either more or fewer copies, then there is no cause for some traits to become more prevalent than others in the population and evolution won't occur. You'll just get a jumble of random traits without any trend toward fitness (aka more efficient copying).

It is a fact of nature/mathematics that those four rules are necessary and sufficient for evolution to occur. If you set them up, evolution will happen, guaranteed. Those rules constitute an algorithm, carried out by the laws of nature on the substrate of physical matter, whose result is evolution— a trend toward increasing reproductive fitness.


Certain genetic algorithms could fulfill this criteria.

Could software be alive?


I'd say yes.

Note that "alive" by the given definition doesn't imply "conscious" or "intelligent" (though that is certainly logically possible for software too, though I doubt that has happened yet).


Excuse the cliche question, but who is Razengan?

If we make an HN account whose thoughts and communication are practically indistinguishable from Razengan's, and it makes similar comments in all the same threads, is this account actually Razengan's?

The answer to "Who is Razengan?" is probably this: Whoever other HN users think Razengan is.


The concept of life is not unique in this regard. One can form a similar argument for any concept whatsoever.

It's more useful to situate concepts such as 'life' at the center of constellations of other concepts such as 'self-determination', 'will', 'evolution', and so on.


I'm guessing they're using the scientific definition:

"A distinctive characteristic of a living organism from dead organism or non-living thing, as specifically distinguished by the capacity to grow, metabolize, respond (to stimuli), adapt, and reproduce"

- https://www.biology-online.org/dictionary/Life


Yes, I'm saying that definition may very likely be challenged once we actually start discovering extraterrestrial entities.


Going from the 'swapping spit' hypothesis. We can just say it suffices for there to be cells that divide and grow and contain DNA or RNA




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

Search: