Download
in .pdf format (alternative
link)
Download
in .html format
Download
in .odt format
A
case against "The Existence of God": a reply to Swinburne's
case for theism
0.
Introduction
1.
Simplicity, scope, metaethics, and the probability of God
1.1.
Simplicity, scope and intrinsic probability
1.1.2.
Simplicity
1.2.2.
Scope
1.2.
The concept of God
1.2.1.
Freedom
1.2.2.
Moral perfection
1.2.2.1.
Some implicit assumptions
1.2.2.2.
Swinburne's account of morality
1.2.2.2.1.
An alien threat
1.2.2.2.2.
More aliens
1.2.2.2.3.
Omniscience and perfect freedom without moral perfection
1.2.2.2.4.
Morality, objective values, and reasons
1.3.
The intrinsic improbability of theism
1.3.1.
Simplicity
1.3.2.
Scope
1.3.3.
Consequentialist deities
1.3.4.
Estimating the intrinsic improbability of God
1.4.
The final improbability of theism
1.4.1.
More alternative deities
1.4.1.1.
Simplicity
1.4.1.2.
Probabilistic assessment
1.4.2.
Choice
1.4.3.
Gambling deities
1.4.3.
Simplicity, revisited
1.5.
Conclusion of Part 1
2.
Power, moral assessments and probability
2.1.
What can God do?
2.2.
What would God likely create, or refrain from creating?
2.3.
Conclusion of Part 2
3.
More probabilistic assessments, and more theistic improbability
3.1.
The improbability of God's creations: possible worlds as an example
3.2.
Theism and the probability of humans
3.2.1.
What's a 'type' or 'kind' of action?
3.3.
'Humanly free' agents, physicality, and dimensionality
3.4.
Dimensionality and some probabilistic assessments
3.5
Conclusion of Part 3
Notes
and references
0.
Introduction
In
this article, I will argue that the case Swinburne defends in "The
Existence of God" [1],
fails
to provide any support for theism.
The
article is divided in three main parts, each of which constitutes a
case against Swinburne's case for theism, and each of which is
independent of the two others.
In
the first part,
my
strategy will be to focus on the issues of simplicity, probability,
and some metaethical claims he makes, arguing that either the case
for theism is blocked by a division by zero, or the conclusion would
be that the probability that theism is
false
is
either 1 or arbitrarily close to 1.
In
the second part, I will challenge
Swinburne's take on power, most of his moral assessments, and part of
the reasoning he uses to try to back some probabilistic assessments.
In
the third part, I will concede for the
sake of the argument some of Swinburne's claims, and from that I will
argue from there that God does not exist.
In the course of my arguments,
and unless otherwise specified I will make assume for the sake of the
argument that the properties Swinburne posits (e.g., omniscience,
"perfect freedom") are coherent. That does not mean I'm
taking a stance on that.
Also,
I
will not challenge all of the controversial claims Swinburne makes.
For instance, I will not challenge the distinction between personal
and scientific explanations.
On
a terminological note:
a)
I will only refer to different
parts of this article as "sections" or "subsections"
- i.e., no sub-subsections, etc.
I
think links between the relevant parts of the article will prevent
any ambiguity.
b)
All references to page numbers are references to Swinburne's book.
[1]
c)
If I talk about dimensionality and dimensions without further
specification, I'm talking about spatial dimensions; in other words,
I'm not counting time.
Finally,
and as usual, I don't claim there is any novelty in the ideas on
which I base these arguments.
1.
Simplicity, scope, metaethics, and the probability of God
1.1.
Simplicity, scope, and intrinsic probability
According
to Swinburne, "the simpler a theory, the more probable it
is"[2].
Given
context, it's clear that he means that that's the case all other
things equal, since greater scope may reduce the theory's
probability, as Swinburne asserts [3].
However,
he also claims that greater scope does not have as much relevance as
the other criterion, since allegedly reducing scope also reduces
simplicity, due to the alleged introduction of arbitrary
restrictions.
However,
less scope does not
reduce
simplicity in all cases, all other things equal.
In
fact, it may increase it. For instance, let's consider the following
two hypotheses:
P1:
There exist at least 1000000000 rational particles (or substances,
etc.)
P2:
There exist at least 1000000000000000000000 rational agents (or
substances, etc.)
It
seems clear that P2 has greater scope – you're asserting more,
so you're more likely to be mistaken, following Swinburne's reasoning
-, but it's also less simple, since P1 postulates fewer entities.
In
any case, Swinburne leaves aside for the most part the criterion of
small scope because – he claims – typically a theory that
loses scope loses simplicity [4]
-
but even if that is true, leaving aside the criterion of small scope
requires that one already accepts simplicity as much more important.
In
defense of his claim that simplicity is much
more
important, Swinburne presents Newton's theory as an example of a
theory that, despite its enormous scope, was judged enormously
probable.
However,
that it was judged
enormously
probable as a theory that applies to everything is not a good
criterion. In fact, it turned out that it didn't apply to everything
– i.e., some predictions it makes are mistaken.
Moreover,
Newton's theory, or any theory in physics, does not seem to have
nearly as much scope as theism, which posits an entity that is a
creator of everything,
which
is omnipotent,
etc.
So,
even if simplicity were more important than scope, that does not mean
that any
scope,
no matter how massive, may be ignored.
That
said, let's take a closer look at Swinburne's criteria for assessing
simplicity and scope.
1.1.2)
Simplicity
In
the case of scientific explanations, Swinburne claims that the
criterion for simplicity is that the hypothesis postulates few
entities, kinds of entities, and "few and simple kinds of powers
and liabilities"[5]
He
offers as an example powers describable by simple mathematical
formulas.
His
criterion
for simplicity in the case of personal explanations is based on the
following conditions: [6]
Few
entities, few types of entities, few properties, few constant
intentions, and "continuing basic powers", and "simple
laws – constant predictable ways
in which persons acquire beliefs from their surroundings."
[6]
Yet, the complexity
of a mathematical description involved in the mental properties of
substances in personal explanations is nowhere to be found.
To illustrate the
point, let's suppose that we have a hypothesis H(1) that asserts the
existence of a universe U(1), which is described by some mathematical
model M(H(1)).
Similarly, let's
consider hypotheses H(n), for all natural n,
and the corresponding mathematical models.
As the models
increase in complexity, it seems, so do the hypothesis.
Let's now consider
the following hypotheses:
AH(G(n)): There
exists an entity who fully understands M(H(m)), for all m≤n, and
who is the creator of all things.
Now, clearly, for
all n, AH(G(n)) is no more complex than theism, since AH(G(n))
does not postulate more entities, kinds of entities, or more or more
complex powers and liabilities.
Moreover, the scope
of AH(G(n)) is much less than that of
theism.
Also, obviously it's
not less probable than theism – it's entailed by theism.
Now, let's consider
the following hypotheses:
H(n): There
exists a universe describable by M(H(n)).
For all n > 1,
it's clear that the mathematical description in H(n) is simpler than
the mathematical description in AH(G(n)).
Yet, AH(G(n)) is
considered much simpler than H(n) - since AH(G(n)) is simpler than
"God exists" -, in spite of the fact that the former would
include a more complex mathematical description.
In the case of
mental properties, it appears that, for some reason, Swinburne's
position is that no unpacking of mathematical models is needed when
considering simplicity.
If
we needed to unpack, that would make theism more complex than a claim
of existence of any universe,
no matter how complex its description might be – and that would
be enough to block Swinburne's case for theism.
But let's grant
Swinburne's criterion for the sake of the argument.
1.2.2.
Scope
Unlike
the case of simplicity, Swinburne does not say much about how to
assess scope, perhaps because he rejects that criterion as important
when it comes to assessing the intrinsic probability of theism and
rivals, which – he maintains – have the same scope as
theism, as they are all "worldviews". [7]
This is not
clear, since – for instance – Christianity makes a lot
more claims about the world than theism. Wouldn't Christianity have
greater scope?
Apparently,
Swinburne counts that as greater complexity, not scope. For instance,
he claims that adding fallen angels to the hypothesis that theism is
true, complicates it – not that it increases its scope. [8]
So,
let's grant that that would increase complexity.
Even
then, the
rivals of theism do not have to be physicalism, or naturalism, etc. -
even assuming that such hypotheses have the same scope as theism.
Someone
might posit the universe as a brute fact[9],
and
make no claims whatsoever about whether or not there are other
realms, maybe other universes causally disconnected from ours, and
also brute facts, etc.
On
the other hand, theism posits a creator of everything, making a claim
about an entity that controls the whole of reality.
So,
it seems to me that theism has greater scope than some alternatives.
1.2.
The concept of God
In
this section, I will take a look at some aspects of the concept of
God, as proposed by Swinburne – in particular, to moral
perfection.
According
to Swinburne, theism can be defined as follows: [10]
Theism: "There exists
necessarily a person without a body (i.e. a spirit) who necessarily
is eternal, perfectly free, omnipotent, omniscient, perfectly good,
and the creator of all things"
That
being is called "God", and the hypothesis that theism is
true is called "h" in Swinburne's case – I will call
it either "h", or simply "theism".
By,
"moral perfection", Swinburne means that "God always
does a morally best action (when there is one), and never a bad one.
[10]
That
definition does not state that God would
never
do a bad action, but that should be understood in that way, in
context.
As
for "perfectly free", that means that his choices aren't
causally influenced[11].
1.2.1)
Freedom
While
I don't agree with Swinburne's conception of freedom, I'm not going
to challenge it in the main part of the argument. I will make some
brief comments in the first
appendix,
but for now, I will just point out the following:
According
to Swinburne "an
inbuilt detailed specification of how to act is a much more complex
person than one whose actions are determined only by his uncaused
choice at the moment of choice"[12]
Yet,
Swinburne posits a person who is morally
perfect, which
amounts to a detailed specification of
how
to act, in infinitely many specific situations.
In
fact, if an inbuilt detailed specification as to how to act is
incompatible with "perfect freedom", then moral perfection
and "perfect freedom" are mutually incompatible, and thus
theism is logically
impossible.
Still,
let's leave that aside, and assess Swinburne's argument from
omniscience to perfect freedom to moral perfection.
1.2.2)
Moral perfection
In
order to argue for the claim that omniscience and "perfect
freedom" entail moral perfection, Swinburne makes the assumption
that moral judgments are propositions that are true or false.
I
will grant that metaethical assumption, but there are implicit
assumptions that require further consideration.
1.2.2.1.
Some implicit assumptions
Apart
from the explicit assumption I just granted, Swinburne makes other
assumptions, which aren't trivial but he does not state. In this
subsections, I will consider three of them.
a)
He assumes not only that non-cognitivism is not true, but also that
speaker relativism and culture-relativism aren't true. I will grant
these implicit assumptions as well.
b)
He implicitly rejects a moral error theory; for some reason, he
addresses and rejects the claim that moral judgments aren't true or
false, but he does not mention the claim that all moral judgments
like "X is immoral", "X is morally obligatory",
etc., express propositions but are false.
I
will grant this assumption too.
c)
He assumes that any omniscient, "perfectly free" agent
would be a moral agent or moral being – i.e, a being with moral
properties, and/or whose actions have moral properties.
Moreover,
given his
conception of moral goodness,
and in context, he seems to assume that any rational agent – or
at least any sufficiently intelligent rational agent – is a
moral agent.
Even
under the previous assumptions, this is not at all clear.
For
instance:
Let's
suppose an alien species evolved on another planet, and instead of
morality, they have other intuitions and standards – let's call
that "morality*.
They're
free in whatever sense in which humans are free, but they have no
sense of right and wrong, even though they have a sense of right* and
wrong* - which yields different results from a sense of right and
wrong, in some cases.
For
instance, a behavior might be morally bad, but morally* neutral, or
morally* bad, but morally neutral.
If
they decided to hunt and kill humans for sport, would they acting
immorally, assuming that they're not acting immorally*? Or would they
be not morally anything, like a lion or a tiger?
Or
let's consider an alternative:
On
yet another planet, some aliens design a smarter organism that ends
up overtaking their whole planet, and becomes nearly their entire
biosphere.
This
being – let's call it "Guk" - sends probes to Earth,
abducts humans and conducts all sort of experiments on them, some
horribly painful.
After
it's done experimenting on a human, it discards her – i.e., it
kills her.
Guk
has
a mental makeup that its very different from that of its makers, and
from that of humans.
In
particular, it has no concern whatsoever for the well being of other
beings, no sense of right or wrong – nor some analogue -, no
sense of guilt – nor some analogue -, etc.
However,
it is extremely intelligent; it can design and make computers,
starships, genetically engineered organisms – even some smarter
than humans -, and so on, and is free to act as it pleases – at
least, as free as humans, and as free as its makers.
Would
this being, Guk, be evil?
Or
would it be a non-moral entity, so that its actions aren't morally
anything (like those of a lion or a tiger)?
It's unclear to me.
1.2.2.2.
Swinburne's account of morality
Swinburne's
argument from omniscience and perfect freedom to moral perfection is
based on his conception of moral goodness and reasons, which the
following passages briefly explain:
Swinburne:
[13]
Having
a reason for an action consists in regarding some state of affairs as
a good thing,
and the doing of the action as a means to forwarding that state.
Swinburne:
[14]
I understand by an
action being morally good that it is overall better to do it than to
refrain, that there are overriding reasons for doing it; and by an
action being morally bad that it is overall better to refrain from
doing it than to do it, that there are overriding reasons for
refraining.
Swinburne:
[15]
But
the suggestion that someone might see refraining from A as
overall better than doing A, be subject to no non-rational influences
inclining him in the direction of doing A, and nevertheless do A is
incoherent. But
the suggestion that someone might see refraining from A as
overall better than doing A, be subject to no non-rational influences
inclining him in the direction of doing A, and nevertheless do A is
incoherent.
However,
whether a behavior is better than another one – even "overall
better" - depends on the standards of evaluation. The same goes
for refraining from a behavior.
For
instance, behavior A might be better than behavior B as a means to
achieve certain goal G.
Or
it might be that A is overall better than B, in the sense that there
are a number of goals G(1),..., G(n), and some order between the
goals – some standard S -, and according to standards S, A is
overall better than B.
Also,
it may be the case that the standards are implicit. In particular, it
may be that the implicit standard is morality, so in that context,
"overall better" would mean "morally better". And
it might even be that that's what humans usually
mean by that.
However,
there is no guarantee that other rational beings would care about
moral standards; for instance, they might care about some other
standards, say
moral*
standards.
There
would be no irrationality on
the part of said beings.
So, let's consider
some possibilities:
1.2.2.2.1.
An alien threat
For
instance, let's consider the
aliens
I introduced above, in some more detail.
Some aliens evolved
on a different planet.
Instead
of morality, as a result of their evolutionary process, they have
other intuitions and standards – let's call that "morality*".
They're
free in whatever sense in which humans are free, but they have no
sense of right and wrong, even though they have a sense of right* and
wrong* - which yields different results from a sense of right and
wrong, in some cases.
For
instance, a behavior might be morally bad, but morally* neutral, or
morally* bad, but morally neutral.
So,
they find Earth, and decide to hunt humans for sport. They don't see
anything immoral* about their behavior, and in fact, moral* standards
are such that it's not immoral* for those aliens to hunt humans for
sport.
Whether
it's immoral for them to do so is another matter, but the point is
that they do not care. They care about morality*, not about
the standards that humans care about; in particular, they do not care
about morality.
In
fact, even if the aliens, by means of studying their human prey, come
to the conclusion that there are some standards that humans care
about, and that according to said standards, their actions –
i.e., the aliens' actions – are very negatively regarded, the
aliens just won't care.
It
seems clear to me that there is nothing irrational
about
the actions of these aliens.
They
might be evil – or not; maybe they're not moral beings at all.
But
that's beside the point.
Moreover,
the conceivable aliens do not even have to be social beings:
1.2.2.2.2.
More aliens
Let's
consider the extraterrestrial
being that
I posited earlier, and make the story a little more detailed:
Guk sends
probes to other planetary systems, and studies a number of species,
including humans.
In
particular, by studying the humans his probes abduct, Guk
gains knowledge about the human species, including some specific
psychological traits, etc.
So,
Guk
takes
some humans to its planet for further study, and comes to know that
humans tend to describe it as "gray, blue and red".
It also concludes that those are
judgments based on the normal human visual system, and that, given
the light conditions, the judgments are very likely true, since the
system tends to be reliable.
So,
Guk
learns
he's (very probably) gray, blue, and red – though Guk
can't
experience the perceptual sensation of seeing those colors: its way
of perceiving the world around it is different and vastly more
complex than that of humans.
By
studying humans, and due to its knowledge of biology in general, Guk
also
reckons that basic human intuitions are usually reliable.
Also,
due to its studies of human psychology, Guk
finds
out that humans tend to make judgments and say behaviors are "morally
good", "morally evil", etc. - in different languages,
but in the same context -, and that those judgments appear to be
based on human basic intuitions.
Guk also
learns that all of the humans he studies - abducting them, sometimes
subjecting them to horrible experiments, in the end always killing
them - consistently state that it - i.e., Guk - is morally evil
and/or that its actions are evil, etc.
So,
Guk
comes
to the conclusion that the judgments are probably true, so that it
probably is evil.
However, that does not motivate
it to stop any more than learning it's gray, blue and red does –
i.e., that does not motivate Guk at all.
It also comes to realize that
some the actions it performs on the humans motivate the judgment.
Guk continues
studying them and it then tests kind of torments, and also rewards,
etc., puts humans into a variety of complex situations, etc., in
order to study what moral judgments its human guinea pigs will make -
and it continues studying them.
Eventually, Guk has satisfied its
curiosity with regard to humans, and it kills the subjects of its
study that are still alive.
Also, it reckons that, if humans
are left alone, in millions of years they or their successors might
advance enough to be a serious nuisance.
So, it designs some biological
weapons, and sends robots to attack the Earth.
The biological weapons kill
nearly all of the human population, and then the robots launch a more
traditional assault against the survivors to finish the extermination
– and succeed, wiping out the human species.
The
fact that Guk
has
learned that its actions are probably evil, however, fails to
motivate it in the least to stop them. It does not care about moral
evil, or good.
Clearly,
Guk
does
have reasons for acting – studying those other species
entertains it.
Moreover,
Guk
does
not have any reasons whatsoever
for
refraining from acting as
it does.
Someone might object that Guk
can't really grasp the meaning of moral terms.
But the point is that Guk does
not care about the standards humans care about – whatever those
are -, except to the extent that it's curious about humans and want
to know more about them.
Guk
cares
about its own entertainment, not about the well-being of anyone else.
1.2.2.2.3.
Omniscience and perfect freedom without moral perfection
Earlier
in this section, I posited a conceivable entity, Guk,
who does
not care at all about
the suffering of humans or other beings, and subjects some humans and
other beings to a lot of suffering in order to learn about them,
because that entertains it.
Also,
I posited some other
aliens who
have a different set of standards, other than morality –
namely, morality*.
Now,
clearly Guk
is
not omniscient or perfectly free, and neither are the other
aliens.
However,
that would not change the relevant point: omniscience and perfect
freedom would guarantee that a being would know what the best course
of action to achieve its goals is, but it does not guarantee that it
will have a particular set of goals; in particular, it does not
guarantee which standards it cares about.
If
the being is also omnipotent, that will guarantee that its goals –
if compatible with each other – will be accomplished, but that
too does not tell us what those goals are.
For
instance, conceivably, an omnipotent, omniscient, perfectly free
being could create humans and other intelligent physical beings
because it enjoys watching them struggle. In other words, it wants to
have enjoyment – that's its goal -, and it enjoys the show.
So,
the creator would create a universe with painful parasites, deadly
earthquakes, etc, because it enjoys watching lions, dinosaurs,
humans, etc., fight, suffer, succeed or fail, etc.
There
is nothing contradictory
about
such a being.
Such
being would surely know what's immoral – and what's immoral*
as
well, if those aliens
were
also created.
But
– once again – it
just wouldn't care – not
beyond the entertainment value of watching those beings in such a
situation.
Whether
such a being would be a moral
agent at
all is debatable, but one way or another, what's apparent is that it
wouldn't be morally perfect.
1.2.2.2.4.
Morality, objective values, and reasons
Swinburne
claims that there are objective values [14],
so perhaps someone might raise the objection that Guk
and
the other alien beings
in
question would fail to see the true, objective value, which is
given
by moral standards – not by moral*
standards, or by whatever standards Guk
may
have.
While
I've granted a number of Swinburne's metaethical
assumptions, they
do not entail anything about what agents value,
and what an agent values depends on its particular mental makeup.
Moreover, it's not
entirely clear at all what it would mean for "values" to be
objective in that sense.
If by "objective
values" Swinburne means moral judgments that are true or false
independently of who makes them, etc., I've granted that already, but
it has no bearing on my points.
If
Swinburne means that there are things (actions, agents, properties,
etc.) such that it would be immoral for
any sufficiently intelligent, knowledgeable and free agent not to
value them, whether or not that's the case has no bearing on my
points.
On
other other hand, if Swinburne is saying that there are things
(actions, agents, properties, etc.) such that it would be irrational
for
any sufficiently intelligent, knowledgeable and free agent not to
value them, and that moral goodness is among them, then I don't see
any reason to grant that assumption.
Moreover,
the entities I posited – say, those behaving according to
moral*
standards -, would not appear to be irrational in any way, or
making any mistakes about the properties of anyone or of any
behavior.
But
moreover, even if it turned out that the word "rational"
actually picked some specific final or terminal values – i.e.,
the things, behaviors, etc., that an entity values for their own sake
-, that would only mean that "rational" also increases the
complexity of the claim that theism is true, as it is giving a
description of the terminal values of the entity that is not
contained
in the description "omniscient" or "perfectly free".
In
other words, in any case, from "omniscient and perfectly free",
nothing
follows
about what the terminal values of the entity in question are.
For that matter, an
entity might value its own fun for its own sake, might find the
torture of humans fun, and might not have any stronger terminal
value. So, that entity might torture humans for fun.
There is nothing in
"omniscient" or "perfectly free" that would imply
or even suggest otherwise.
1.3.
The intrinsic improbability of theism
In
this section, I will grant for the sake of the argument that there is
such thing as an objective intrinsic probability of theism –
which may or may not have a precise numerical value – and argue
that it's very improbable.
1.3.1.
Simplicity
As
I explained earlier,
Swinburne's criterion for simplicity in the case of personal
explanations is based on the following conditions: [6]
Few
entities, few types of entities, few properties, few constant
intentions, and "continuing basic powers", and
"simple
laws – constant predictable ways
in which persons acquire beliefs from their surroundings."
[6]
My
hypotheses will only posit one type of entity and one entity each, no
more properties than the theistic hypothesis, no ways of acquiring
beliefs other than theism (I will only stipulate omniscience), and
the same basic powers than God has – i.e., omnipotence.
Thus,
given Swinburne's criterion for simplicity, my hypotheses will be no
more complex than theism.
Granted,
someone might posit an alternative criterion for simplicity, or for
when to unpack, etc., but picking an arbitrary criterion won't do. It
would be up to the claimant to justify the choice of the criterion.
1.3.2.
Scope
Another
criterion that Swinburne's consider in order to assess the intrinsic
probability of a hypotheses, is scope – but Swinburne claims
it's far less important than simplicity. [7]
However,
Swinburne says the scope is always the same for all "world
views", so we can just ignore it when comparing the intrinsic
probability of the hypotheses in question.
1.3.3.
Consequentialist deities[16]
In
this section, I will propose some alternatives to God.
For
any entity EN that I propose, by "AH(EN)" I mean the
hypothesis that AH(EN) exists.
So,
let's propose first a type of consequentialist deity that I will call
"Con(1)".
Con(1)'s
definition is like God's, minus moral perfection, plus the property
that it is consequentialistically perfect, in the sense that it
always does a the consequentialistically best action when there is
one, and never a consequentialistically bad one – though it
might do a morally bad one -, according to some system that rates
actions depending on some of its consequences.
More
precisely, on this consequentialist system, an action X is
consequentialistically better than an action Y if it satisfies more
desires.
If
the number of desires they satisfy is the same, then X is
consequentialistically better if X satisfies desires that are more
intense, on average.
If
they satisfy the same number of desires, and neither of them
satisfies desires that are overall more intense, then neither action
is consequentialistically better than the other – they're
consequentialistically equivalent.
So,
the hypothesis AH(Con(1)) can be defined as follows:
AH(Con(1)):
There
exists necessarily a being who necessarily is eternal, perfectly
free, omnipotent, omniscient, consequentialistically perfect, and the
creator of all other things.
Given
Swinburne's account of simplicity,
we can tell that that AH(Con(1)) is no less simple than theism. Since
the scope is the same, the prior probability of AH(Con(1)) is no less
than that of theism.
Next, let's introduce also
consequentialist deities that focus on some type of beings.
Let
T(n)
be
types of possible beings who can have desires and such that T(n) ≠
T(m)
if n ≠ m.
A
being is n-consequentialist if it will always maximize the
satisfaction of desires of beings of type T(n), in the same way
explained in the previous case – in other words, equal numbers
of desires and/or intensity are handled as
before.
AH(Con(1,n)):
There
exists necessarily a being who necessarily is eternal, perfectly
free, omnipotent, omniscient, n-consequentialistically perfect, and
the creator of all other things.
Given
that there cannot exist more than one creator of all other things,
clearly the hypotheses AH(Con(1,n)) are pairwise disjoint.
Also,
given
Swinburne's account of simplicity,
we can tell that AH(Con(1,n)) is no more
complex than theism, and if k is only tautological evidence, we
can tell:
P(AN(Con(1,n))│k)
≥ P(Theism│k)
[17]
Now,
considering that the hypotheses are pairwise disjoint, if
there is some integer r such
that P(h│k)
≥ (1/r), then we obtain:
P((AN(Con(1,1))
or AN(Con(1,2)) or... AN(Con(1,n)))│k) = P(AN(Con(1,1))│k)
+ P(AN(Con(1,2))│k) +... + P(AN(Con(1,n))│k) = ((r+1)/r)
> 1.
Since
that is impossible, it follows that there is no r such that
P(h│k) ≥ (1/r).
Moreover,
we can further simplify the hypotheses, by introducing fewer claims
about the properties of the entities:
AH(Con(2,n)):
There
exists a being who is the creator of all other things, and who is
n-consequentialistically perfect.
Since
AH(Con(2,n)) is entailed by AH(Con(1,n)), it's no less
probable, and given that we're leaving aside the conditions of
omnipotence, omniscience, etc., that makes the alternative simpler.
At
this point, someone might suggest that n-consequentialistialistic
perfection is a condition on the persons' inclinations, and thus
precludes perfect freedom.
However,
with that criterion, the same would apply to moral perfection, making
moral perfection and perfect freedom incompatible, and thus making
God logically impossible.
Also,
someone might say that behaving in that way is an imperfection, not a
perfection.
However,
that objection would miss the point: I'm simply positing beings who
always behaves in the way I described above, and who always would
behave that way.
If
there is some conception of perfection that those beings would not
match, that's beside the point.
If
someone prefers not to call it "n-consequentialistically
perfect" but something else, that's a matter of taste, but it
does not affect the point I'm making.
1.3.4.
Estimating the intrinsic improbability of God
Given
the results of the previous subsections, we've concluded that
P(Theism│k)
is
no greater than 1/r,
for any natural number r.
That
leaves us with the following options:
a)
P(Theism│k)
= 0
b) P(Theism│k)
= i
Where
i is a nonzero infinitesimal hyperreal, assuming probabilities can
have hyperreal values.
c)
P(Theism│k)
> 0,
somehow mysteriously does not have a numeric value, even though it's
no greater than (1/r),
for all natural numbers r.
Even
then, n*
P(Theism│k)
is
also no greater than 1/r,
for any natural numbers r
and
n,
so P(Theism│k)
still
is some kind of infinitesimal – even if a mysterious one.
d)
P(Theism│k)
= i
Where
i
is
some sort of nonzero infinitesimal – whatever that is -, but
the point is that n*
P(Theism│k)
is
also no greater than 1/r,
for any natural numbers r
and
n.
Moreover,
the alternative
hypotheses I
introduced are just a few – well, infinitely many few, but the
point is that there are many more.
So,
the above shows that even if one assumes that for some reason, theism
is somehow a simple hypothesis, the math still shows that theism is
intrinsically really
improbable.
1.4.
The final improbability of theism
In
the previous
section,
I concluded that the prior probability of theism is no greater than
1/r,
for any natural number r.
In
this section, I will show that the final probability of theism
is also no greater than that, under an assumption that Swinburne
requires to make his case.
So,
if that assumption fails, Swinburne's case is blocked. And if that
assumption is true, then the probability of non-theism, given
our evidence, is greater than x, for any real x < 1,
and so we can conclude beyond a reasonable doubt that non-theism is
true.
The
assumption is question is P(e&k)
>
0,
where e
is
– following Swinburne [18]-
the conjunction of e1,
e2,...,
en,
which are different propositions that people bring forward in
arguments for or against theism, and k
is
mere tautological evidence.
Without
P(e&k) > 0, it would not be possible to apply Bayes'
theorem, since it's impossible to divide by zero, and while Swinburne
divides his argument in several steps, in the end he has to estimate
P(theism│e&k) using Bayes' theorem, and that
requires dividing by P(e&k).
So,
let's assume P(e&k) > 0, and introduce some alternative
hypotheses:
1.4.1.
More alternative deities
Let's
introduce a class of universes:
U(1):
A universe where e obtains.
In
other words, U(1) is a universe where all the propositions in
question are true (i.e., the universe is ordered, there are physical
beings who are moral
agents,
etc.)
Then,
let's introduce U(n), for all natural n > 1, which
are possible alternate physical universes, which behave in accordance
to rules described by some mathematical model M(n), such that
the rules are different for any two numbers n and m.
No
rational being lives in any of those alternate universes.
Now,
let's consider a class of standards:
V(n):
a)
Creating at least one universe of each of the types U(1), U(2),...,
U(n) is V(n)-good.
b)
Creating a universe of type U(n+m) is V(n)-bad, for any natural m.
c)
All other actions are V(n)-neutral.
d)
A being is V(n)-good if and only if it does at least one V(n)-good
action, and never does any V(n) bad action.
Finally,
let's consider hypotheses asserting the existence of certain deities,
called "Alternative Deity(n)":
AH(Alternative
Deity(n)): There exists a being who is omnipotent, omniscient,
V(n)-good, and the creator of all other beings.
Let
"AH(AD(n))" stand for the hypothesis that Alternative
Deity(n) exists; as usual, "h" stands for theism, so I use
"h" and "theism" interchangeably; "k"
is tautological evidence.
1.4.1.1.
Simplicity
Given
Swinburne's account of simplicity,
it seems that AH(AD(n)) is no less simple than theism:
It
does not posit any more entities, kinds of entities, properties, or
constant intentions.
As
for the powers, they're just as "simple" as in the case of
theism – i.e., omnipotence -, and they do not have liabilities
– since they are omnipotent.
Also,
while Swinburne claims that a detailed specification of how to act
makes a person more complex [12]
than
one that is perfectly free, in this case I've stipulated that the
entities in question are perfectly free.
The
fact that there is a specification given by the fact that Alternative
Deity(n) is V(n)-good would not seem to preclude perfect freedom,
just as the specification that God is morally perfect does not
preclude perfect freedom, either – and if someone claimed
otherwise, the burden would be upon them: why would morality be the
specification that just does not preclude perfect freedom? [19]
That
aside, someone might object that AH(AD(n)) are more complex than
theism since they entail, with probability 1, that there will be a
physical universe of a certain kind.
However,
if that counted as complexity, then if Swinburne were correct that
God would create other beings with probability 1, that would make
theism less simple than positing just one being.
Also, Swinburne
claims that theism entails a probability of a physical universe
greater than 1/2, so if that counted as reducing simplicity, it
should count in Swinburne's argument too.
Still,
I will present an alternative
in the
following subsection.
1.4.1.2.
Probabilistic assessment
For
the reasons given above,
theism is no simpler than AH(AD(n)), for any natural n.
[20]
Also,
given that they are "world views", it seems that AH(AD(n)),
for any natural n, has the same scope as theism.
Hence:
a)
P(AH(AD(n)))│k) ≥ P(Theism │k)
b)
P(e│k&AH(AD(n))) = 1.
c)
P(AH(AD(n))│e&k) = P(e│k&AH(AD(n)))*P(AH(AD(n))│k)
/ P(e&k) ≥ P(e│k&Theism) * P(Theism │k) /
P(e&k) = P(Theism │e&k)
Given
that AH(AD(n)) and AH(AD(m)) are disjoint if n≠m,
if there is some natural r
such
that P(Theism
│e&k)
≥ 1/r,
a contradiction follows as
before.
So,
the result is established.
1.4.2.
Choice
Someone
might object that I haven't defined the U(n), explicitly, and maybe
infinite choices is then somehow a problem.
I
don't see how that would be a problem, but just in case, one may as
well proceed as follows: first, let's assume that there is some
natural r
such
that P(Theism
│e&k)
≥ 1/r,
and let's consider U(1), U(2),...,U(r+1), and the corresponding
hypotheses, and a contradiction follows as above.
1.4.3.
Gambling deities
First,
let's define a class of actions.
A(n):
A coin-toss like action that might create a universe of
type U(k),
for some k≤n, with probability 0.5/n, or no universe at all.
For
instance, if an omnipotent being in question were to carry out a
A(n)-type action, he would choose to leave it at random which
universe to create with his action, if any – there would be a
0.5 chance that no universe is created by that action.
Now,
let's define a class of standards W(n):
a)
An A(n)-type action is W(n)-good.
b)
An A(m)-type action is W(n)-bad, if n≠m
c)
All other actions are W(n)-neutral.
d)
A being is W(n)-good if and only if it does at least one W(n)-good
action, and never does any W(n)-bad action.
Finally,
let's consider hypotheses asserting the existence of certain deities,
called "Creator(n)":
AH(Gambling
God(n)): There exists a being who is omnipotent, omniscient and
W(n)-good, and the creator of all other beings.
The
new conditions do not entail that, with probability 1, there will be
a physical universe.
They
do guarantee, however, that P(e│k&AH(Gambling
God(n))) ≥ 0.5/n.
Given
Swinburne's account of simplicity,
we can tell that that AH(Gambling God(n)) is no less simple than
theism, for all n.
The details are the same as
in
the previous case.
[21]
Also,
given that they are all "world views", it seems they all
have the same scope as theism.
Let
"AH(GG(n))" stand for AH(Gambling God(n)) - as usual, "h"
stands for theism, so I use "theism" and "h"
interchangeably.
Hence:
a)
P(AH(GG(n)))│k) ≥ P(h │k)
b)
P(e│k&AH(GG(n))) ≥ 0.5/n.
c)
P(AH(GG(n))│e&k) = P(e│k&AH(GG(n)))*P(AH(GG(n))│k)
/ P(e&k) ≥ 0.5*P(AH(GG(n))│k) / (n*P(e&k)) ≥
0.5*P(e│k&h) * P(h│k) / (n*P(e&k)) =
0.5*P(h│e&k)/n
Given
that AH(GG(n)) and AH(GG(m)) are disjoint if n≠m, if
there is some natural r such that P(theism│e&k) ≥
1/r, a contradiction follows because the harmonic series is
divergent.
An
objection regarding infinite choices is handled essentially as
before,
with the only difference that assuming P(theism│e&k)
≥ 1/r for
some fixed r,
we
derive a contradiction by considering U(1), U(2),..., U(m) - and the
corresponding hypotheses -, taking – for example – m
= 3^((2*r)+1).
1.4.4.
Simplicity, revisited
Someone
might raise the following objection to the previous results: they
might claim that even though specifications as to how to act do not
preclude perfect freedom, it still remains the case that including
specifications as to how to act in specific situations does make a
hypothesis more complex, and the more detailed the specifications,
the more complex the hypotheses becomes.
Moreover,
they might insist that, while morality is somewhat complex, it's less
complex than some of the different
alternatives
I
posited.
But
why would that be?
It's
not because of how long it would be to, say, write the number n
if
unpacked, or how long the descriptions of the mathematical models in
my alternatives would be: as we saw before,
Swinburne's criterion for simplicity does not require unpacking
numbers or mathematical models when they're in the minds of the
entities posited as personal explanations – and rejecting that
criterion would require unpacking omniscience, making
theism an enormously complex hypothesis.
Why
would then the specifications of behavior in the different
alternatives
I
posited be more complex than the specification of behavior in the
theistic hypothesis?
Someone
might say that unpacking is sometimes needed, and that that affects
my alternatives, but not theism.
It
would be up to the claimant to defend that claim, though.
In
addition to that, by admitting that specifications of behavior
increase the complexity of a hypothesis, they would have to concede
that that applies to the theistic hypothesis.
Perhaps,
someone will come up with a system of assessing simplicity that just
happens to make only theism very simple, but not any of the
alternatives I presented. But as usual, they would have to defend
their choice of criterion.
So,
at this point, I can rest my case.
1.5.
Conclusion of Part 1
Given
the previous considerations, the conclusion is that Swinburne's
arguments in "The Existence of God" do not provide any
support for the hypotheses that theism is true, even without
considering a number of other claims – e.g., about freedom and
causality, or about personal vs. scientific explanations – that
are controversial to say the least.
A
theist might modify Swinburne's case in order to block some of the
objections I raised, or defend Swinburne's
account of morality,
but in any case, the burden would be on them.
2.
Power, moral assessments and probability
Leaving
aside the previous objections for the sake of the argument,
Swinburne's case rests, to a considerable extent, on his moral
assessments of certain actions, plus some reasons to back them up.
My
moral assessments of the behaviors in question, however, are in most
cases radically different from Swinburne's; in what follows, I will
explain my take on some of those behaviors, using the same method he
uses (i.e., I will use my sense of right and wrong, give reasons,
etc., as usual in moral discussions), and so presenting a different
perspective to readers – who can of course make their own
assessments on these matter.
In
addition, I will challenge much of Swinburne's reasoning in support
of some probabilistic assessments, and also challenge Swinburne's
claims about God's power. The issue of power,
however, is not required for the rest of the arguments in this
section to succeed.
2.1
What can God do?
Swinburne
defines God as omnipotent, yet he claims that
God
cannot
do
evil.
Also,
according to Swinburne, God does not have the freedom to choose
between good and evil. [22]
However,
Swinburne's definition of omnipotence is that "he is able to do
whatever it is logically possible (i.e. coherent to suppose) that he
can do". [10]
Now,
it's perfectly coherent to suppose that he can,
for instance, torture everyone else for eternity, and that he is able
to
do it.
What
is not coherent is to suppose that he would
do
it.
If, however, we
accepted Swinburne's conception of freedom and power for the sake of
the argument, then it seems that God indeed would not have the power
to bring about evil, because that would somehow entail that it's
logically possible that an entity E is God and does evil, and that is
a contradiction.
But that would be
very odd, and it would have other odd consequences, such as the
following one:
Let's consider the
following hypothesis:
AH(Gin): There
exists a being who is omnipotent and who has the property that he
will never communicate with other intelligent beings.
Let "Gin"
be the being whose existence is asserted by AH(Gin).
Then,
by Swinburne's rationale to conclude that God cannot do evil. it
follows that Gin is omnipotent, but it cannot
communicate
with other intelligent beings. How
would that be omnipotence, in any interesting sense of "omnipotence"?
Note
that the definition is not contradictory if that of God isn't, since
I merely stated some propensity that Gin will always follow –
i.e., not to communicate with other intelligent beings – just
as there is a propensity – i.e., doing always what's morally
good – built-in the definition of "God".
2.2.
What would God likely create, or refrain from creating?
I
will now assess Swinburne's moral assessments, both by challenging
his arguments to back up some of those assessments, and by presenting
my moral intuitions in opposition to his moral intuitions.
Let's
consider some of Swinburne's claims, then:
a)
There is no world W such that no other world W' is better. [23]
Since we're making moral
assessments, that would be a claim that no world is morally better.
Normally, we assess moral goodness of people and/or their behaviors,
so this is rather odd.
Perhaps, one could judge the
moral goodness of a world counting how much moral goodness there is –
or how much evilness.
But Swinburne argues that the
moral goodness of a world consists in having a finite or infinite
number of conscious beings who will enjoy it, so the more, the
better, and so there is no world W such that no other world is
better. But why would that be so?
I see no good reason to believe
so.
Regardless
of more or less obscure metaphysical issues, the only information
about God's motivations provided in the definition of 'God' is that
he's perfectly [morally] good[10], which
means he'll never do a morally bad action, and he'll do a morally
best action when there is one.
That's Swinburne's only
definition.
If we go by that, whenever an
action is neither morally best nor morally bad, it appears that we
simply have no means whatsoever of assessing whether God would do it,
or assigning any probability to it.
On the other hand, if we extend
moral perfection to a propensity to do morally good actions even if
not the morally best when there is no morally best action, it might
be argued that that would give us, in some cases, a means of
rationally assigning probability that God would carry out some
morally good action.
But the question is: how?
How do we make such assessment,
where there are infinitely many possibilities?
Moreover, even granting for the
sake of the argument that, sometimes, we can make rational
probabilistic assessment about God's potential actions even if such
actions are only morally good and not morally best, we still would
have no means of making any assessment when it comes to morally
neutral actions, since we do not know anything else about the
character of God that would allow us to make such an assessment.
Indeed, the only mental trait of God that tell us anything about his
motivations in perfect moral goodness, so only moral motivations can
be used (at best) for probabilistic assessments.
That
aside, this particular claim, combined with other claims Swinburne
made, allow us to raise another objection, which I will raise in the
next part. But for now, let's get back to
some of Swinburne's moral assessments.
b)
God does not have the right to permit or impose unlimited suffering
to anyone against that agent's choice. [24]
I agree.
Moreover, it seems to me that
unless it's a sacrifice to prevent a greater evil that might come to
others, it would clearly be irrational for an agent to choose to
suffer for all eternity when she can choose otherwise.
Also, it
is obvious – I hope – that making an irrational decision
to suffer forever does not make a person deserving of infinite
punishment, and that it would be morally bad for a moral agent to
impose such infinite suffering on someone who makes such an absurd
choice.
c)
It would be better to create lions and tigers than just lions, and
generally more the more species, the better. [24]
I strongly disagree. Frankly,
that's just puzzling.
Swinburne gives no argument other
than calling it "plausible". I would rather call it
"extremely implausible".
Intuitively, there is no moral
advantage in doing so as far as I can tell.
In fact, I'd say that creating
any such predators create unnecessary suffering, and so God would not
do so. But leaving aside lions and tigers and considering other
beings if required, if it's not immoral for an omniscient, omnipotent
moral agent to create such beings, then making more or fewer kinds
would appear to be morally equivalent, all other things (e.g.,
suffering) equal.
In
fact, under the assumption that in some cases it wouldn't be immoral
for an omnipotent, omniscient moral agent to bring about predators
that cause horrible suffering on other beings (an implausible
assumption), then it would seem to me that that's not a moral choice,
but more like choosing between eating an apple or a banana. The
choice of creating in any such circumstances is neither morally good
nor morally bad. It's morally neutral – of course, assuming
that there aren't any other
factors
involved that have moral relevance. [25]
However,
if it's not morally
good
or bad but neutral, we cannot rationally make any assessments about
the prior probability that God would do it, since moral perfection is
the only information about the motivational aspect of God's mind that
the definition of 'God' provides, and so only moral reasons ought to
be used to assess how he would behave.
In other words, if there is a
category C of potential actions by God such that any action in C
would be morally neutral, then it seems that we do not have the means
to assess the probability that God would carry out an action in that
category.
d)
Plausibly, it's better for God to create than not to.
[24]
Swinburne
claims that a perfectly morally good being would inevitably try to
make other good things. However, intuitively, I don't see how not
being inclined to make anything would mean a being is not morally
good, or any less good. There appears to be no immorality in not
creating other beings, as far as I can tell.
Perhaps,
it could be argued that making other morally good beings is morally
good, even if not morally obligatory. That is not at all clear if
there already is an omnipotent, omniscient morally perfect being.
Still, even granting for the sake of the argument that, in
some cases, an
omnipotent, omniscient moral agent would be doing something morally
good (even though not obligatory) by creating other moral agents,
clearly in other cases it would be immoral (e.g., it would be immoral
to create a moral agent who would suffer horribly forever no matter
what she chooses), so this at best would lead us to the question of
whether it's morally
good for
an omnipotent, omniscient moral agent to create the
kind of moral agents that actually exist.
I
will later argue against
that
idea.
e)
Swinburne: [26]
"Animate substances are
substances of a better type than inanimate ones".
That is very obscure. What
does Swinburne even mean?
What is a substance of a 'better
type'?
Swinburne oddly classifies
substances in types, according to which ones are allegedly better.
But that seems obscure at best, and in any case not related to moral
intuitions and assessments.
In any case, what is relevant to
the matters at hand is what would be morally good or even morally
acceptable for an omnipotent being to do, not some obscure statements
about 'better type' of stuff, even if we assume them to be coherent.
Still,
if Swinburne means that it's always morally
better to create animate substances that inanimate ones, that's
surely false. For instance, it's morally better for an omnipotent
being to create a rock than an intelligent being (even a moral agent)
that always suffers horrible pain she cannot stop, no matter what she
does.
But
if he does not mean that, then what does Swinburne mean?
A
lion is not of morally better than a rock. There is no moral
dimension to compare them, as far
as I can tell. Nor is it generally morally better to create
lions than rocks, as far as I can
tell.
f)
Swinburne: [26]
"Humanly
free agents are substances of a better type than animals".
Again, that's very obscure.
If Swinburne is saying that
somehow it's always morally better to create – essentially
that it's morally better to create entities with limited moral
awareness than with no such awareness at all -, I can see no
particular reason to believe so.
In fact, under certain
circumstances – which I maintain apply to the case of an
omnipotent, omniscient moral agent -, it would be immoral to create
humans or similar beings, but creating beings with no moral awareness
is not always immoral.
So, it's not always morally
better to create beings with moral awareness than without it.
But if he means something else,
what is it?
In any case, and as before, what
is relevant to the matters at hand is what would be morally good or
even morally acceptable for an omnipotent being to do, not some
obscure statements about 'better type' of stuff. But let's consider
the following category of actions:
C1: Moral agent A brings moral
agent B into existence.
Some actions in such category are
immoral. For instance, it would be immoral for an omnipotent,
omniscient moral agent to carry out any action in the subcategory
'moral agent A brings into existence a moral agent B who will suffer
horrible pain no matter what she does'.
On the other hand, it's usually
not immoral for humans to have children, so some actions in some
category C1 aren't immoral.
Plausibly, also, some actions in
C1 are morally good.
Now, let's consider the
subcategories:
C2: An omnipotent, omniscient
moral agent A who exists in a world devoid of any other moral agents,
brings about moral agent B, who is a morally flawed agent who will
behave immorally sometimes (a behavior that A predicts).
C3: An omnipotent, omniscient
moral agent A who exists in a world devoid of any other moral agents,
brings about moral agent B, who is a moral agent who will endure
suffering she does not deserve.
Using
the same method used by Swinburne – who makes his claims by
means of his own moral intuitions-, I would say that any actions in
subcategories C2
and
C3
appear
intuitively morally
bad,
and so God would not carry them out. I will address this matter again
later.
Note that claiming that there
might be a morally sufficient reason is not enough to block an
assessment that all actions in categories C2 and C3 are
morally bad. For that matter, the following category contains only
morally bad actions:
C4: An omnipotent, omniscient
moral agent A who exists in a world devoid of any other moral agents,
brings about moral agent B, who is a moral agent who will endure
immense suffering for all eternity, regardless of how B behaves or
tries to behave.
Surely,
any action in category C4 would
be immoral. And saying that God might
have morally sufficient
reasons to bring about an action in C4 would
not provide any reasons to undermine our intuitive moral assessment,
for the simple reason that there is no good reason whatsoever to even
suspect that God might have morally sufficient reasons to bring about
an action in C4.
But
then again, there appears to be no good reason to suspect our
intuitive assessment of categories C2
or
C3:
while
the degree of immorality of many actions in either of those
categories is much less than the degree of immorality of any action
in category C4,
they still appear clearly immoral, going by my sense of right and
wrong.
Swinburne assesses otherwise
(though he appears also confused by introducing the matter of the
'better type' of agents agents); I suggest that readers make their
own assessments of these matters.
g)
A world with God alone would be a bad state of affairs.
[22]
That is puzzling. What would it
mean for it to be a 'bad state of affairs'?
A state of affairs where there is
a lot of immoral behavior, perhaps?
But in this scenario (i.e, God
alone), there surely is no morally wrong action involved, and
the state would be one in which the only substance in existence is
morally perfect.
If he means something else, what
is it?
h)
God needs to share, love, interact, etc.
[22]
Actually,
that is not entailed by
the concept "God", which
gives no details about the inclinations of the agent other than moral
perfection.
Instead,
one would need to include more
claims about the psychological makeup of God to
reach that conclusion, making the theistic hypothesis even more
complex than it already is.
As
it stands, we have no information on whether God needs
to
share, interact, etc., or even love.
Perhaps,
someone might claim that a morally good agent has a need to love,
share and interact. However, after a brief analysis. we can tell that
that is not plausible. To see this, let's consider the following two
scenarios:
Scenario
SC1:
During
an expedition to colonize another planetary system, a human being,
Bob, who has always been a good person, ends up alone, stranded on a
spaceship at drift.
Bob
has no way and will never have a way of contacting anyone else. But
he has some frozen GM dog eggs and sperm, and a machine that can use
that to make GM dogs using artificial wombs, etc. (GM dogs are just
like regular dogs, but sterile; they can only reproduce by artificial
means).
The
ship's reactor and recycling system are still operational, and can
sustain Bob for the rest of his life; in fact, they were designed to
sustain a much larger community for thousands of years, so that won't
be a problem.
The
ship AI can also sustain a good number of dogs, if Bob is dead and
can't do it himself.
Knowing
that, Bob makes a dozen dogs to keep him company, so that he can
share, interact, etc., at least to some extent. Dogs, of course,
aren't the most suitable animal for Bob; humans would do much better.
But Bob does what he can.
Scenario
SC2:
Similar
to scenario SC1,
but instead of Bob, the person who gets stuck on the ship is Alice,
who also has always been a morally good person. However, Alice is not
human. She's a Vulcan-like alien who does not have the same
propensities as Bob. Alice might make the dogs as Bob did (they were
originally both humans and aliens of Alice's species on board), but
she does not find that option appealing, and chooses to spend the
rest of her life meditating instead. So, she does exactly that.
The
point here is that, going by scenarios sc1
and
sc2, we
should not conclude that Bob was a morally better person than Alice,
or that his choice of action was a morally better one – nor the
other way around. Nor
are we entitled to conclude that Bob was probably
a
morally better person, or probably
took
a morally better course of action.
The
point here is that there is nothing in the concept of moral goodness
that suggest that a morally good agent would need to share, interact,
love, etc.; moreover, even if it's true that a morally good agent
would love at least some other agents that already exist,
there is no way of telling, from the meaning of the concepts alone,
that a morally good agent would have an inclination to create other
beings to share, etc.
Still,
perhaps someone might say that a morally good agent would have an
inclination to create other morally good agents to share,
interact, etc., and dogs do not count since they're not moral agents.
Even
if that were true, that would only lead us to the issue of creating
other moral agents, and not non-moral agents, or things that
aren't agents.
So,
much of Swinburne's reasoning would be blocked anyway, since
he's claims that it would be morally good for God to make all sorts
of things other than moral agents.
Moreover,
it's not at all clear to me that we can tell a morally good agent
would have an inclination to create other morally good agents,
regardless of any other features of the morally good agent's
psychological makeup.
In
fact, that looks counterintuitive to me.
To
see why, we can modify the previous
scenarios,
as follows:
Scenario
SC3:
During
an expedition to colonize another planetary system, a human being,
Bob, who has always been a good person, ends up alone on a spaceship,
as a result of a serious accident.
Many
ships already reached their destination, and more are on their way.
Bob
can use the tech on board the ship to clone some of the dead humans
and make more humans, whom he would raise.
Now,
regardless of whether or not Bob makes other humans, the ship will
reach its destination, with its cargo. Moreover, colonization will go
on one way or another. So, Bob decides to make more humans, raises
them, and some of his descendants reach the target planetary system.
Scenario
SC4:
Similar
to scenario sc3,
but instead of Bob, the person who gets stuck on the ship is Alice,
who also has always been a morally good person.
Alice
is not human but an alien (though we might as well pick a human
here), who decides not to clone anyone, since she does not feel like
reproducing and raising children. The ship gets to its destination
regardless, and colonization goes on unhampered.
As
in the case of scenarios 1
and
2, it seems to me that scenarios 3
and
4
do
not entitle us to conclude that Bob was or probably was a better
person than Alice, or that his course of action was morally better.
If
so, then there is nothing in the concept of 'morally good' that
suggests that a person would need to share, interact, or love.
Sometimes, a moral agent sometimes has a moral obligation to interact
with other moral existing agents; maybe, sometimes, she would have a
moral obligation to love them.
However,
the question in the context of this particular
part of Swinburne's argument
is
whether we can tell, from the assumption that an agent is morally
good, that he would need or at least feel inclined to create other
morally good agents in
order to interact with them, share things with them, love them, etc.,
and the answer appears to me to be negative.
Still,
even if we were entitled to infer such propensities, the question
would still be what kind of moral agents, and my assessment would
still be that an omnipotent, omniscient, perfectly morally good agent
would not
create
the kind of moral agents that we observe.
i)
Significant
free choice is evidently good. [22]
Swinburne
claims that what he calls 'significant free choice' is evidently
good, and from that he goes on to say that it's something God would
want to do.
The
issues at heart here are what Swinburne means by "significant
free choice", whether it would be morally good for an
omnipotent, omniscient creator to bring about such beings in some
cases.
So,
a category to consider here is:
C5: A moral agent brings about
the existence of one or more entities with what Swinburne calls
'significant free choice' into a world previously devoid of such
agents.
A subcategory is:
C6: An omniscient, omnipotent
moral agent brings about the existence of one or more entities with
what Swinburne calls 'significant free choice' into a world
previously devoid of such agents.
The questions are whether at
least some actions in the subcategory would be morally good, and
whether some of them would be
morally neutral if none of them would be morally good.
If all actions in the subcategory
would be morally bad, then God would not do any of them, and then we
can tell that God does not exist.
If some actions in the
subcategory would be morally neutral but none morally good, then we
have no means of telling whether God would do any of them.
If some actions in the
subcategory would be morally good, it's still not clear that we have
the means to assess whether God would do them.
Now,
Swinburne argues that what he calls "humanly free agents"
do have what he calls "significant free choice", but –
surprisingly – God, angels and some other beings do not.
Swinburne
seems to be saying that an entity that has a fixed morally good
character does not have significant free will. That is really odd,
because of the following reasons:
First,
if there is an omnipotent moral agent, he can
do
evil, even if he wouldn't
because
he's morally perfect; I already made that point earlier.
Second,
even leaving the previous point aside, Swinburne himself claims that
God could
create
a universe without what he calls 'humanly free agents', and thus
without beings with what he calls 'significant free choice'[27],
but he also claims that God could also create a universe with humanly
free agents.
So,
by
Swinburne's own account, God
could create a universe that will have moral evil, or refrain from
doing so, which seems to suggest that God can make choices that make
a difference between good or ill, which seems to imply that God has
significant moral freedom.
In order to avoid that, it seems
our conclusion here should be that Swinburne's stipulation that a
being with what he calls 'significant free choice' means that said
being sometimes (under some circumstances) would behave immorally.
But if having 'significant free
choice' is having a character such that, under some circumstances,
one would behave immorally, then it seems intuitive that an
omnipotent, omniscient moral agent would be behaving immorally if he
were to create such beings when he can instead create beings without
such flaw, and without any propensity to make beings with such flaw.
Regardless
of what Swinburne meant to say,
if someone with a fixed
morally good character does not have what Swinburne calls
"significant free will", because they do not and would not
act immorally, then it seems to me that creating beings with that
feature – which shouldn't be called 'significant free will',
but rather something like 'limited depravity', or something like that
– would be a morally bad thing, all other things equal, and for
an omnipotent, omniscient moral agent who could choose to make
morally perfect beings instead, as I said earlier.
Swinburne
also claims that significant free choices are those that can make
real differences for good or ill, but oddly claims that humans have
such choices because of non-rational influences and
temptations to do what's not good...which make them not what
Swinburne calls 'perfectly free'.
So,
in brief, what Swinburne's claim entails is that even when one could
just create morally perfect beings, it's morally good to create
beings with limited moral knowledge and temptations to act immorally,
in order to achieve the morally good end of...having entities with
limited moral knowledge and a temptation to act immorally, and who
would on occasion behave immorally!
Calling
those limitations and temptations "significant free choice"
might make the act of creation sound the behavior morally good, but
once the obscurity is removed to a sufficient extent at least, it
should be clear – I hope – that the claim ought to be
rejected.
Swinburne's
claim entails that, in a world of moral perfection, it's morally good
to introduce moral imperfection, even if one could choose not to do
so, and indeed to introduce more morally perfect beings instead, if
one so chose.
Granted,
many, plausibly most humans often do morally good actions, and I
wouldn't say that making humans or similar creatures would always be
morally bad if the creator were limited – but we're talking
about an omnipotent, omniscient one, who has the power to create
beings who do not have any inclination to do evil, and whose moral
beliefs are always true.
On
top of that, in order for the choices to be "significant",
it seems – according to Swinburne's idea – that it's also
a good thing to give these morally imperfect beings a certain even if
limited amount of power to do evil, and to inflict suffering on the
innocent.
How
can it be a good thing not only to introduce evil temptations into
the world, but even to allow people who act on them to make good
people, or non-moral agents, suffer?
What
would be the greater good? That someone resisted the temptation?
How
would bringing about a world full of beings of imperfect moral
knowledge and tempted to do evil – who often do good and
sometimes do a lot of evil – be morally good, or even
morally acceptable, if one can instead bring about a world of
morally perfect beings who would never do any evil?
After
carefully considering Swinburne's claim, I would say – for the
aforementioned reasons – that that claim not only isn't
evidently true, but it appears intuitively evidently false – to
me, at least.
Now,
Swinburne defends this claim by saying that it's a good thing that
our children make their own choices in good or ill, and that their
choices influence whether they're good or bad.
It's
true that humans usually want their children to learn and become good
people in the way limited beings like humans do learn, but generally
human parents do not want their children to be morally bad, or to
make immoral choices.
Still,
let's suppose we could choose to have children who will always make
morally good choices, or children who might do morally good or
morally bad choices, for all we know. Would it be okay to choose the
latter?
Moreover,
a human instinctive desire to have normal offspring may affect our
first impressions on the matter, but that does not affect the
previous considerations.
But
let's consider the following scenario, as an analogy:
Let's suppose we can make a
strong AI: an artificial being with far greater intelligence than a
human being – or even that all humans put together. We can do
the following:
1) We can make sure the AI is
going to be friendly, and in particular, it's going to be similar
enough to humans to be a moral agent (even if a vastly more
intelligent one), and will always do a morally good action.
2)We can make sure that the AI
will be a moral agent, but a certain number of times (say, at least
10% of the times, for instance), the AI would make a morally bad
action.
3) We can make part of the
programming random, so that we won't know in advance whether the AI
will always do a morally good action, or sometimes or even often
morally bad ones.
4) We can refrain from making any
such AI.
Isn't
it clear that both 2) and 3) would be immoral
courses of action for us to take?
Of course, this is only an
analogy, and hence not a perfect match.
So, someone might point out to
differences between that and the situation of an omnipotent,
omniscient moral agent in a world devoid of any imperfect moral
agent, from the differences in power and knowledge on one hand to the
fact that we don't live in a universe devoid of imperfect moral
beings in the first place, and suggest that they make a relevant
moral difference. But that was an analogy to reply to Swinburne's
analogy about parents.
In
the end, what matters is the actual case, not the analogies, and for
the reasons I've been explaining and my intuitive assessment of the
situation, I would say that even if we could tell that God
would in fact create other morally good beings, he would not create
what Swinburne calls 'humanly free agents'; in particular, God would
not create humans.
Granted, someone might say that
there might be some hidden sufficient moral reason to create humans
or humanly free agents that I have missed, that we're not omniscient
but Go is, and so on.
However, as I argued earlier, simply
claiming that there might be sufficient reasons is not enough to
block my point. Moreover, for that matter, someone might claim that,
perhaps, God would have sufficient reasons to bring about the actions
of, say, Pol Pot, who might have been for all we know rightfully
following his orders. Of course, such a claim is preposterous, and we
can tell that Pol Pot was a very immoral agent.
The
point of the analogy is not to imply anything about the degree
of
immorality involved in the creation of imperfect moral agents by a
moral agent that is omnipotent and omniscient – which would
actually depend on the specific action in the category
-, but to stress the point that if someone objects to a moral
assessment by saying that God might
have
sufficient moral reasons, they actually ought to provide plausible
reasons,
otherwise the clear moral assessment stands.
All that said, even if the
previous considerations were insufficient to conclude that God would
not create humans or what Swinburne calls beings with 'significant
free will', a weaker result is that no good reasons have been
provided to suspect that he would do so, and that weaker result is
enough, on its own, to block Swinburne's case for the existence of
God, independently of other considerations.
So, while my position is that the
arguments I've given above show that God would not create what
Swinburne calls 'humanly free agents', even if I'm wrong about that,
I would say that at least they show that Swinburne has not
provided good reasons to suspect that God would create such
beings.
In particular, if one were to
accept the objection to my previous argumentation based on the claim
that an omnipotent, omniscient moral agent just might have
morally sufficient reasons to bring about what Swinburne calls
'humanly free agents', even if we are unable to ascertain what those
reasons might plausibly be, then a similar objection would work in
the other direction: for that matter, an omnipotent, omniscient moral
agent just might have morally sufficient reasons to never,
ever create what Swinburne calls 'morally free agents', even if we
are unable to ascertain what those reasons might plausibly be (though
we actually are able to ascertain what those reasons plausibly are,
which is what I've been arguing for above, but leaving that aside now
just for the sake of the argument).
So, if that particular objection
were to be accepted, it seems we wouldn't be able to tell whether God
would have sufficient reasons one way or another. That would block
probabilistic assignments – a 0.5 assignment in such cases
would not work since, for that matter, the same "reasoning"
(i.e., mysterious sufficient moral reasons) would yield the same
assignment for, say, humans (not
just what Swinburne calls 'humanly free agents'), or a number of
other beings, as well as any sort of behavior, such as the chance
that, perhaps, Pol Pot was God behaving like that for mysterious
reasons.
We
may also add the problem of creating moral agents who endure
suffering they do not deserve – another
kind of action that, in my assessment, God would not carry out.
In
addition to the previous objections, there is yet another problem for
a theist defending Swinburne's argument, which is the following:
While
I do not think that what Swinburne calls 'perfect freedom' has
anything to do with the actual concept of freedom, the fact is that
Swinburne says that God is 'perfectly free' and is not limited by
irrational desires, and then goes on to say that what he calls
'significant free will' is 'a good thing' (apparently implying that
it would be morally good for God to create them), where that
so-called 'significant free will' would be a kind of freedom that a
'perfectly free' being would not
have,
since he's not limited by irrational desires...
So,
once again, the choice of the term 'significant free will' might make
it look the idea look appealing at first glance, and even give the
impression that it would be morally good for an omnipotent,
omniscient, moral agent to create beings with such 'significant free
will', but once the obscurity of Swinburne's words is removed to a
sufficient extent, it's clear that this so-called 'significant free
will' amounts – and based on Swinburne's own account freedom –
to a kind of imperfect freedom, which falls short of perfect freedom
due to irrational impulses. How would it be morally good for an
omnipotent, omniscient moral agent to create that,
when he can create other perfectly free agents?
j)
It would not be worse for God to create humanly free agents than to
create other divine or semi-divine beings. [27]
For
all of the reasons given above, I disagree.
It
would be much better for an omnipotent, omniscient moral being to
create beings with full moral knowledge and no inclination whatsoever
to do evil (i.e., morally perfect ones), than to create "humanly
free" agents (i.e., beings with limited freedom and moral
awareness, and apparently with inclinations to do evil and such that
at least some of them will surely do some moral evil).
k)
God
has reason to create a beautiful physical universe, independently of
any reasons to create 'humanly free' agents, and even if there were
no one but God himself to see that world. [28]
While
Swinburne says that "of course" God has reasons to make a
physical universe, actually Swinburne does not have any good reason
whatsoever for even suspecting that God would have such reasons,
since whether God would be inclined to make such a thing would depend
on features of his mental makeup other
than moral perfection, or anything that is part of the definition of
"God", or follows from it.
Therefore, an even
more complex hypothesis than theism – such as some particular
version of theism – would be required if we are to make an
assessment as to whether God would create a material world.
That extends beyond
what Swinburne apparently calls 'the physical universe', since it
applies to lifeforms as well.
But purely for
example, and to illustrate my point here, let's consider the case of
a beautiful rose, which Swinburne considers as part of the inanimate
world, apparently, as the quote above implies (though whether a rose
counts is beside the point here; we could pick anything else in what
Swinburne calls the 'inanimate world').
Do we have any good
reason to suspect, a priori, that God would have any motivation
whatsoever to create such a thing?
Let's consider the
following scenario:
On
a distant planet, some intelligent aliens (say, kelorans) evolve. As
a result of a different evolutionary past, any normal keloran would
find anything that looks like a rose nauseating,
and some alien plant-like organism (let's call it a 'k-rose') very
appealing, while any normal human would find a k-rose nauseating as
well.
Assuming that kelorans also have a moral sense and are
assessing the prior probability of God (I see no good reason to think
they would, but leaving that aside), then we could ask:
Would kelorans be
justified in assessing that, say, God would be likely to make
k-roses, or more likely to make k-roses than something as nauseating
to a keloran as a rose?
Clearly,
the answer is no,
kelorans would not be so justified.
But then, and for
the same reasons, humans are not justified in assessing that roses
are more likely created by God than k-roses. We have no information
about God's predispositions to behave other than his moral
perfection.
Note
that the previous example does not require 'assuming' unguided
evolution. It's enough to point out that the kelorans are
conceivable,
and conceivably rational, even if they would find a rose nauseating.
Of course, the
species 'rose' is not really the point here.
The point is that,
as far as we can tell by the meaning of the words:
1) There might be
rational and moral agents (and even morally good ones) who find roses
appealing and k-roses nauseating.
2) There might be
rational and moral agents (and even morally good ones) who find
k-roses appealing and roses nauseating.
3) There is no good
reason to suspect that agents such as those described in a) are more
or less probable than those described in b), given that words like
'rational', 'moral', 'morally good', etc., have no implications
whatsoever regarding how appealing an agent would find either roses
or k-roses.
Of course, the
previous example of roses is merely that: an example.
We
may well also conceive of rational, moral and morally good beings
that would look at what Swinburne calls the 'inanimate world' and
find it nauseating.
The
crucial point
here is that
there is nothing
in the concept
of
moral goodness, and/or the concept of rationality, that would provide
any information as to whether an entity would find the 'inanimate
world' (or, for that matter, roses) appealing, nauseating, or would
just be indifferent to them.
So,
Swinburne's claim that God has good reason to create a 'beautiful
inanimate world' is unfounded. In fact, we ought
not to make that assessment; rather,
and for the aforementioned reasons, we ought to conclude that we
cannot tell or even probabilistically assess whether God would have
reasons to create an inanimate universe that looks at all like the
inanimate parts of our universe (I'm limiting this to the inanimate
parts to assess the matter without contamination from the problem of
suffering, pain, etc.).
Granted,
Swinburne also claims that God has moral
reasons to
create such a universe, as that would be required to make what he
calls 'humanly free agents', and it seems Swinburne believes it would
be morally good for God to bring about such agents, at least under
certain conditions.
However,
that's another matter, which I have already
addressed.
l)
In general, it's better to create more humanly free agents, more
animals, etc. In other words, in general (though not always), the
greater the number, the better. [29]
There appears to be
no good reason to think there is any moral dimension to the numbers,
if we assumed that creating such entities when one can make morally
perfect ones instead would be morally acceptable.
However,
in the case of "humanly free" agents, it seems to be
immoral for an omnipotent,
omniscient
moral agent to
create them, for the reasons I explained
above,
so God would not create them. Given that such entities exist, we can
tell that
God
does not exist.
m)
The probability that there will be other beings, if God exists, is 1.
[30]
For
the reasons explained previously
in this section,
it appears to me that there is no good reason to assume so – at
least, no good reason provided by Swinburne.
n)
The probability that there will be "humanly free" agents,
if God exists, is 1/2. [30]
For
all the reasons I
gave
above,
in my assessment the probability that God, who is omnipotent and
morally perfect, would create what Swinburne calls 'humanly free'
agents, is 0,
or at least nearly 0.
2.3.
Conclusion of Part 2
Based
on the previous arguments in this section,
it seems we're justified in concluding that God would not create
humans, and so he does not exist.
However,
that conclusion is not required: a far weaker one suffices, namely
that the aforementioned arguments at
least show
that Swinburne fails to give good reasons for his assessments about
what God would likely create.
3.
More probabilistic assessments, and more theistic improbability
In
this part, I will tackle Swinburne's probabilistic assessments from a
different perspective.
I
will not challenge Swinburne's moral assessments.
In
fact, I will even grant some of them for the sake of the argument,
and reason from there. Moreover, I will also grant some of his
probabilistic assessments for the sake of the argument. Of course,
those concessions are limited to this part of my reply to Swinburne's
case, unless otherwise specified.
So,
let's consider some of Swinburne's claims:
In
order to make his assessments about what kind of worlds God would be
likely to bring about, Swinburne makes the following claim:
P1:
There is no best possible world. [31]
According
to Swinburne, if there were a best possible world, God would bring it
about, since bringing about such a world would be the only morally
best action, and God only does such an action when there is one (that
follows from the definition of 'God' [10]).
However,
Swinburne claims that there is no such unique best world, and
supports that claim by saying that a world with another Swinburne
that does exactly what he does, all other things equal, would not be
a better or worse world, and the same would happen for other worlds.
So,
Swinburne concludes that there is no unique best act of creation of a
world.
Let's
grant that for the sake of the argument; Swinburne also makes the
following, stronger claim:
P2:
There is no world W such that no other world W' is better. [23]
Swinburne
says that even if there is no unique best world, if there were worlds
W such that no world is better (even if others are just as good),
then it would be an equal best act to create such a world, and
allegedly, God would carry out at least one of those acts of
creation.
However,
he concludes that there is no such world W: for every world W, there
is a better one W'. Let's also grant that for the sake of the
argument, and let's consider two more claims Swinburne makes:
P3:
If God faces a choice n possible, mutually exclusive and equally best
actions, the probability that he will carry out any specific one of
them is 1/n. [30]
P4:
If God faces a choice n possible, mutually exclusive and equally best
kinds of actions, the probability that he will carry out at least one
action of any specific one of those kinds is 1/n. [30]
For
the sake of the argument, let us grant P3 and P4 as well.
3.1.
The improbability of God's creations: possible worlds as an example
Let's
suppose that God has a choice between two equally morally best
actions, A(1) and A(2), that {W(n,1)} is a finite set of equally good
worlds, and that the cardinality of the set is m.
What
can we say about the probability that God would bring about W(j,1),
for a specific j?
Given
that it's equally good to bring about W(j,1), for any j, mirroring
the reasoning that apparently led Swinburne to P3,
it seems that the probability of God's bringing about W(j,1) is no
greater that the probability of any other W(i,1) in the set.
Hence,
the probability of each one is no greater than 1/m.
Let's
introduce a variant: {W(n,2)} is a finite sequence of worlds such
that W(j+1,2) is better than W(j,2), and the cardinality of the set
is m.
Then,
and for the same reasons, it seems that the probability that God
would bring about W(j,2) is at least no greater than the probability
that God would bring about a better world, namely W(j+1,2).
In
particular, the probability that God would bring about W(1,2) is no
greater than 1/m.
Combining
this with the claim that, for every world
W, there is a better one W', it follows at once that, for every
world W, the probability that God would bring about W is no greater
than 1/m, for any natural number m.
This
result is merely an example, but the kind of reasoning used to
establish will be useful as a means to establish more relevant
results.
3.2.
Theism and the probability of humans
As
Swinburne claims [35],
if there is a best action to bring about, God will bring it about.
But
Swinburne also maintains that if there is a best kind
of
action to carry out, then God will carry out an action of that kind.
Based on that, Swinburne separates actions in different kinds, in
order to make his probabilistic assessments.
In
particular, Swinburne concludes that God will create something else,
because – Swinburne says – the kind of action consistent
in creating something is a better kind than the kind consistent in
not creating anything, and given that they are mutually exclusive,
the kind of action of creating something is a best kind of action.
Also,
Swinburne makes a case in support of his contention that the category
(or type, or kind) of action of bringing about HFA
is neither a better type or a worse type than bringing about other
kinds of worlds. In other words, Swinburne maintains that to create
and not to create HFA are equally best kinds of
actions. [27]
Based
on that, he assigns probability 0.5 to the hypothesis that God will
create HFA.
Now,
let P(k) be the intrinsic probability of theism P(HFA) be the
probability that there would be some 'humanly free' agents given
theism, and P(Hu)
for humans.
Going
by the same reasoning, someone might say that to create some humans
and
not to create some humans
(in
addition to whatever else he brings about) would be acts of equal
best kinds), resulting in a 0.5 probabilistic assessment for humans.
In other words, P(Hu
│k)
= 0.5.
Granted,
humans are a subcategory of HFA, but that's not the point: I'm merely
defining categories or types or kinds of actions that aren't the same
defined by Swinburne, but then making probabilistic assessments about
them by his own reasoning.
Alternatively,
there is another way to see that Swinburne's reasoning and claims
entail that P(Hu │k) = 0.5 Given that the existence
of humans entails the existence of HFA, then P(Hu │k)
≤ P(HFA│k) = 0.5.
Hence,
P(Hu │k) ≤ 0.5.
So,
the kind of action of 'to create humans' is either of equally best or
of worse kind than the kind of action 'not to create humans'.
Now,
if the kind of action 'not to create humans' is a better kind of
action than the kind of action 'to create humans', then the kind of
action 'not to create humans' would be a best kind of action, and
then God would never create humans. It would follow, then, that the
kind of action consistent in not creating humans would be a best kind
of action, and thus God would never create humans. But Swinburne
claims that God created humans.
Then,
we have P(Hu│HFA&k) = P(Hu&HFA&k)
/ P(HFA&k)= P(Hu&k) / P(HFA&k) = (P(Hu│k)
* P(k)) / (P(HFA│k)*P(k)) = 1.
That
would mean that the probability that, on theism, God would create
humans given that he creates 'humanly free' agents (henceforth, HFA)
is 1. Moreover, picking humans was just a choice among infinite
possibilities for HFA; the point here is that this would mean that
the probability of any
specific HFA
is 1 on theism, given that there are at least some HFA, at least as
the powers of the specific HFA in question are limited as those of
humans are. Now, let H1
be some
conceivable HFA, with such limited powers.
Then,
as in the case of humans, P(H1│k)
= 0.5, and P(H1│HFA&k)
= 1.
Then,
P(H1│HFA)
≥
P(H1&k│HFA)
=P(H1&k&HFA)
/ P(HFA) = P(H1
│HFA&k)
* P(HFA&k) / P(HFA) = P(k│HFA).
In
other words, that implies that the probability that theism is true
given that there are HFA (as there are) is no greater than the
probability of any
conceivable
species H1
of
HFA given that there are HFA, as long as the powers of H1
are
sufficiently limited.
Granted,
the theist might argue that other pieces of evidence (apart from the
existence of HFA) increase the probability of theism, but still, this
result is very
problematic
regardless.
Moreover,
there is another problem: since
P(H1│HFA&k)
= 1, that would commit a theist to accepting the existence of H1,
for any such conceivable beings, as long as they accept Swinburne's
reasoning to the conclusion that P(HFA│k) = 0.5, since the same
kind of reasoning leads to P(H1│k)
= 0.5.
There
is yet another problem, which I will address below.
3.2.1.
What's a 'type' or 'kind' of action?
Someone
might raise the following objection to the
previous reasoning: they might suggest that 'to create humans'
and 'not to create humans' aren't kinds of actions in the sense
Swinburne uses the term. But how would that be so?
Swinburne
picks categories, or kinds of actions, as he consider them useful to
make the points. For instance, he considers the types of action of
creating only members of particular species, such as lions or tigers,
even if in that case only to conclude that none of them constitutes a
best kind. [24]
So,
it seems that Swinburne defines any category of actions as the
actions that have a certain property, and then he assesses whether
that category is a best kind, which is no more or less than
what I'm doing here by dividing the categories 'to create humans' and
'not to create humans'.
The
previous conclusion can be supported also by Swinburne's reasoning on
page 116, and what follows:
In
particular, he contends that in order to be able to assess what God
must bring about, we would need to separate substances into finitely
many kinds such that creating substances of each of those kinds
beyond a certain amount of goodness is either better or just as good
as creating members of any level of goodness of a class that does not
include those types.
But
that's exactly what I'm doing in the case of humans:
First,
I divide substances in two types, kinds or categories: humans and
non-humans.
Second,
I ask whether creating substances of type 'non-humans' beyond a
certain level of goodness, is better than creating any number of
substances in the category 'humans', which is pretty much what
Swinburne is doing in the case of HFA, only defining a different
category.
If
it is better, then not creating humans is a best kind of action, it
seems, so God would do it with probability 1.
Otherwise,
since not creating humans is not of worse kind than creating them
(else, God would create humans with probability 1), then it follows
that creating humans and not creating them are actions of equal best
kinds, and so going by Swinburne's own method for assigning
probability in such cases, the intrinsic probability that God would
create humans seems to be 0.5.
Granted,
Swinburne claims he does not rely on exact values, so perhaps the
probability isn't exactly 0.5, but very close to it. But that does
not affect my reasoning in any significant
manner,
since the problem for theism that I raised earlier
would only be very slightly modified, and would remain a serious
problem, and the problem I will raise later
would not be affected in the least, since it only requires that we
conclude from Swinburne's position that the intrinsic probability
that God would create humans be greater than 1/m, for some natural
number m – any
m,
not necessarily 2 -, and that conclusion is reachable by the
reasoning given in this section, regardless of whether 1/2 is the
exact number.
It
is true that Swinburne focuses on four types of substances, which he
considers "important": namely, inanimate substances,
animate substances without free will or moral awareness, animate
substances with limited knowledge, free will and power, plus moral
awareness, and animate substances with unlimited knowledge, free will
and power. [26]
Oddly,
Swinburne also claims that a substance with unlimited free will and
power would lack what he calls 'significant free will'. But I've
dealt with that in part 2. so leaving that aside, the point here
is that Swinburne chooses to focus on those types substances that he
considers important to make his probabilistic assessments. That's
fair enough, but I'm just as entitled to focus on types of substances
that I consider important as a means to refute his case for theism.
Still,
if someone objected to my choice of types or categories and claimed
that only some specific categories of actions can be used to assess
the probability that God would engage in one of them, the burden
would be on them. If we have the means to make probabilistic
assessments in the case of those categories of actions, why not in
other mutually exclusive categories?
As
a side note, I will point out that Swinburne's types do not
appear
to be exhaustive of all conceivable substances, and Swinburne
provides no good reason to suspect that they're exhaustive of all
possible ones.
In
particular, a substance of limited power but unlimited knowledge
would not fall in any of Swinburne's categories or types. Yet, such a
substance appears to be conceivable is an omniscient one is, at least
prima
facie,
and Swinburne has not made a case against either the conceivability
or the possibility of such substances.
3.3.
'Humanly free' agents, physicality, and dimensionality
According
to Swinburne HFA need to have physical bodies [28][30].
Let's
say, for the sake of the argument, that Swinburne is right about
that, and so HFA need to have physical bodies.
That
does not tell us anything about the number
of spatial dimensions of
those physical bodies. We have 3-dimensional physical bodies[32],
but surely if God exists, he would not be limited to creating
3-dimensional bodies.
While
our limited cognitive capabilities prevent us from imagining
say,
4-dimensional physical bodies, even we can conceive
of
them, and God, being omnipotent and omniscient, could make
4-dimensional physical bodies, or (3+k)-dimensional ones for that
matter, for any natural number k.
3.4.
Dimensionality and some probabilistic assessments
Let
us now consider the following categories or types of worlds, and
corresponding types of actions, for each natural natural number k. If
HFA
require at least k(0) dimensions for some k(0), then let's stipulate
that k is no less than k(0).
Cl(k):
A world W is of type or category Cl(k)
if and only if there are k-dimensional HFA in W,
but there are no j-dimensional HFA for any j<k.
Tl(k):
An action is of type Tl(k)
if it's the action of bringing about a world of type Cl(k).
Clearly,
the types Cl(k)
are
mutually disjoint, and the types of actions, incompatible ones.
It
seems that Cl(k)
is not a better or worse type of worlds than Cl+1(k),
and the corresponding types of actions are not actions of a better or
worse type, for the following reason:
If
a world W is of category Cl(k),
we may consider another world W', as follows.
Let a j-dimensional realm be any
maximal j-dimensional part of W (though maximality is not required
for my argument, it simplifies matters a little):
i)
If there are HFA in a j-dimensional realm in W,
there is a (j+1)-dimensional realm that corresponds to it in W' and
which will also have HFA, in similar or greater
numbers (whatever is better), and restricting their power to do evil
just as much (i.e., as good as possible).
ii)
If there are no HFA in a j-dimensional realm in W,
then there is either a similar realm in W' as
well,
or a (j+1)-dimensional realm with no HFA, either,
but much greater numbers of things. Assuming with Swinburne non-moral
agents can be categorized as 'good' somehow in a sense that's
relevant to Swinburne's case[34], we
also stipulate that W' has many more good things that aren't moral
agents in those realms.
iii)
In addition to the above, W' also has many (say, 1000000000) extra
k-dimensional physical universes without HFA, full
of good things if it make sense to count such things and/or their
existence as 'good'.[34] If only one
such universe is possible, than W' has one extra such universe.[34]
So, W' is no worse than W.
Still,
if the above reasoning is not enough to show that Cl+1(k)
is no worse than Cl(k)[34],
then at least it shows that there is a subcategory C'l+1(k)
of Cl+1(k)
that is no worse than Cl(k)
-
namely, the subcategory constituted by the W' that correspond to each
W in Cl(k)
-,
and a corresponding subtype of actions, and the following
probabilistic result obtains as well:
The
intrinsic probability that God would carry out an action of type
Tl(k)
is
no greater than 1/m, for any natural number m, since actions of type
Tl(k)
and
Tl(p)
are
mutually exclusive if p≠k,
and neither of them is more probable than the other.
Someone
might object to the previous reasoning and say that it's not possible
for there to be beings of different dimensionality in the same world.
I do not see why that would be impossible, but if it is impossible,
then the construction is even simpler, because in that case any world
W in Cl(k)
only
contains k-dimensional physical beings, and we may consider a world
W' with only (k+1)-dimensional ones, also with similarly limited HFA,
but many more good things, just with one more dimension. Then, at the
very least, W' is no worse than W, and the previous
conclusion follows as before.
An
alternative objection might claim that that a world W of type Cl(k)
may
have some good that a world of type Cl(k+1)
is
deprived, like k-dimensional HFA, and that makes
the world W better than any alternative W'.
However, that second objection
does not succeed, either, for the following reasons:
First,
assuming with Swinburne (against good
reasons, as argued in part 2) that there is something good about
an omnipotent, omniscient moral agent bringing about HFA
into a world devoid of them, and that that goodness consists in what
Swinburne (improperly) calls 'significant
free choice', then the alternative world W' contains that as well,
and in similar or greater numbers (whatever is better).
Second,
any kind of non-moral
goodness
(whatever that is) that those k-dimensional moral agents may have in
W can be matched or surpassed either by some (k+1)-non-moral goodness
in W', or by the goodness of some non-moral
k-dimensional
things in W', which we may add as required since we're talking about
some non-moral
goodness,
assuming that somehow that is relevant to the argument at hand. [34]
So,
the conclusion remains: the intrinsic probability that God would
carry out an action of type Tl(k)
is
no greater than 1/m, for any natural number m.
In
particular, this implies that the intrinsic probability that God
would create humans
is
less than 1/m, for any natural number m, since the category of worlds
with humans is contained in the union of the types Cl(j),
for all natural numbers j equal or less than the dimensionality of
humans, and so the probability that God would create humans is no
greater than the probability that he would carry out an action of
some type Tl(j),
for some natural number j equal or less than the dimensionality of
humans.
Since
the probability of the latter is no greater than 1/m, it follows that
the probability that God would create humans is less than 1/m.
But
as I argued above, Swinburne's own
reasoning leads us to the conclusion that the probability that God
would create humans P(Hu│k)
= 0.5, contradicting the conclusion that P(Hu│k)≤(1/m),
for all natural numbers m.
3.5.
Conclusion of Part 3
Given
the arguments given in this part of the
article, we ought to reject Swinburne's case for theism, on the
grounds that the reasoning leading to his probabilistic assessments
contains decisive errors.
Notes
and references
[1] Swinburne,
Richard "The Existence of God", Second Edition.
Clarendon
Press Oxford
[2]
Page 53.
[3]
Page 55.
[4]
Page 56.
[5]
Page 86.
[6]
Page 65.
[7]
Page 72.
[8]
Page 239.
[9]
Of course, we don't need to claim it's a brute fact, or that it is
not. We may remain undecided on the subject.
[10]
Page 7.
[11]
Page 94.
I
would rather call that "perfectly random", but that would
be beyond the scope of this article.
[12]
Page 98.
[13]
Page 100.
[14]
Page 101.
[15]
Page 104.
[16]
I will not define "deity", even though the concept is very
vague at best.
If
that's a problem, we can call call these entities "other beings"
- for instance -, and the arguments are not affected.
I
just thought "deities" was more fitting, since it stresses
the fact that I'm providing alternatives to theism.
[17]
In fact, one needs much less than P(AN(Con(1,n))│k)
≥ P(Theism│k)
It
would be enough to take – for instance – something like:
P(Theism│k)
≤P(AN(Con(1,n))│k) * (n*(log(n+20)+(10^100)))
[18]
Page 17.
[19]
Actually, Swinburne
does claim that an "inbuilt detailed specification of how to
act" would make a hypothesis more complex than one stipulating
perfect freedom.
However, moral
perfection is
such as specific account, so we have to assume that such a
specification does not preclude perfect freedom – else, God
would be logically impossible.
[20]
We can even further simplify the hypotheses, if we so choose:
AH(Alternative
Deity(n,1)): There exists a being who is V(n)-perfect and the creator
of all other beings.
Since
there cannot be two different creators of all other beings,
AH(Committed Deity(n,1)) and AH(Committed Deity(m,1)) are disjoint if
n≠m, and the rest of the argument follows as before.
[21]
We
can even further simplify the hypotheses, if we so choose:
AH(Gambling
God(n,1)): There exists a being who is the creator of all other
beings and W(n)-perfect.
Since
there cannot be two creators of all other beings, AH(Gambling
God(n,1)) and AH(Gambling God(m,1)) are disjoint if n≠m,
and
the rest of the argument follows as before.
[22]
Page 119.
[23]
Page 115.
[24]
Page 116.
[25]
Someone might object that, under some circumstances, it's not morally
neutral to choose an apple instead of a banana – there might be
a disease that spreads through apples, etc.
However,
I'm talking about a case in which all other things are equal.
In
other words, my point is that the choice of – say – an
apple over a banana, when there are no other factors that might be
morally relevant, is morally neutral.
The
same applies to other cases in which I say that some actions are
morally equivalent, or neutral.
[26]
Page 118.
[27]
Page 120.
[28]
Page 121.
[29]
Page 122.
[30]
Page 123.
[31]
Page 114.
[32]
Someone might raise the issue of multiple dimensions in our
universe and suggest that, perhaps, our physical bodies are not
3-dimensional after all. However, that issue is not relevant to the
points I'm trying to make. If our bodies have m dimensions, for any
natural number m, for that matter God could create physical beings
with (m+1)-dimensional bodies.
[33]
In this and other cases, unless otherwise specified, I'm considering
all the physical beings at that world across time. For instance, if
at world W there are only 3-dimensional physical beings before time
t1
and
only 4-dimensional ones after t1,
that world would not be in category C(3) or C(4).
[34]
In reality, as I argued in part 2, I
think all this talk about better types of action and that conflation
of morality and things completely unrelated to morality is a
confusion.
Moreover,
the talk of value seems confused as well.
Swinburne
seems to have conflated the issue of whether moral statements state
true propositions with the existence of some mind-independent value
(whatever that might be), and then conflate moral and non-moral
'value'.
However,
at this point in my argument I'm following Swinburne just for the
sake of the argument.
[35]
Page 113.
