This is an outdated argument. The latest version is here.
The Moral Argument: Why Craig's metaethical case for theism fails
5) Moral semantics
In this article, I will show that William Lane's Craig metaethical argument does not provide any support for theism (I also posted a more comprehensive reply to theistic metaethical arguments – which including a reply to Craig's argument at least as detailed as this one – here).
In older versions, the argument is formulated as:
P1: If God does not exist, objective moral values do not exist.
P2: Objective moral value exist.
P3: God exists.
In more recent versions, he adds 'moral duties' to the formulation:
P1: If God does not exist, then objective moral values and duties do not exist.
Objective moral values and duties do exist.
P3: Therefore, God exists.
My strategy will be to show that Craig's arguments provide no good reason to suspect that the first premise is true, in any of the versions of the argument.
Before I go on, I'd like to point out that this counter-argument is not based on any novel and/or original ideas.
In most cases, I got the ideas from various sources; in a few cases, I came up with them, but this metaethical argument has been discussed for quite some time, so I have no good reason to think I'm the first person to come up with them.
1) Craig calls his argument the "Moral Argument", but I will refer to it as a metaethical argument, for the sake of accuracy.
2) Unless other wise specified, I will use the symbol " when I mean to quote Craig, and ' when I'm not.
For instance, if I post "herd morality", I mean to quote Craig, but not if I say 'objective greenness'.
To be clear, moral agency is not about whether it's morally good or bad to treat a being in some way.
For instance, it's immoral for humans to torture cats for fun, but that does not mean that cats are moral agents. They are not.
4) I will use the word 'argument' loosely, to refer to both the formal argument, and the informal arguments used to support the premises of the formal argument. I think this is a common way of speaking, and context should prevent any ambiguity despite some notational abuse.
In this section, I will analyze the premises of Craig's metaethical argument, in particular the meaning of some obscure terms.
The first premise in Craig's metaethical argument states:
William Lane Craig:
P1: If God does not exist, objective moral values do not exist.
William Lane Craig
If God does not exist, objective moral values and duties do not exist.
I will analyze the terms 'objective', 'moral values', and 'moral duties' later, so in this subsection, I will focus on the meaning of 'God'.
Now, there are definitions of 'God' (e.g., 'an omnipotent, omniscient, morally perfect being, creator of all other beings'), but it's not clear whether Craig provides one in the context of his metaethical argument:
On one hand, in his debate with Sam Harris, Craig says that asking "Well, why is God good?" is like asking why bachelors are unmarried, and some of the language suggests that he uses "the greatest conceivable being" as the meaning of 'God'. 
On the other hand, Craig also says that Anselm "saw" that God is the greatest conceivable being, which suggests Craig is making a claim about one of God's properties, not providing a definition.
While we don't need to define most words we use, since there is a usually a general understanding of the meaning by competent speakers of a language due to the common usage of such words to describe and/or name familiar things (in a broad sense of 'things', like cars, pain, trees, music, etc.), that does not appear to be the case with the word 'God', at least outside philosophy of religion.
For instance, in addition to Harris' claim – which did not appear to be seen as absurd by the audience -, personally, I've encountered plenty of non-theists – including many former Christians – who have no problem saying that God is not good, not great, etc., - and by 'God' of course they don't mean something like 'an omnipotent, omniscient, morally perfect being, creator of all other beings', but probably something like 'the entity described in the Bible and presented as the creator'.
Granted, in the context of this metaethical argument, it's clear that by 'God', Craig does not mean anything like 'the entity described in the Bible'.
However, the lack of precision adds to the general obscurity of the argument; I will have to use the word 'God' nonetheless, despite the significant ambiguity, but I will clarify when needed.
The second premise in Craig's metaethical argument states, in the recent formulation:
William Lane Craig:
Objective moral values and duties do exist.
In the older one, duties are not mentioned.
But let's see what Craig says about those terms:
On the issue of objectiveness, Craig says:
William Lane Craig:
Second, there’s the distinction between being objective or subjective. By “objective” I mean “independent of people’s opinions.” By “subjective” I mean “dependent on people’s opinions.” So to say that there are objective moral values is to say that something is good or bad independent of whatever people think about it. Similarly, to say that we have objective moral duties is to say that certain actions are right or wrong for us regardless of what people think about it.
Craig uses the Holocaust to provide an example, saying that it was morally wrong regardless of the Nazis' belief that it was right, and it would have been morally wrong even if the Nazis had won the war and brainwashed everyone into thinking that it was right.
I will assess the issue of moral ontology and moral semantics later, but here, I will just point out that 'X is immoral' does not mean 'Agent A believes that X is immoral' (that would be circular), so that makes Craig's requirement somewhat odd.
Perhaps, if he had used words such as 'feel', 'approve', etc., in the definition, that might have matters more clear; perhaps, 'subjective' should be defined in terms of how similar the use of a word is, for different speakers.
1) Different people mean the same thing by moral terms (such as 'immoral', 'morally good', etc., or terms that are translated as 'immoral', 'morally good', etc., from other languages into English).
Different languages all have basic moral terms that can be correctly
translated to other languages.
For instance, terms like 'morally good', 'morally bad', 'immoral', 'morally neutral', and 'morally wrong' have equivalents in all languages.
3) There is no circularity (i.e., 'X is immoral', does not mean 'A believes that X is immoral', or anything like that).
This condition should go without saying, but given Craig's definition of 'objective' and 'subjective', I include it just in case.
4) Moral statements are not, in any way, about the speaker in her capacity of speaker.
For instance, the claim that the Holocaust is morally wrong is about, well, the Holocaust, not about the feelings of the speaker, or anything like that.
This condition seems to follow from 1), but I thought I'll specify it separately, just in case someone might interpret 1) differently.
5) Moral statements like 'X is immoral', 'X is morally good', etc., are sometimes true; for instance, the Holocaust was immoral.
Then, if those conditions are met, the objectivity condition Craig posits is met: everything in his arguments in support of the second premise indicates that what he's claiming is that some behaviors are indeed immoral, no matter what people think, how they feel, etc.
On this note, it's important to realize that despite Craig's talk of objective "values", or "duties", the question of what the truth-value of moral statements depends on is a semantic one.
Granted, someone might argue that, without a creator with such-and-such properties, there would not be basic moral terms with the same meaning across speakers, cultures and/or languages, but that would have to be argued for, and the burden would be on the theist.
Moreover, that would be a very implausible claim: for example, species-wide senses do not appear to require the existence of God, and as long as there is some species-wide sense, it seems it's entirely possible for there to be shared basic terms that refer to the properties tracked by that sense.
Also, generally, when it comes to something that is part of our daily experience, such terms usually do exist. For instance, terms such as 'adult', 'child', 'alive', 'dead', 'pain', 'fear', 'pleasure' seem to exist across cultures, and have counterparts in all languages.
Someone might argue that all of that requires the existence of God, but that would have to be argued for.
Alternative, someone might say that in absence of God, we could have the language, but the properties wouldn't exist, so condition 5) would not be met. That also has to be argued for, of course.
So, let's stipulate that morality meets the requirements I explained above – and so, let's grant that premise 2 is true -, and let's see whether Craig's arguments show that God exists.
Another distinction that Craig makes is between moral values and moral duties.
According to Craig, the former has to do with good and bad, and the latter with right and wrong, and with moral obligation. 
He presents some examples as a means to try to clarify that, but his examples are puzzling: for instance, he says that it would be good for you to become a doctor, or a diplomat, or a firefighter, or a homemaker, but you have no moral obligation to become any of those, specifically, or all of them – and you can't do all of them.
However, while it's true that there is no moral obligation in that case, it's not at all clear how it would be morally good for you to become any of those things – not in all cases, anyway.
In fact, in most cases, it seems to me that becoming a doctor, a diplomat, etc. - or for that matter a mathematician, or a physicist, etc. -, would be not be morally good or bad, but morally neutral.
There may be cases in which one has a moral obligation to become something in particular: perhaps, that's the only profession that is within one's reach and can provide an income sufficient to fulfill one's obligation towards one's children – and it's not harmful to others.
I think one can also construct hypothetical scenarios in which it would be immoral to become one of the above – even if, perhaps, not too realistic -, but there is no need to.
The point here is that the general claim that those actions are morally good appears very weak. Readers will of course use their own intuitions, but it seems counterintuitive to me, especially as a broad claim, apparently applicable to most if not all cases.
That is not to say that there is no distinction between what's morally good and what's morally obligatory, though.
For instance, it may be that some heroic actions are morally good, but refraining from doing them is not morally bad, but morally neutral.
So, that would be a case of an action that is morally good, but not morally obligatory.
In any event, the non-theist need not take a stance on this distinction, or try to elucidate it.
In order to accept the second premise, it's sufficient for the non-theist to accept the conditions I mentioned earlier, and then the Holocaust was morally wrong regardless of what everyone believes, etc.
As I mentioned earlier, Craig's metaethical argument for theism is an argument about moral ontology, not about moral semantics, as he explains in his debate with Sam Harris.
In particular, Craig is not claiming – at least, not in that context – that, say, 'X is morally wrong' means the same as 'God forbids X', or something along those lines.
Craig illustrates the distinction by means of the example of light: he points out that people understood the concept 'light' well before there was any understanding of what Craig calls its "physical nature".
He says that one can similarly understand the meaning of moral terms without being "aware" that moral goodness is "grounded in God".
Now, there are a number of interesting points here, but the first one is the following: why would an atheist be expected to provide any account of moral ontology?
After all, if an ontology of light can be provided now (which is debatable, but let's say so), that was not the case for the vast majority of time that humans have been around.
The same goes for an ontology of, say, color – even assuming that a correct ontology of some of those entities, properties, etc., can be given.
On the other hand, if we consider other properties and/or phenomena, there seems to be no good answer to that kind of questions – not yet, anyway: what's the ontology of, say, love? Desire? Intelligence?
However, in any event, an atheist does not have to propose an ontology in order to reject the claim that there is no objective morality without God.
Just as an atheist would have
good no reason to accept a claim that there is no light, or no
objective color without God just because she does not know
anything about the ontology of light, or of colors (or color
properties), she has no good reason to accept the claim that there is
no objective morality if God does not exist just because she
knows nothing about moral ontology.
Instead, it is incumbent on the claimant – in other words, the theist defender of the metaethical argument – to show that if God does not exist, then there is no objective morality, by showing that any such ontology would be impossible – or at least, very implausible.
That aside, it's not the case that the theist is somehow in a better position, epistemically speaking, just because he claims that moral values and duties are "grounded in God", in case the atheist does not have any theory of the ontology of morality.
For that matter, someone could have claimed that green, red and other colors are 'grounded in the gods', or in God, or something else, but unless the claimant has good reasons to believe that the claims in question are true, he is in fact in a worse epistemic situation than the person who recognizes that she does not have knowledge of the ontology of the properties, things, etc., under consideration – since he's making an unfounded claim.
4.1) Color and morality: a useful analogy
As an example used to illustrate the issue of objectiveness , Craig points out that even if everyone were colorblind, there would be a difference between green and red, and the same goes for morality: there would be a difference between right and wrong, good and bad, etc., even if no one could see that.
While not necessary to refute Craig's arguments, the color analogy turns out to be useful to illustrate a few relevant issues:
After claiming that even in a world of colorblind people, there would be a difference between red and green, Craig also goes from that to say that somehow moral values are external to the body, and exist "out there" so to speak.
In a sense, that is acceptable, but we have to be careful not to fall into a trap of inferring some ontological claim about some objects 'greenness', or 'moral goodness' which exists somewhere 'out there', above and beyond green or morally good beings.
Let me explain:
If there is an account of color ontology available to us, it seems it's in term of light of certain frequencies, etc.
But imagine a universe where there is no such light: for instance, the early universe right after the Big Bang would be an example.
Now, in that universe, there would be no red or green objects. Another way to say the same is to say that the properties 'greenness' and 'redness' are not instantiated. But that's only a change in terminology. It does not suggest at all that there is a need to posit some object 'greenness' as the 'foundation of greenness', which exists necessarily and independently of green objects, or anything of the sort.
And if the 'foundation of greenness', of redness, etc., are precisely photons with certain frequencies, etc., then the fact that a universe without any of them is conceivable, and seems to have existed, is no problem for a claim that there is objective color, in the sense that is relevant in this thread: for instance, some Nazi uniforms in World War II weren't red, and they would not have been red – given the same composition – even if the Nazis had won the war and indoctrinated everyone into believing that they were red.
Furthermore, while I said 'universe' above, even if at some point there was not a single green object in the whole of reality – the actual world, if you like -, that too would be no problem for color objectivity.
Now, imagine a universe without humans, or any other moral agents.
In such a world, there is a sense in which there would be no moral goodness or badness – namely, there would be no good or bad people, actions, etc. -, but – as in the case of color -, that is no problem for the stance that there is a difference between right and wrong, good and bad, objective morality, and so on.
In fact, if the universe was like
that in the past – if there was no good or bad entity; in other
words, if moral properties weren't instantiated then -, that too
would be no problem for moral objectivity.
If someone claimed otherwise, they would have to make a case for it.
But their claims would have to be based on moral semantics, since they would have to argue that the there is something in the meaning of moral terms that sets them apart from color terms when it comes to the adequacy of an ontology that does not require that the properties in question – namely, moral properties vs. color properties – be instantiated necessarily.
That said, let's take a closer look at color and moral ontology:
An ontology of color might say – for instance – the property 'greenness' is the same as the property of emitting and/or reflecting photons in wavelengths within certain ranges, gives a list of such frequencies, etc..
We can observe the following facts:
First, the account is given in technical terms, and after considerable advances in physics. There is no ontological theory of color given in non-technical, non-color terms, unless one counts things like 'whatever elicits the experience that we usually have when we use our respective color terms' as giving an ontological account of color (or, if you like, of greenness, redness, etc.).
That shouldn't be surprising: after all, why would non-technical languages have different words that refer to the same property?
Second, as I explained above, the possibility of a world without any green objects, red objects, or generally objects with any color at all is not a problem for objective color.
There is no need to posit a metaphysically necessary green object, or anything of the sort.
Third, science can be used to learn about color properties, but in order to discover the connections between frequencies and colors, we need to trust human color intuitions, at least under certain controlled conditions – else, we simply did not know what it is that we need to match, and no scientific study of color would be doable.
Fourth – maybe not important, but just to be thorough -, there appears to be no semantic equivalence between claims about color statements and statements about wavelengths.
For a competent English speaker, the question 'I know it reflects such-and-such wavelengths, but is it actually red?' remains open, at least from a semantic point of view – i.e., before learning between the connections between them, and even if the person understands both the technical and the non-technical terms.
That, however, is not a problem for an ontology of color, it seems.
There is a crucial point that I will leave for later, but first let's turn back to moral ontology.
First – and keeping the color example in mind – there appears to be no good reason to assume that a correct ontology of, say, goodness, would involve a simple description in terms of other non-technical terms, like greatest happiness, pleasure, satisfaction of desires, etc. - at least, a claim that we should expect that would have to be argued for.
Moreover, for most of the history of our species, no such account of color was available, and it took a lot of development of physics in order for one to be available.
Given the above, it might be that psychology research has to advance considerably, introduce new terms, etc., before we can come up with a plausible account of moral properties in terms other than non-moral terms, especially considering that the science of psychology appears not to be nearly as developed as physics yet.
So, there seems to be no good reason to assume one such account is right around the corner, either, or that it's doable only with non-technical terminology, or mostly non-technical terminology – unless, perhaps, one is willing to have a very long and convoluted system.
Second, it seems that we shouldn't assume, just because of the objectiveness of morality, that an ontological account of morality, moral properties, etc., would account for moral properties in terms of properties that would necessarily be instantiated, or entities that would exist necessarily, or even that always existed.
The point is that we should not assume that a lack of instantiation of a property such as moral goodness – i.e., the lack of morally good agents or actions -, in some possible scenario is not a problem for moral objectiveness in the sense which is relevant here: namely – and to use two examples Craig brings up -, that the Holocaust was immoral regardless of what anyone believed, a man who tortures children for fun is morally bad, etc.
While a theist might claim that, because of the semantics of moral terms, moral properties are different from color properties in some significant sense, that would have to be argued for, and the burden would be on the claimant.
My point here is that that it shouldn't be assumed just from objectiveness that moral properties were always instantiated.
Third, it seems there is no good reason to think that future science can't be helpful in discovering moral truths – as long, of course, as there are such truths.
On this point, someone might insist on the usual claim that science tells us what is, not what ought to be – or something like that.
However, if science can be used to discover what is morally good, or what is immoral, etc., the moral 'ought' follows at once.
But there appears to be no good reason to assume that science can't find answers to questions about goodness, immorality, etc., as it can in the color case.
Again, those claiming otherwise would have to justify the difference.
Granted, just as the science of color needs to trust human color vision, at least in controlled conditions, the same would be the case with regard to the human moral sense.
However, that alone is not an objection to the use of science to elucidate moral questions – else, it would be an objection also in the case of color, or light, or pretty much anything.
Given morality's greater complexity, we should expect that a science of morality would be a lot more difficult than a science of color, but the 'ought' vs. 'is' objection has nothing to do with that, and it seems the objection based on it amounts to a baseless assertion – again, as long as there is objective morality.
Alternatively, someone might also say something like: 'You can make a machine that detects green objects. But how would a machine ascertain whether a behavior is morally good?'
That's more a question related at most to moral epistemology, which is not what Craig's argument is about, but we can consider the matter:
It seems that humans often can ascertain whether a certain behavior is morally good, and they do that in a clearly finite number of steps.
If future scientists can figure out what that algorithm is – for instance -, and/or come up with a more precise description of goodness, etc., than that given by our usual moral terms, and develop an alternative algorithm, then a future supercomputer would probably be able to ascertain what's morally good, bad, etc., much faster than any human could, without the difficulties associated with human weaknesses and propensities that the computer would not have, and even in cases that would be very difficult for humans.
Of course, it's also conceivable that no future scientists will ever figure that out.
However, the point is that apart from the difficulty of its complexity, there does not appear to be anything particularly salient in the case of morality (or moral goodness, etc.) that would make it any more beyond scientific understanding and/or detection than, say, color is.
By the way, the above (i.e., the supercomputer, etc.) does not seem to be even incompatible with theism.
Fourth, the fact is that we already have words to describe moral properties, and that's precisely our usual moral terms.
There appears to be no pressing reason to come up with an alternative way of describing them – apart from no good reason to think we can do so in the short term, given the previous points about the difficulties involved.
Granted, it might be useful to have a more fine-grained description.
It would probably be very useful to have a supercomputer like the one I described above.
However, none of that is any cause for concern for the non-theist, who does not have any burden to present any ontology of morality, for all the aforementioned reasons.
Fifth, in particular, if a theist presents an ontological theory such as some Divine Command Theory (DCT), it's up to the theist to defend it, and the non-theist has any burden to present an alternative.
Of course, if the theist can successfully argue for DCT, that would establish theism in some of its variants (which one would depend on how 'God' is construed, and what that particular version of DCT actually says), but showing that DCT succeeds is, again, a burden exclusively upon the theist, and that's a burden that cannot properly be discharged just by pointing out that the atheist has no alternative hypothesis.
Moreover, just as – if no one had come up with the Theory of Evolution yet -, a theist who posited a biblical creation story as a hypothesis about the origin of humans wouldn't be in an epistemically better position just because he has a hypothesis, neither would having a hypothesis – such as DCT -, on its own, put the theist in a stronger epistemic position at all.
In fact, if the theist can provide no sufficient reasons to justify his belief that his hypothesis is true, then his position is far weaker than that of the person who simply recognizes that they do not have any such hypotheses.
Sixth – though perhaps not important, but just for the sake of thoroughness -, we shouldn't assume that an ontological account needs to semantically close moral questions, in order to be correct.
For instance, the question 'I know it reflects such-and-such wavelengths, but is it actually green?' seems to open, at least from a semantic point of view:
Even if we specify the 'such and such', and even if a person grasps both the non-technical color terms and the technical physics terms, they would still need to do some experiments – or study those already done -, in order to learn the connections between the two; it seems they can't make the connection just by means of conceptual analysis.
So, generally speaking, open questions are not so problematic for ontological accounts put in technical terms.
Perhaps, someone might suggest that the case of morality is different in a relevant sense in that regard, given the difference in the way we assess, say, goodness vs. greenness.
However, that would have to be argued for, since it seems it does not follow from the general conditions for having an ontology that questions of the sort must be semantically closed.
While Craig does not claim in this context that 'X is immoral' means something like 'X is forbidden by God', and in general is not concerned with moral semantics but with moral ontology in the context of his metaethical argument, I will dedicate this section to some moral semantics, since that will be relevant to my counterpoints later on.
Now, I will not make any claims about what 'X is immoral', 'X is morally good', etc., mean; rather, I will make claims – and will support them – about some of the things that such terms do not mean.
Somewhat more precisely, in this section I will argue that 'X is immoral', does not mean 'X is forbidden by G' (or 'G forbids X'), or anything of the sort, for any creator, any conscious being G, etc.
I will argue that 'X is morally good' does not mean 'X resembles entity G', for any such entities, that 'X is morally obligatory', does not mean 'G commands X'; that 'A has a moral obligation to do X' does not mean 'G commands A to do X', and so on.
First, I will consider the case in which 'G' stands for 'God', since it's the most relevant here by far.
Now, the concept of God is unclear in the context of Craig's metaethical argument, but in any case, there are examples of groups of people that don't have any of the concepts of God used in present-day philosophy.
Granted, they often believed or believe in some powerful beings that are in English called 'gods' sometimes, or 'spirits', 'demons', etc., and they have the corresponding concepts, but those entities do not resemble anything that Craig or other philosophers would call 'God'.
Indeed, the concepts are (or were) very different: in many cases, there was nothing in the concepts they used that entailed omnipotence, or omniscience, or moral perfection, or even moral goodness, or that those entities were the creators of all things, or the greatest conceivable being, maximally great being, etc.
So, all human social groups use moral terms – at least, basic ones like 'immoral', 'morally good', 'morally bad' -, so humans clearly can and do grasp the meaning of 'morally good', 'immoral', etc. (or equivalents in other languages), even if they do not have a concept of God.
Now, if 'X is immoral' means something like 'God forbids X', how come all those people made claims of immorality while they didn't believe in God (in any of the usual present-day philosophical meanings of the word; I'll just say 'God' for short), or even have a concept of God?
Someone might suggest that they had a concept of God, but just didn't bother to invent a word to name it.
However, that seems implausible. How would they come about such a concept, in a way that would not prompt them to make up a word for it?
Still, even that would not be a problem:
If 'X is immoral' meant something like 'God forbids X', people would have immediately realized that they were affirming the existence of God all the time.
Someone might object that some semantic identities aren't transparent. That may be true, but that is not plausible in this case: if someone is actually making a claim that God issued a command forbidding some behavior, it seems difficult to see how they would all fail to realize that they're saying it.
To make the matter more concrete, let us consider a specific example: Japan.
Japan is a country in which there is no tradition of belief in God: the main religions – both traditionally, for a long time, and in the present – are Buddhism and Shinto, not Christianity, or Islam, or any other religion that posit the existence of God.
That fact has not changed: in fact, today as in the past, the vast majority of people do not believe in God – while different polling methods yield different results, all of them agree in that it's a significant majority. 
Yet, clearly, and with the exception of cases of severe mental illness, Japanese adults grasp the meaning of moral terms, can and do use them competently, etc.
Also, clearly, they're not making claims about commands or prohibitions issued by God, or anything of the sort.
So, given the previous considerations, we can tell that there is no semantic connection of the sort I described: in other words, it is not the case that 'X is immoral' means 'God forbids X", or anything of the sort.
Second, and going to the general case of any creator and/or intelligent entity G, similar considerations apply:
Someone might suggest that there is some property P such that 'X is immoral' means 'An entity with property P bans X'.
But that is not the case, since P clearly would not be a property any human being or less intelligent being has, and there are groups that do not have beliefs about superhuman entities, but nonetheless understand moral words, make claims that some behaviors are immoral, and so on.
As before, someone might object that some semantic identities aren't transparent, but again, f someone is actually making a claim that there is a being issuing a command and forbidding some behavior, it seems that they would realize that that's what they're saying.
As for other moral judgments, such as 'X is morally obligatory', 'X is morally good', etc., similar cases can be made, so I will not repeat the arguments, but will provide more examples:
First, there are many theists, who believe that Yahweh had or has a moral obligation to honor his covenant with the ancient Hebrews.
It is apparent that their belief that Yahweh has a moral obligation to honor his covenant with the ancient Hebrews is not a belief that someone commanded Yahweh to honor his covenant with the ancient Hebrews.
Second, there are also theist philosophers who say that God has moral obligations; for instance:
Richard Swinburne: 
God has a moral obligation to make himself known
Swinburne is most certainly not saying that God commanded God to make himself known, or that some other being commanded God to make himself known.
As I explained earlier, Craig is concerned with moral ontology, not with moral semantics.
However, I will describe a crucial connection between ontology and semantics, and explain how that affects Craig's metaethical argument.
Earlier, I made some points on color ontology, as an analogy to moral ontology; in this subsection and the next one, I will use that analogy again, so illustrate a few more issues that are relevant to understand what it is that the theist defender of the metaethical argument would need to show, and generally what the burden on each side is.
First, it seems the property 'greenness' is a property of some objects and/or light that our color vision perceives or tracks; in any event, it's a property tracked by our visual sense.
Second, when an ontological account is given, the proposed account actually makes a list of categories of objects that have that property.
We try to make the complete list, such that any object that has the property 'greenness' is in at least one such category.
For instance, an object can be green because it, say, under certain conditions it reflects light in some wavelength l1, or a combination of wavelengths l2 and l3 that our color vision perceives very similarly.
All the wavelengths posited in the theory have to match our color vision – i.e., a normal human color vision is the guidance.
Third, there is no one single account: if we find a more general description of the universe than present-day physics, we could come up with a description in those terms, and that too would be correct, if the description in terms of wavelengths is.
Fourth, even if the account in terms of wavelengths I suggested above is completely wrong, there is still some property that our color vision tracks, and which under normal conditions elicits our judgment 'green', and that property is the property 'greenness'.
The point is that as long as there is some property that our color sense is tracking – whatever that property is -, and which elicits our judgments 'green' under normal conditions, that appears to be in general sufficient for having objective color green (in the sense of 'objective' that is relevant in the context of the metaethical argument), and unless the semantics of the word "green" prevented that, for some specific property.
For instance, if 'green', meant something like 'that which evokes a certain feeling in normal humans, and which was created by Odin', then even if our visual system is tracking a property, and even if the perception of it is what normally elicits our judgment 'green', that property wouldn't be greenness, because Odin does not exist, so we would have an error theory of greenness – in other words, nothing would be green; obviously, there is no Odin requirement in the word 'green'.
It's important to note that issues such as how our visual system came to be, its inner workings, why it tracks the property that it tracks, but not another one, etc., do not matter with regard to whether there is objective color, and of course there is no need for a Supreme Green Commander, as long as the semantics of the word 'green' do not require it – which, of course, they do not.
Now, let's take a loot at moral properties:
As in the case of color, let's say - which Craig accepts - that we have a sense that tracks certain properties, and when we perceive them (not necessarily in person, but considering hypothetical scenarios, etc.), that perception (or understanding, if you like) elicits corresponding judgments like 'morally good', 'immoral', 'morally obligatory', and so on.
Again, it seems, as in the case of color, that as long as the system is species-wide, and it tracks some properties, then we would have objective goodness, objective immorality, etc.
In other word, in that case, morality is objective in the sense of 'objective' in the sense that is relevant in the context of this metaethical argument, regardless of things such as how our moral sense came to be, and what the properties that our moral sense is tracking actually are, except for semantic constraints: For instance, if our moral sense is tracking some property, and when we perceive such property it normally elicits our judgment 'morally obligatory', but when we say 'X is morally obligatory' we mean something like 'Thor commands X', that would give us a moral error theory (of course, 'X is morally obligatory" does not mean 'Thor commands X", or 'God commands X', for that matter).
So, what would those properties be?
Actually, we already have a language for them – precisely, moral language -, so we can simply say those are the property of moral goodness, moral wrongness, etc.
But if one wants to say something more about them, then one might suggest that they may be complicated mental properties, involving attitudes, intentions, dispositions, etc., and in the case of actions that are good, bad, etc., those carrying out the actions would have such mental properties.
However, and regardless of whether that is the case, the crucial point is that there is no need for me or any non-theist to come up with a theory about them: whatever the properties that our moral sense is tracking are, those are the moral properties, as long as there is no semantic barrier to that.
In other words, as long as the meaning of moral terms does not involve ontological claims that the properties that our moral sense tracks do not match, then there are moral properties, regardless of what they might be – well, regardless of what their description in non-moral terms would be; the properties are actually moral goodness, moral wrongness, etc.
So, the previous conditions are enough, unless there is some semantic restriction that gets in the way.
However, arguing for such a restriction requires arguing moral semantics, and the burden to show that that there is a restriction that affects only non-theism is on the theist defender of the metaethical argument.
In other words, the theist arguer would have to show that if theism is not true, then the properties that our moral sense is tracking – regardless of what they actually are -, do not meet the ontological commitments entailed by moral language, but if theism is true, then the properties that our moral sense is tracking do meet such commitments.
That's a burden that theist defenders of metaethical arguments – including Craig – come nowhere near meeting.
Burden aside, a non-theist may of course choose to go on the offensive and argue that some or all the potential semantic objections fail. In particular, earlier I showed that moral judgments do not involve claims about a commander of any kind, or a creator, etc.
Based on all the reasons explained in previous sections, we can tell that a non-theist moral objectivist may simply posit a species-specific moral sense – just as there is color vision, for instance -, and claim that moral properties are whatever properties that moral sense tracks, and which our moral language refers to.
She may just claim that the conditions I mentioned earlier are met, and that is it.
Now, as I explained earlier, ontological challenges would fail, in absence of a successful semantic challenge.
But there are other potential challenges, so let's take a quick look at the main alternatives:
Semantic challenges that the non-theist objectivist might face can be divided in two main groups:
a) Semantic challenges to moral objectivism:
Those challenges do not concern us here, since a theist can't challenge use them to support their metaethical case.
b) Semantic challenges to the compatibility of moral objectivism with non-theism which do not challenge the possibility of moral objectivism under theism.
These are the kinds of semantic challenges that the theist defender of a metaethical argument might want to raise.
Given that, it's hard to see which challenges the theist still has at his disposal, but in any event, the burden would be on the theist claimant.
Here, too, we may distinguish between two main types of challenges:
a) Empirical challenges to moral objectivism:
Those challenges do not concern us here, since a theist can't use them to support his metaethical case.
b) Empirical challenges to the compatibility of moral objectivism with non-theism which do not challenge the possibility of moral objectivism under theism.
Once again, these are the kinds of empirical challenges that a theist might want to raise, but it seems clear that the range of challenges is very limited, and at most they would be challenges for some types of non-theism; in any case, the burden of arguing for them is on the theist.
A possibility for the theist would be to argue that if theism were not true, then either humans wouldn't have a species-wide sense but one that varies from person to person, or that they would have some kind of species-wide sense, but it would be very different from what it is, so it wouldn't track the properties that it actually tracks, but some other properties.
However, that kind of argument would be a heavy burden on the theist; I seriously doubt they can meet that, but in any case, here is enough to point out that they have not.
Another potential theistic tactic would be to claim that a good number of non-theist philosophers deny that there is objective morality.
Of course, that tactic goes nowhere:
First, there are also plenty of non-theist philosophers who do not deny that there is objective morality, including plenty who affirm that there is.
Second, many of those who deny that there is objective morality make arguments to the conclusion that objective morality is actually impossible – rather than possible only on theism -, so theists can't consistently use those arguments in their favor.
Third, even theist philosophers are divided on whether premise one is true.
Fourth, in any case, that wouldn't matter.
A non-theist objectivist who understands the matters at hand realizes that she can just posit a human universal moral sense and moral properties tracked by it, and has no extra burden; why should she care about arguments from authority?
Having surveyed the main potential objections and arguments available to the theist, let's now assess Craig's argument in detail.
Even though Craig is concerned with moral ontology rather than moral semantics, Craig contends that, on atheism, there are no moral obligations of prohibitions because there is no competent authority to issue moral commands or prohibitions:
William Lane Craig: 
Moral obligations or prohibitions arise in response to imperatives from a competent authority. For example, if a policeman tells you to pull over, then because of his authority, who he is, you are legally obligated to pull over. But if some random stranger tells you to pull over, you’re not legally obligated to do so.
Now, in the absence of God, what authority is there to issue moral commands or prohibitions? There is none on atheism, and therefore there are no moral imperatives for us to obey.
Yet, Craig does not provide any good reasons at all to even suspect that having an obligation would require having an authority issuing commands.
Note that the burden is on him.
It's not up to the non-theist to show that that is not the case, and that obligations do not require someone issuing commands.
Now, the fact is that there is no semantic requirement.
Now, Craig's argument is an ontological one, not a semantic one.
However, as I explained earlier, the non-theist objectivist is not challenged in the least by an ontological challenge in absence of a successful semantic challenge.
Why should we even suspect that without such a competent authority issuing commands, moral obligations would not exist?
Granted, someone might posit that such a being – that we might call a 'Supreme Commander' - exists, and then claim that that would somehow provide an adequate 'ontological foundation' for morality.
I will later show that that is not true, but leaving that aside for the moment, that suggestion would make no progress whatsoever towards meeting the theist's burden in the context of this metaethical argument, since it would not present any challenges to the possibility of moral objectivism without a Supreme Commander.
Furthermore, the person making such a suggestion would have the burden of showing how such a Supreme Commander would provide an adequate 'ontological foundation' for moral obligations, since he would be the one making the claim in an attempt to support the first premise of this metaethical argument.
But even if he managed to show that the Supreme Commander, if he existed, would provide the adequate 'ontological foundation', for moral obligations – which is false, but leaving that aside for now -, that alone in no way would imply that such a being is required for moral obligation, so even that would not provide support for the first premise of this metaethical argument.
Finally, Craig's police officer analogy does not work, either:
it's plausibly not meant to suggest a semantic
since Craig's argument is an ontological one.
Still, some of Craig's quotations of Richard Taylor suggest that Craig might be raising a semantic challenge after all.
It's not entirely clear, but in any case, I will simply point out that if the police officer analogy is an attempt to introduce a semantic challenge, suggesting that moral obligations entail a Supreme Commander by the meaning of the words, then the challenge fails, for the reasons I explained earlier.
Second, if it's not meant to suggest any semantic requirement, then Craig provides no good reason to suspect that morality and legality are indeed analogous in the case under consideration.
So, why should we suspect that they are?
So, in brief, Craig's argument about a "competent authority" provides no good reason to even suspect that a lack of a Supreme Commander would be in any way a problem – let alone that any Divine Command Theory might be true.
First, a clarification is in order:
There are several theories, about moral ontology and/or moral semantics, that may be called 'Divine Command Theory'.
In a previous section, I've shown that semantic DCT – which posit that 'Agent A has a moral obligation to do X' means 'God commands agent A to do X', or something along those lines – are not true.
In this section, I will show that ontological DCT are not true.
In other words, I will show that theories that claim that moral obligations/duties are constituted by the commands of God – which is what Craig claims in his debate with Harris – are not true, regardless of whether they also are semantic DCT.
It's not entirely clear what 'constituted' means in this context, since moral obligations aren't some entity that may have a certain composition, and the claim is not one of semantic equivalence, either. Still, I will let that pass and in any case present an objection that succeeds in spite of the obscurity of the claim, since it works under any plausibly understanding of the word 'constituted'.
Now, before I go any further, I will point out that there is no burden on the non-theist to show that DCT are not true.
Rather, if a theist claims that a DCT is true, it's his burden to show that it is.
In particular, this section 9) is not at all required to properly conclude that Craig's metaethical argument provides no support for theism, since Craig does not provide any good reasons to even suspect that the first premise might be true, for the reasons explained in the rest of the sections.
So, even if we ignore the arguments I will make in this section, the case against Craig's metaethical argument succeeds.
Still, since Craig posits a DCT, and since refuting DCT also has interesting implications for other theistic metaethical arguments and hypotheses, I now proceed to do so:
Before I address the heart of this matter, I will address a specific issue about metaphysical possibility, in order to preempt certain potential objections.
Let's consider the following hypothetical dialogue:
Alice: Water is H2O.
Tom: I don't believe it.
I believe that that's a scientific conspiracy.
Water is not H2O, but Hg2Po.
Alice: What? Hg2Po? That's absurd!
Tom: That's easy to say, but do you have any evidence?
And don't tell me to look at papers or textbooks. They're all in on the conspiracy.
If you want to persuade me that water is not Hg2Po, then show me that it is not, and then maybe you can try to convince me that it's H2O.
Alice: Hmm...let's see: Do you know what the composition of sulfuric acid is?
Tom: Yes, that one is real. Sulfuric acid is H2SO4.
Alice: Good. Let's see:
If water were Hg2Po, then the molecule of water would be heavier than the molecule of sulfuric acid.
Maybe we can use that to test the theory.
Bob: I'm sorry, Alice, but that's impossible.
I agree with you, of course, that water is H2O.
However, given that water is H2O, it's metaphysically impossible for water to be Hg2Po, so your conditional has a metaphysically impossible antecedent.
So, I'm afraid that you're constructing a metaphysically impossible scenario.
Tom: Well, water isn't H2O, but if it were, then Alice would indeed be constructing a metaphysically impossible scenario. So, Alice, your suggestion fails. Try again.
Alice: What are you two even talking about?
I'm not suggesting that Alice's test is a good one, or that entertaining Tom's absurdities is a good idea, either, but my point here is that Alice's claim 'If water were Hg2Po, then the molecule of water would be heavier than the molecule of sulfuric acid' is clearly a true claim.
The objections raised by Bob and Tom are very confused.
That it is metaphysically impossible for water to be anything but H2O has nothing to do with the truth value of Alice's conditional.
Incidentally, as a side note, if someone is a theist and shares Bob's confusion, he might object to Craig's metaethical argument on the grounds that the first premise has a metaphysically impossible antecedent, since theists usually hold that 'God does not exist' is metaphysically impossible.
If there is no personal creator of all other personal beings, then God does not exist. Then, it is not the case that our moral obligations are constituted by God's commands, and so DCT are not true.
So, let's assume in the rest of this subsection, and for the sake of the argument, that there is a personal being, creator of all other personal beings.
Let's name that being 'Alex'.
In other words, by 'Alex' I mean 'The personal being who created all other personal beings.'
I don't mean anything else by 'Alex'.
So, let's see that Alex actually has moral obligations.
To show that, my strategy is in a sense similar to Craig's strategy in support of the second premise, which appeals to people's intuitive assessment that, say, the Holocaust was morally wrong, that torturing a child just for fun is immoral, and so on.
So, Alex is a person who created all other personal beings.
My claim – which I would ask readers to please assess by their own sense of right and wrong – is the following:
It seems to me that P1 is obviously true.
I'm not even talking about eternal punishment in Hell – I maintain that that would also be immoral, but that's a matter for another article.
In the case under consideration in P1, there is no punishment for any sin; we're talking about a person creating personal beings with the specific, deliberate and exclusive purpose of torturing them for all eternity.
So, if Alex created other personal beings for the specific, deliberate and exclusive purpose of torturing those beings for all eternity, then Alex would be acting immorally.
Someone might object that if Alex is God, then the antecedent of the conditional P1 is metaphysically impossible.
However, that would be a very confused objection, as explained in the previous subsection.
Alternatively, they might claim that my argument is circular, because I'm somehow assuming that there can be objective morality without God.
But that would be confused as well: I'm not assuming that there can be objective morality without God. In fact, I am:
a) Assuming – or rather, granting – that there is objective morality.
b) Further assuming that Alex exists. In other words, I'm further assuming that the personal creator of all other personal beings exist.
c) I'm not making further assumptions about Alex; in particular, I'm neither assuming that Alex is God, nor that he or she is not God.
d) I'm using my sense of right and wrong to conclude that P1 is true – and asking readers to do as well.
In other words, I'm concluding, by means of my sense of right and wrong and under some assumptions that are entailed by theism, that if Alex were to create other personal beings for the specific, deliberate and exclusive purpose of torturing those beings for all eternity, then Alex would be acting immorally.
So, this objection would fail as well. There is no improper assumption or circularity on my part.
Now, I will appeal to the reader's grasp of moral terms, and claim that – just by the meaning of the words – P1 entails:
Readers will use their own grasp of the terms to make their own assessment, of course, but I contend that P2 follows from P1 just as 'Barack Obama is not a bachelor' follows from 'Barack Obama is married'.
So, Alex has a moral obligation.
But that moral obligation is not constituted by one of Alex's commands – hopefully, that is clear.
In other words, Alex's moral obligation not to create other personal beings for the specific, deliberate and exclusive purpose of torturing those beings for all eternity, is not constituted by Alex's 'command to Alex' not to create other personal beings for the specific, deliberate and exclusive purpose of torturing those beings for all eternity: there is no command from Alex to Alex.
Hence, it is not true that moral obligations are constituted by Alex's commands.
Now, if God exists, then God and Alex are the same person, since God is the creator of all other personal beings, and Alex is the creator of all other personal beings.
Therefore, it is not true that moral obligations are constituted by God's commands.
Therefore, Divine Command Theories are not true.
William Lane Craig: 
On the naturalistic view, human beings are just animals, and animals have no moral obligation to one another. When a lion kills a zebra, it kills the zebra, but it doesn’t murder the zebra. When a great white shark forcibly copulates with a female, it forcibly copulates with her but it doesn’t rape her–for none of these actions is forbidden or obligatory. There is no moral dimension to these actions.
First, a brief reply:
When a lioness kills a zebra, she doesn't fly.
And when a zebra escapes from a lioness, he doesn't fly, either.
Zebras, lionesses, gazelles, rats, cats – none of them flies.
So, mammals do not fly.
On a biologist's view, bats are just mammals.
But mammals do not fly, and many bats do.
That refutes biologists' claim that bats are just mammals...
Parodies aside, and leaving also aside Craig's contemptuous "just", the fact that some animals are not moral agents is no good reason to believe that no animals are moral agents.
Of course, there are obvious psychological differences between different species, and in particular, between humans and great whites, or any other species for that matter. 
What actually matters, when it comes to the question of whether a being is a moral agent, is not the ontology of the agent's mind, but the psychology of her mind.
The ontology could only have an indirect effect, to the extent to which it conditions the psychology.
I think that, after reflexion, that should be clear, but still, in case someone is not persuaded, let's consider the following scenario:
First, let's assume theism for the moment, and let's suppose that God creates a universe in which there are no souls.
Instead, he creates some sort of panpsychist universe, where there are some basic, essential particles with some sort of basic phenomenal consciousness, but no intelligence, no will, no pain, essentially much simpler than a mosquito's mind – they shouldn't even be properly called 'minds', I think.
Then, in that universe, through theistic evolution, complex beings arise.
They do not have souls, but they do have minds, with the full range of emotions, knowledge, etc., of the animals we're familiar with.
In fact, one of the species God creates has a psychological makeup similar to that of humans, and is capable of making moral assessments, just like we can.
Let's suppose that some of those beings engages in torturing other such beings for fun.
Would they not be acting immorally?
It's pretty obvious that they would be, even though their minds is made up of the same kind of basic stuff that the rest of the minds of the other animals in that universe – including all of those without a moral sense -, and even made up of the same kind of basic stuff as tables, chairs, and the like.
Of course, the tables, etc., do not have table-minds, etc: the basic particles that make them up do have the most basic phenomenal consciousness, but they're not combined in a way that makes up any less basic consciousness. They're only connected externally, so to speak.
So, chairs, tables, etc., are not moral beings – obviously -, and neither are mosquitoes or similar entities, but on the other hand, entities with human-like minds are moral beings, just as humans are, and regardless of what their minds are made of.
We may also consider an alternative scenario, just like the above but in which there is no phenomenal consciousness in the basic particles – i.e., no panpsychism -, and minds emerge when some kind of combinations of particles happen – God just made the particles with the properties that when they combine in a certain manner, they combination acquires awareness.
The conclusion is the same: beings with human-like minds are moral agents, regardless of the ontology of the mind.
So, as I pointed out, it's not the ontology of a mind what matters when it comes to the question of whether a being is a moral agent, but its psychology – or, to put it simpler, what matters is what kind of mind it is, not what kind of stuff the mind is made of.
Someone might insist that the scenarios in which there are people without souls are metaphysically impossible, or something along those lines. That would have to be argued for, but even then, that would be beside the point.
The scenarios still show that there is no semantic requirement of souls, or anything like that.
In other words, the meaning of moral terms is not such that if we say, 'Agent A acted immorally', we're implying that A has a soul, or in any case that the mind of agent A is made up of some kind of basic substance that is different from the kind of basic substances that chairs, tables, and mosquitoes are made of.
A theist might object to that point and claim that the meaning entails that, but the meaning is non-transparent, so competent users of moral terms will not notice it.
However, for that matter, someone can say that the meaning of moral terms entail that there aren't souls, but that that's non-transparent.
The point here is that there is nothing as far as one can tell that would even suggest that our moral language is such that souls are a requirement for moral agency, or that some ontological difference between the kind of basic stuff that chairs, mosquitoes, lions, and people are made of, is required for moral agency to exist.
Given that there is no semantic requirement for souls or any other such entity, then it seems that what would be left to the theist here would be an empirical challenge: namely, they might argue that if theism is not true, or at least that if naturalism  is true, then the kind of mind that we have would not have developed.
that would have to be argued for.
Moreover, if such an empirical challenge were successful, then the metaethical argument would play no role anymore – the conclusion would be obtained right away from the empirical case.
On the other hand, if such an empirical challenge is unsuccessful, then obviously it provides no support for the metaethical argument.
In line with his "just animals" argument, Craig actually considers the possibility of a moral sense resulting from the evolutionary process, and makes disparaging remarks about it, like calling that a “herd morality”,
But let's address his arguments, and see what the objection actually amounts to:
William Lane Craig:
As a result of socio-biological pressures, there has evolved among homo sapiens a sort of "herd morality" which functions well in the perpetuation of our species in the struggle for survival. But there does not seem to be anything about homo sapiens that makes this morality objectively true.
William Lane Craig:
If there is no God, then any ground for regarding the herd morality evolved by homo sapiens as objectively true seems to have been removed. After all, what is so special about human beings? They are just accidental by-products of nature which have evolved relatively recently on an infinitesimal speck of dust lost somewhere in a hostile and mindless universe and which are doomed to perish individually and collectively in a relatively short time.
"By-products of nature"?
One wonders what the main products would be...but in any case, let's begin with a quick reply:
1) As a result of environmental pressures, there has evolved among Homo Sapiens a sort of “human color vision”, which functions well in the perpetuation of our species in the struggle for survival. But there does not seem to be anything among Homo Sapiens that makes our color statements objectively true...
2) If there is no God, then any ground for regarding the statements about colors based on the color vision that evolved among Homo Sapiens as objectively true seems to have been removed. After all, what's so special about human beings?
Parodies aside, the fact is that strong rhetoric and displays of contempt do not provide any support for premise one of the metaethical argument.
The fact is that the claims from Craig I quoted above completely fail to even present any relevant objection.
Again, however our moral sense came to be what it is, is not relevant to the matter at hand.
As long as we have it and it tracks some properties, then those properties are moral properties, in absence of a successful semantic challenge.
And also, again, a theist could challenge that such a sense would develop under non-theism, or at least under naturalism, but that burden would be on him.
According to Craig, naturalists are usually physicalists or materialists, and he uses that to try to defend his claim that in that case, there is no objective morality.
Before I go on, I will point out that hundreds of millions of people do not believe in God, but are not materialists, physicalists, or naturalists (e.g., most Buddhists).
That aside, Craig claims that without a soul, humans are not qualitatively different from other species, and that prevents moral objectivism. I already explaned why the claims about "just animals", and the like, fail, so let's assess his other argument in this context – namely, the one about determinism:
William Lane Craig: 
Secondly, if there is no mind distinct from the brain, then everything we think and do is determined by the input of our five senses and our genetic make-up. There is no personal agent who freely decides to do something. But without freedom, none of our choices is morally significant.
While is true that freedom seems to be required for moral responsibility, it's not true that determinism is a problem for freedom.
I will address that in the following section, but in this one, I will argue that even assuming that libertarian free will is coherent and actually is a correct understanding of freedom, there is no good reason to think that that would be a problem for the non-theist.
In fact, there is nothing in non-theism or even in materialism that entails determinism, and some interpretations of quantum mechanics are non-deterministic.
Someone might suggest that that quantum non-determinism does not provide the adequate kind of non-determinism to allow libertarian freedom.
In reality, non-determinism might hinder but never help freedom, and libertarian 'freedom' is no freedom at all, but leaving that aside and assuming otherwise, there is no good reason to think that quantum mechanics prevents brains from having whatever kind of freedom souls are supposed to have.
To see this, let's say that there are particles, not souls, and the universe is indeterministic.
Now, the fact remains that humans have minds. We can love, believe, feel, and so on.
So, the conclusion is that particles can interact with each other in ways that result in minds.
Hence, it remains the case that not all the properties of particles are those described by present-day physics, since present-day physics says nothing about either forming minds or interacting with them.
In addition to that, it remains the case that not all the properties of particles that are causally effective are described by present-day physics – particles have the property of being capable of forming minds, for instance.
Moreover, any arguments and/or evidence that we – by assumption – have in support of libertarian freedom would remain, since modern physics says nothing about minds, or what kind of freedom they may or may not have.
Someone might claim that only non-mental properties of particles, or of combinations of particles, would be causally effective in that scenario. But that would have to be argued for.
Else, why should we suspect that causal efficacy is limited to non-mental properties – some described by present-day physics, some not – of particles and of certain configurations of particles, against the obvious evidence that we can cause events by, say, making choices?
Whoever claims that brains made of particles and no souls would somehow prevent freedom would have to make their case, rather than just claiming so.
As it stands, and for the aforementioned reasons, there appears to be no good reason to suspect that that would be the case.
Also, the fact that quantum mechanics could be used to make probabilistic assessments about human behavior given sufficient information and computing power would not be a problem, either:
For that matter, we can make probabilistic assessments about human behavior all the time, for instance using economic models, psychological profiles, etc., or even intuitive assessments in daily life, and that does not suggest in any way that we have no freedom – even under the false assumption that freedom is libertarian freedom.
As I explained in the previous section, Craig's arguments – or similar ones – fail to show that EN, materialism, or physicalism entail determinism and/or a lack of freedom, even assuming that a libertarian free will hypothesis is a correct account of human freedom.
In this section I will show that libertarian free will is not freedom, and should more properly be called 'random will'.
Let's consider the following scenario:
Alice has been a good police officer for ten years.
She's kind, committed her job, good to her children, and so on.
Now, one morning, Alice goes to work as usual.
The police get a call about a domestic disturbance, and Alice and another officer are sent to the address they're given.
When they arrive there, they encounter Harry, a thirteen-year old kid high on drugs, acting completely irrationally.
He tells Alice: 'You're a police officer, so you're evil. Why don't you shoot me?'
Alice has no reason at all to shoot Harry.
He poses no threat to her, and can be easily arrested if needed.
However, it's clear that she has the power to shoot him, and is free to choose whether to shoot him.
All she'd have to do is pull her gun, point it at Harry, and shoot.
No one would see that coming, so no one could stop her if she did that – no human, anyway; the point is that she wouldn't be stopped.
But Alice – of course – feels no inclination whatsoever to shoot Harry, does not shoot him, and follows procedure.
The point is that saying that Alice can shoot Harry, that she has the power to shoot him, that she is free to choose whether to shoot him, etc., means that she would shoot him if she chose to do so, that she's not being coerced, etc.
It does not at all mean that, even given Alice's mental state at the time she chose to follow procedure, and even given all the conditions of the world at that time and previous times – including Alice's goals, beliefs, character, etc. -, it was still possible that Alice would shoot Harry.
On the contrary, if, given all those previous states, it was possible that Alice shot Harry, then it seems that there is a possible world W with the exact same past as ours prior to Alice's decision to follow procedure, at which Alice shot Harry instead.
But that is not an exercise of freedom, in the usual sense of the words. Rather, it's an unfortunate event that happens to Alice.
To see this, let's consider Alice's mental processes leading to her 'decision' to shoot Alice – say, decision D. Alice never considered shooting her, and had no desire, intention, etc., before decision D happened.
However, at some time, earlier states of the world, including her earlier mental processes did not determine her later mental processes. There is an event "Alice decides to shoot Harry" that happens irrespective of any previous states of Alice's mind, and no matter how much Alice would loath being a murderer.
All of Alice's previous reasoning, desires, behavior, intentions, etc., are incapable to stop 'decision' D from happening. But how's that Alice's decision?
It seems D is not a decision Alice made, but rather, it's something that happened to Alice.
It's not something Alice could have anticipated, or prevented: at some point her mental processes changed from normal to 'shoot Harry', without forewarning, and without any cause in previous mental processes.
Someone might claim that necessarily, there is always some hidden reason to shoot people, or to do anything one can do, but that would have to be argued for, and even then, that would not change the fact that, in that case, Alice could not have prevented his mental processes from changing at some point from normal to 'shoot Harry', no matter what she did before – and that change could not be reasonably said to be her decision, since she had never considered that before, and the change took her by surprise.
Those considerations show that that kind of thing should not be called 'freedom', but more like 'an unfortunate kind of randomness'.
That does not mean that human non-determinism isn't true. But that is surely not required for freedom, and in fact, it might undermine it, as the previous scenario shows.
Perhaps, there are situations after which, after assessing the pros and cons, a human is undecided between A or ¬A; if so, maybe there is a truly random outcome generator for such cases (which might involve also several mutually exclusive options: A1, A2, A3,...)
However, if that is the case, that is not required for free will: a random generator that delivers 'decisions' in cases in which the mind remains undecided clearly does not result in more freedom than a mind that actually makes decisions.
So, if there is such indeterminism, as long as the indeterministic events happen when a person is undecided (based on her previous feelings, desires, reasoning, etc., she is undecided and does not cause any outcome), maybe that randomness is compatible with free will, but that's all.
On the other hand, if there is an indeterministic feature of human behavior that happens to be like Bob's example above – i.e., if it happens against everything that the person stood for, his previous considerations, etc. -, then, and as the previous example shows, that kind of indeterminism – at least, when it happens – would actually preclude free will; rather, the 'decision' would be an unfortunately random will.
There is another way to see this, taking into account that even under the exact same preexisting conditions – including, of course, the previous mental states of the libertarian-free agent.
So, let's consider the following scenario (relativizing time as required):
Alice is a libertarian-free human, and at t(s), the state of worlds W and W' is exactly the same – that includes, of course, Alice's mental processes.
Later, Alice libertarian-freely chooses A at W, and B at W', even though the states of the worlds prior to Alice's decision were the same (A is different from B).
In other words, W and W' are exactly the same until Alice's mental processes diverge.
Now, let p be a Planck time, and n a non-negative integer, starting with 0.
Let's consider times t(s)+n*p, and the states of W and W', W(n) and W'(n) respectively.
Let n(l) be the last n such that W(n) = W'(n).
Since the 'decision' was made even given the exact same prior conditions, it seems that the 'decision' happened between t(s)+n(l)*p, and t(s)+(n(l)+1)*p = t(s)+n(l)*p+p, in other words, the 'decision' was made in not more than a Planck time.
That's way too fast for any human conscious decision, though. So, it becomes clearer that the first indeterministic event E that distinguishes between W and W' is same random alteration of Alice's mental processes.
Someone might suggest that, previous processes in her mind made E in probable, but weren't enough to bring it about, something still altered her mind randomly; let's assume that that would be a coherent interpretation of probability (else, this objection fails already).
Even then, the fact would remain that her mind was altered without a cause, and with nothing she could do earlier to stop it; moreover, in some cases, the improbable 'decision' might happen. And in those cases in which the improbable 'decision' happens – i.e., the decision that her previous mental processes made improbable – we're back with something like the unfortunate case of the libertarian-free police officer.
Someone might still object that, if such a random change in her mind happened, she still could have changed her mind, and refrained from carrying out the decision – in the case of the first example, the shooting.
The problem is, though, that if you can have such a random event between t(s)+n(l)*p, and t(s)+(n(l)+1)*p, it seems you can have another one at every single Planck time that follows, until the "decision" that was completely against everything the person previously stood for, actually happens.
But let's suppose someone introduces some fuzziness in some way – which they would have to explain, of course; else, the previous reasoning stands. Even then, the fact would remain that the agent would have a random component – a change in her mind she can't bring about, because it happens no matter what she tried previously; it's just that we wouldn't be able to see that by means of analyzing the process step-by-step, but all of the other reasons I've given above remain.
A theist might say that that's 'actually the agent acting', or something like that but – whatever that means -, the fact would remain that that would a partially random agent acting, not one in which mental processes are sufficient to bring about behavior; it would be an agent with a randomly altered mind – i.e., a mind that suffers some alterations that have no sufficient causes; it's akin to dice-throwing, and in some cases, it might go against everything the agent had stood for up till then.
So, for all of the previous reasons, the claim that non-determinism is required for freedom ought to be rejected.
That does not mean we can't act of our own accord, of course. We can and sometimes do have freedom; it's just that indeterminism is not required for that.
Now, there is an objection available to the theist, which seems to be Craig's position: namely, that is lack of causal determination that is required for freedom to exist, not lack of determinism.
However, if an event is determined by previous conditions, then it seems it's causally determined too, since some the previous conditions would be causes. How would it be otherwise?
It seems puzzling.
But regardless, we can make a case against the requirement of causal indeterminism independently. If causal indeterminism is true, then no matter what Alice does up to some time t, all of her thought processes, intentions, desires, memories, reasoning, are all insufficient to bring about her decision. So, it seems that the "decision" might just happen to her, and she might still shoot Harry.
Again, the theist might say that that's the agent acting. But how can she act so quickly – indeed, instantaneously?
Regardless, as I explained in the previous section, whatever the correct account of freedom is, if a soul can do it, particles can do it as well.
Another point Craig raises is based on the possibility that someone can get away with evil – Craig calls that a 'practical argument' for believing in God.
This is not a part of the metaethical argument under consideration, strictly speaking, since Craig does not claim that it supports the premises, as far as I know.
However, he makes this and other arguments in the same context, so I will address them anyway.
According to Craig, if there is no God, there is no moral accountability, and it does not matter how we live. 
However, even without any kind of afterlife, Craig's claim is false: our choices affect our future, and the future of others; they can cause happiness, suffering, etc., to us and/or to other people, and of course that normally matters to us and to others.
Moreover, there is in many cases moral accountability even if God does not exist, like bank robbers going to prison.
So, there does not need to be an afterlife for justice to be done in many cases.
There may not always be accountability, but there is in many cases.
That said, the previous considerations, while correct, are actually minor in this context, since we're talking about a 'practical argument' for belief in God, and that's epistemically disastrous.
It would be irrational for a person to come to believe that God exists just because they conclude that there is no accountability in all cases without God, and that makes that person feel sad – or however it makes them feel.
That would be some kind of wishful thinking.
The fact that having belief B would make a person feel better does not warrant having belief B.
In fact, it's not clear to me how this is even psychologically doable.
How would someone trying to engage in that kind of wishful thinking go about it?
Bob: Let's see: I do not have the belief that God exists, or enough reasons that would convince me of that. However, without God some people will probably get away with evil, and that is disheartening. So, from now on, I will believe that God exists.
That's just not doable – i.e., I don't think that that would actually result in belief.
Someone might suggest some kind of Pascal Wager-style conversion, in which people practice the rituals of a religion in order to somehow gradually convince themselves that said religion is true.
I'm not sure that that would be doable.
At least, I'm pretty sure that for many of us, it wouldn't be, though it might be for others.
However, in any event, that would be another irrational course of action.
To be fair, Craig does not attempt to use the practical argument alone.
Instead, he proposes to use practical arguments to "back up or motivate" the acceptance of what he believes are sound theoretical arguments.
However, that would be irrational as well: if the person has not been persuaded by the theoretical arguments, they would still be engaging in wishful thinking to come to believe in God.
In short, it's still an epistemic nightmare.
In addition to that, of course, Craig provides no good reason to even suspect that the first premise of his metaethical argument might be true, and thus no good reason to even suspect that his metaethical argument might be sound...and that's without counting the fact that Divine Command Theories are not true...
Another 'practical argument' Craig gives is based on the issue of motivations for doing the right thing.
Of course, as in the case of the previous 'practical argument', it would be irrational to believe on account of this.
That aside, Craig contends that sometimes self-interest is in conflict with morality. 
Of course, Craig is using 'self-interest' in a way that excludes a person's interest in doing what's right, simply because it's the right thing to do.
That usage is common, so it's not a problem, but we need to keep in mind that that's what's meant by 'self-interest'; it's not the only interest people have, of course.
Because of their own psychological makeup, human beings are motivated to do the right thing; that's also one of our interests, even if not covered under the label 'self-interest'.
That motivation may be defeasible, but it's there, with the possible exception of some psychopaths.
Moreover, it seems to me that in order for an action to be morally good, motivation counts.
For instance, it seems clear that helping people out of fear of damnation would not be morally good. It wouldn't always be morally wrong, either. But it wouldn't be morally good.
It is true, though, that fear can prevent some people from behaving immorally.
On the other hand, if we're engaging in considerations such as this one – which have nothing to do with whether God exists or whether he's required for objective morality -, then we should also assess the likely consequences of coming to believe that God exists.
The fact is that, in addition to the irrationality of adopting a belief for practical reasons, usually such beliefs are not just some kind of unspecified theism, but some version of Christianity or Islam, with all the baggage of false beliefs – including false moral beliefs -, attached to them.
False moral beliefs tend to cause people to behave immorally, believing that they're doing the right thing. Someone might point out that some non-theists have engaged in terrible behavior, perhaps in the name of communism or some other ideology.
That is true, but the point remains that adopting false moral beliefs generally results in more immoral behavior, regardless of whether the false ideology the false moral beliefs come from is religious or not.
In practical terms, people who become non-theists are not likely to engage in such actions, whereas people who become theists are likely to become Christians or Muslims aren't likely to kill in the name of their religion, either, but are likely to follow some of its false moral teachings, so conversions do tend to have such negative consequences.
Still, none of that is the main point here.
The main point is the irrationality of using so-called 'practical arguments' for belief, regardless of what the belief is about.
Craig uses strong rhetoric in defense of his metaethical argument, but he actually does not provide any good reason to even suspect that the first premise of his argument is true.
In this case, he does not defend the full metaethical argument, but only the first premise.
 At least not if we assume theism is compatible with the existence of morally imperfect beings, humans, etc., but the evidential problems of evil and suffering are beyond the scope of this article, so let's leave them aside for the sake of the argument; the point is that the supercomputer scenario does not appear to add any new difficulties for theism.
 Someone might extend the term 'Divine Command Theory' to other theistic metaethical frameworks, even if commands do not play a central role. In this article, the term is limited to either the kind of semantic or ontological theories that I've described.
Else, let's pick another example, as long as the concept of metaphysical necessity is coherent – if it isn't, the objection I'm preempting fails just because of that.
That does not mean that humans are the only animals that are moral agents.
We know enough about evolution to know that humans evolved gradually from other species, and so there is no clear-cut line.
Some other hominids probably had some moral awareness, and were capable of wrongdoing – i.e.. they were moral agents.
Whether some extant animals – such as bonobos or chimpanzees – are moral agents as well might be debatable, but there is no need to get into that – we may as well assume that they are not.
However, I will leave that aside for the sake of the argument, and show that in any event, Craig's points do not pose any challenge for the non-theist objectivist.