Something is alive iff, within an arbitrary range of time, it replicates such that the replicant is also capable of replicating, and the replicant is different from the original, and those differences are both inheritable and also affect the replicant's ability to reproduce when compared to the original.
This theory is held by RobHarwood
(and no one else, as far as I know :-)
Sounds pretty close to Dawkins.
The attraction of this theory that it is as minimal as possible while still supporting all the conclusions about life that traditional science supports. It also avoids excluding traditionally marginal cases such as virii, languages, ArtificialLife
, and GeneticAlgorithm
s that might provide a deeper insight into the nature of life. Indeed, science itself fits the description of NaturalSearch
, and so could be considered alive by this definition.
One consequence of this definition is that any agent of NaturalSearch
will give rise to a process of evolution of the same nature as the traditional theory of EvolutionByNaturalSelection
. Thus things which have typically been ignored in the context of EvolutionByNaturalSelection
, such as languages, can be characterized under the same single theory of evolution. Predictions about EvolutionByNaturalSelection
will apply equally to agents of NaturalSearch
There's no need to define things like languages and GeneticAlgorithm
s as alive, though, in order to use EvolutionByNaturalSelection
. You just need to give it a broader context than Darwin did. Making evolution the key component of a DefinitionOfLife
is hotly disputed on said page.
On the contrary, when I first began discussing this idea with friends studying philosophy, they repeatedly objected to my use of 'evolution' and 'natural selection' when talking about (proposed) non-natural life forms. Separating out the term NaturalSearch is an attempt to allow me to talk about different
types of life. So, when I talk about natural life, I use EvolutionByNaturalSelection, which is the traditional theory used by biologists today. When I talk about GeneticAlgorithms and EvolutionaryProgramming, I quote things like JohnHolland and the NoFreeLunchTheorem, etc. When I want to talk about the things that natural life and artificial life share in common, I had to come up with some other term, so I decided on NaturalSearch since I couldn't find any other term that encompassed what I wanted it to encompass. -- RobHarwood
You have weird friends, if they are more comfortable with making signet rings alive than claiming EvolutionByNaturalSelection
can apply to non-living things. :)
Making evolution the key component of a DefinitionOfLife
is hotly disputed on said page.
That's fine by me. I wasn't expecting many people to agree; and I didn't explain myself very well at first. (Though I
definitely wasn't expecting such a strongly negative reaction.) Let me try to explain why I think evolution is so important to the DefinitionOfLife:
If we restrict our consideration strictly to natural life, for a moment, I can explain my thought processes. When I think of natural life (humans, dogs, trees, bacteria, etc.) the features I see as most important features are complexity, robustness, responsiveness, etc. I ask myself, how could this have come about?
Typical textbook conditions for natural life include (feel free to add your own and I'll try to respond sensibly):
- Maintains its own structure
- Did the first life form maintain its own structure? Hypothesis: If at all, probably not very well. However, maintaining structure is an important survival task, especially for more complex creatures. In a dynamic environment, there will be many niches where self-maintenance is strongly selected for by evolution. Therefore, the wide degree and variety of self-maintenance we see today in natural life is due to evolution. Experiments confirm that self-maintenance is strongly affected by evolutionary forces.
- Complex structure
- Did the first life form have a complex structure? Hypothesis: It couldn't have been that complex if it arised spontaneously as current theory claims. However, as reproduction is exponential, the environment quickly becomes crowded and competition ensues. As EvolutionByNaturalSelection predicts, complexity accumulates in the population slowly but progressively over long periods of time. Therefore, the extreme complexity we see in natural life today is a result of billions of years of evolution.
- Reproduction is a pre-requisite for EvolutionByNaturalSelection, so let's not consider gene replication an example of growth. However, growth occurs in natural life in the form of anabolic metabolism, generating the complex structures that evolution has selected for. So, the first form of life was probably a simple replicator (my current best-guess is ribozymes) which quickly acquired complex embelishments. But these embelishments take time to construct, of course, so really growth is just a natural extension of evolution's selection for complexity.
- Negative Entropy
- This condition is a result of replication and the anabolic metabolism of growth, which is a natural extension of evolution's selection for complexity.
This is why I see replication and NaturalSelection
as the two primary characteristics for natural life. Most other characteristics (for natural life) can be reduced to these two. If they can't, then of course they deserve to be part of the DefinitionOfLife
, but so far I haven't found any exceptions that couldn't plausibly be reduced. So, I have attempted to strip down the conditions to just these two. Having done that, I noticed that other things such as GeneticAlgorithm
s and viri exhibit the same two characteristics, but do not necessarily exhibit all the textbook characteristics. Clearly (in my
mind), there is a concept that joins these marginal cases to natural life. That concept is what I call NaturalSearch
, and this is why I think that it is more useful to define life as the exhibition of NaturalSearch
. Then, natural life is just a specialization of 'life in general' (see the crude UML diagram on the NaturalSearch
page). You can think of natural life as that life which started on Earth billions of years ago, and is the study of traditional biology today (and which is categorized by the textbook definitions I've mentioned).
Rob, there was no need to re-explain this; it was understood the first time you said it on DefinitionOfLife. Most of what you say about natural searches is fine, but your claim that evolution is a primary characteristic of life doesn't work - and the problem isn't that it includes non-traditional things, but that it doesn't adequately handle classical cases like protozoa. All this has been layed out on the other page, and should be addressed; simple reiteration doesn't help.
I wasn't hoping to duplicate any of the discussion here, simply to refer people to the other page. You should refactor your comments from one page into the other - putting the whole discussion here would probably be better, but the other page might be a little hard to carve up.
I kind of gave up on the DefinitionOfLife
page and was intending to wait until it cooled down before I tried any sort of refactoring. -- RH
I've split that page up into little pieces; the one on NaturalSelection hasn't changed since you left, so help yourself to it. Please, please, please address the issues brought up there before continuing to expouse this position. The stuff here on mules is a start, but by no means all that is problematic.
Ha! I tried to edit the DefinitionOfLife
page but I couldn't. It's too big. I may have to try a different browser if this becomes a recurrent problem.
Anyway, I'll try to address your questions on that page at the bottom of this one.
Just remember that when this is all done, everything should be in one place. Probably here. I'll try and see what I can do about the other page, and let you refactor when that's done. If this page also ends up too big, btw, leave me a note on "JoshuaGrosse".
Oh, and just for the record - I don't think you met an exceptionally negative reaction by and large. In fact I didn't even consider it particularly wrong until we started to consider some of its ramifications.
For a rebuttal of NaturalSearchIsaDefinitionOfLife
, on the grounds of a comparison between "life" on one hand, and "the space shuttle" on the other, and a response to the rebuttal, see FoundDesignedDesignoid
...but is it a useful one ?
The interesting hypothesis made by Darwin and others, then refined in the following decades into a scientific theory with very solid standing, is as follows : we should expect a NaturalSearch
process to result in things which have precisely those characteristics which life exhibits. Life, in other words, could
have been produced by an infinite number (and actually even a finite number) of monkeys working without purpose, provided that
these monkeys operated under constraints equivalent to those of a NaturalSearch
In this context, proposing that NaturalSearchIsaDefinitionOfLife
is a sensible position to adopt, though this "definition of life" might not be the most useful definition to adopt, depending on the context in which we discuss "life".
, I know for a fact that some reputable scientists are quite willing to work with this definition in some contexts. RichardDawkins
is probably a good example.
This isn't a very good definition at all, though. That life is generated by a NaturalSearch, and that life
is that which undergoes a NaturalSearch, are two very different statements...these considerations are already in another page, and it would be very nice to keep them localized.
This is a good point, and this is why I made two separate pages. The NaturalSearch
page is just a description of EvolutionByNaturalSelection
with a broader scope than natural life. NaturalSearchIsaDefinitionOfLife
is my own 'kooky' idea which I happen to think is useful. -- RobHarwood
Mules. How does this definition apply to mules?
Thanks for this example, it has been on my mind lately. Like I mentioned elsewhere, this is the only really big stumbling block I've come across. I admit it. 'Living things' that fail to reproduce are not considered 'alive' by the definition of natural search. Now, being the kook that I am, I'm not going to try to amend the hypothesis in order to fit the 'facts'. Instead, I'm going to try to reinterpret the facts to see if they might support (or at least not contradict) the hypothesis.
Here's a list of 'living things' that the natural search definition wouldn't classify as 'living':
- Mules (genetically sterile)
- Castrated bulls (physically sterile)
- A human who chooses not to have children
- ... (add your own)
There are a few possible directions we could take this. One is to distinguish between the 'process of life' and 'examples of living things'. Another is to forego the condition that reproduction is necessary
and instead talk about the potential
For reasons of personal preference, I don't think the 'potential reproducer' will lead us very far. I suspect that it would bring us down the road of being impossible to verify or disprove.
My current direction of thought is to keep NaturalSearchIsaDefinitionOfLife
, but rephrase it to remove the 'Something is alive iff...' bit and replace it with a definition of the process of life. From that, we can reinterpret those things which we traditionally call 'living things' as 'atrifacts of life' or some such descriptor.
Artifacts of natural
life (I know this is a misleading term, but I can't think of a better one right now) would be those things which are studied by Physiology, Anatomy, etc. In the GeneticAlgorithm
analogy, the artifacts of life would be the candidate solutions (also called genomes, or chromosomes, depending on who you talk to).
The process of life would be the mechanism by which these artifacts are generated, and the explanation for their complexity, robustness, and responsiveness.
Thus a mule is still an artifact of life
, though on an evolutionary scale it would not affect the direction of the process of life
. Those artifacts which do
be affecting the direction. However, since both arise from natural life, it still makes sense to study mules in zoology.
So perhaps this contention over various instances of living things and non-living things has highlighted the distinction between life and its physical manifestation. -- RobHarwood
Rob, you're again co-opting a term which is already in heavy use. Life has two meanings. The first is the quality shared by living things, and the second is the property of living. Living has no relation to 'the process of life' as you define it. Even sterile people go about living without ever participating in 'the process of life'. Further, EvolutionByNaturalSelection
is only one of the possible ways to get 'artifacts of life'. The other way is to design it, which has been done.
The problem is this. What's the difference between a sterile and a brain-dead infant? Neither one undergoes the process of life, and both are complicated globs of living cells that are produced by a parent. So they are both "artifacts of life" in the above sense. But one is alive and the other isn't.
Though there is a difference between life and a manifestation of life, this does nothing to remove the problem. A mule isn't a manifestation of the process of life as you describe it. It is an artifact of it, but so are DNA-containing waste products for that matter. I don't think the distinction can salvage things, and would actually be more willing to trust in some variant of the potential argument (e.g. contains the info necessary for its reproduction). -- JoshuaGrosse
This seems to me to be more of a problem with the English language. There are many interpretations of the word 'life' depending on context. I'll try to be as precise as possible.
The sterile child may still affect the process of life through kin selection. It is also a complex artifact of natural life. It is also medically alive (respiration, and all that).
The brain-dead child is unlikely to affect the process of life much. It is more part of the environment than the population (sounds cold, but unfortunately it's true). It is also medically alive. It just happens to have a major
The problem the I
have with potential is that pure potential (i.e. no actual) will not give rise to those qualities that I think are important (complexity, robustness, responsiveness). A lot of people have the potential to make a huge difference in the world, but the differences in the world only come from those people who actually do something about it.
That's all well and good, but you haven't resolved the problem. What is it that makes one a living things and the other not? Or, if brain-dead isn't dead enough for you, how about a child who is just a dead, shapeless mass of (nevertheless) human cells? Or what if the cells are dead? You can grade right down from St. Augustine to coal; where are you drawing the line?
By 'dead' do you mean 'not an artifact of life, 'not an artifact of
natural'' life', 'not affecting the process of life', or 'not medically alive'? Each of these seem to have fairly decent distinctions at this point. -- RH
Beats me. These are distinctions your model makes, not mine. As far as I'm concerned, dead means not functioning as a living thing. Going by your definition, none of these have processes of life, and all are artifacts thereof. So my question is to you, which of these does your definition of living thing cover?
Most environments are actually quite stable on that scale, e.g. fresh water.
In the natural world, almost every environment is dynamic because, to an artifact of life, other artifacts are part of the environment, and they are continually interacting with and influencing each other. Other causes of dynamism exists, but this one is universal to natural life.
Have you ever looked at protozoa, Rob?
Didinium is a near-perfect efficiency predator on
Paramecium, but the latter has hardly evolved to acknowledge any threats to its existence. It doesn't need to change, it's doing fine the way it is. The universe of a protozoa is too close to being infinite for there to be any real population pressures from such things.
Yes I have, but how is it relevant? Does this show that the environment is not dynamic? I think it shows the opposite.
I don't see how. The presence of predators simply doesn't affect the Paramecia. There are huge volumes of fresh-water out there, and a ciliate will do pretty much equally well throughout most of them.
On protozoa: they've come very, very close to convergence
This thought experiment can be confirmed in a GeneticAlgorithm
simulation: A gene which controls the mutation rate can be modified up and down through mutation to any value between 100% and 0%. (Once set to 0%, all decendants will be identical.) In a static environment, you will eventually reach a stable state in all solutions are identical, and the mutation rate will eventually go to 0% since it is an absorbing state. In a dynamic environment, where the fitness criteria change continually over time, those solutions which mutate to 0% mutation rate will eventually be unable to adapt to the changing environment and will lose out to the non-zero variants. In a dynamic environment, mutating to 0% mutation is a fatal mutation (I hope that parses correctly). -- RobHarwood
Right, but that's irrelevant because not all environments are that dynamic. Skip the though experiment, because I already have a real example. Fossil record for protozoa is understandably scarce, but only the very most complicated groups (e.g. ciliates, dinoflagellates) appeared in the past few hundred million years. Some flagellates are for all practical purposes the same as they were when eukaryotes first appeared.
The same is true for a lot of other fossil groups. Horseshoe crabs are pretty much exactly the same as the were in the Devonian period some 300E6 years ago. Yes, there's been a lot of genetic drift but not much in response to NaturalSelection at all - the over-all animals is pretty much the same. And if you don't like this example, there are plenty more.
Do you mean genetically identical? How has this been determined? How would this show that mutations are not occurring? Not all important differences can be determined from fossil records. Do you have any examples of a species alive today that has a mutation rate of 0%? If you do, I'll rest my case.
In this case I don't mean genetically identical, I mean identical for all practical purposes. In short, I am saying that evolution is happening, but it doesn't matter all that much. In the case of some flagellates not at all: genetic drift is not NaturalSelection. As for 0% drift as well as change, see below.
In a static environment, we would expect to see mutation rate to converge, and eventually reach 0%. So if there are any significant static environments on Earth, I would expect to see at least one species with a mutation rate of 0%.
I would expect no such thing. There is always going to be a non-zero mutation rate because mistakes happen. But what sort of force are you expecting to cause convergence here? As stated, eukaryotes have elaborate mechanisms to prevent mutations but of course such systems prevent their own effective development. You get a convergence on 0%, but that doesn't mean you'll reach it.
That stated, I think you will find that the mutation rates in many protozoa are as close to zero as you might hope. Polymerases all have correction enzymes built in and mistakes are very rare. In fact you will note that those eukaryotes that do happen to live in dynamic environments have had to come up with a complicated system of thwarting their own genetic stability (sex). Seeing as how difficult the procedure is, it speaks a lot on how low the mutation rate is in asexual protozoa (bacteria are, of course, different).
On metaphor: The problem is that running = alive in this sense is nothing like what life means in ordinary biology
This is because I was confusing process and artifact together. You're right, ordinary biology deals with artifacts of natural life. The natural search definition is more general and deals with the process by which those artifacts arose. Algorithm is not subject under consideration per se, but the process which the subject participates in (or doesn't participate in). -- RobHarwood
I'm going to take a break from this discussion for a while. When I originally proposed that NaturalSearchIsaDefinitionOfLife
, I was intending to expose the Wiki community to an alternative definition of life. It has since become an bit of a debate, which honestly I don't have the time for (see WritersBlock
). The various ins and outs of the definition of life isn't really relevant to Wiki, though possibly NaturalSearch
is since it's related to GeneticAlgorithm
s. Good luck to you all and happy refactoring! It was fun while it lasted. :-) :-) -- RobHarwood