Gary Gutting recently interviewed Alvin Plantinga in the pages of The New York Times and brought up the business about Bertrand Russell's celestial teapot. The following response of Gutting to Plantinga comes early on in the interview:
G.G.: You say atheism requires evidence to support it. Many atheists deny this, saying that all they need to do is point out the lack of any good evidence for theism. You compare atheism to the denial that there are an even number of stars, which obviously would need evidence. But atheists say (using an example from Bertrand Russell) that you should rather compare atheism to the denial that there’s a teapot in orbit around the sun. Why prefer your comparison to Russell’s?
Russell's comparison has long struck me as lame, and so I want to revisit and rethink this topic. What follows is an old post from August 2010 amended and substantially expanded:
Gutting, Dawkins, and Russell's Celestial Teapot
In his recent NYT Opinionator piece, On Dawkins's Atheism, Notre Dame's Gary Gutting writes, describing the "no arguments argument" of some atheists:
To say that the universe was created by a good and powerful being who cares about us is an extraordinary claim, so improbable to begin with that we surely should deny it unless there are decisive arguments for it (arguments showing that it is highly probable). Even if Dawkins’ arguments against theism are faulty, can’t he cite the inconclusiveness of even the most well-worked-out theistic arguments as grounds for denying God’s existence?
He can if he has good reason to think that, apart from specific theistic arguments, God’s existence is highly unlikely. Besides what we can prove from arguments, how probable is it that God exists? Here Dawkins refers to Bertrand Russell’s example of the orbiting teapot. We would require very strong evidence before agreeing that there was a teapot in orbit around the sun, and lacking such evidence would deny and not remain merely agnostic about such a claim. This is because there is nothing in our experience suggesting that the claim might be true; it has no significant intrinsic probability.
But suppose that several astronauts reported seeing something that looked very much like a teapot and, later, a number of reputable space scientists interpreted certain satellite data as showing the presence of a teapot-shaped object, even though other space scientists questioned this interpretation. Then it would be gratuitous to reject the hypothesis out of hand, even without decisive proof that it was true. We should just remain agnostic about it.
The claim that God exists is much closer to this second case. There are sensible people who report having had some kind of direct awareness of a divine being, and there are competent philosophers who endorse arguments for God’s existence. Therefore, an agnostic stance seems preferable to atheism.
I have a serious problem with Gutting's response to the Russell-Dawkins tag team. Gutting concedes far too much in his reply, namely, that it even makes sense to compare the claim that there is an orbiting teapot with the claim that God exists. Instead of attacking this comparison as wrongheaded from the outset, Gutting in effect concedes its aptness when he points out that, just as there could be (inconclusive) scientific evidence of a celestial teaspot, there could be (inconclusive) experiential and argumentative evidence for the existence of God. So let me try to explain why I think that the two existence claims ('God exists' and 'A celestial teapot exists') are radically different.
If someone asserts that there there is a celestial teapot orbiting the Sun, or an angry unicorn on the far side of the Moon, or that 9/11 was an 'inside job,' one will justifiably demand evidence. "It's possible, but what's your evidence for so outlandish a claim?" It is the same with God, say many atheists. The antecedent probability of God's existence, they think, is on a par with the extremely low antecedent probability of there being a celestial teapot or an irate lunar unicorn, a 'lunicorn,' if you will.
But this is to assume something that a sophisticated theist such as Thomas Aquinas would never grant, namely, that God, if he exists, is just another being among the totality of beings. For Aquinas, God is not an ens (a being) but esse ipsum subsistens (self-subsistent Being). God is not a being among beings, but Being itself. Admittedly, this is not an easy notion; but if the atheist is not willing to grapple with it, then his animadversions are just so many grapplings with a straw man.
Why can't God be just another being among beings in the way an orbiting teapot would be just another being among beings were it to exist? I hope it is clear that my point is not that while a teapot is a material object, God is not. That's true, of course, but my point cuts much deeper: if God exists, he exists in a way dfferent from the way contingent beings exist.
First of all, if God exists, then God is the metaphysical ground of the existence of every contingent being. Every such being on classical theism is continuously maintained in existence by the exercise of divine power. Thus every contingent being is radically dependent for its existence on divine activity. The same cannot be said about an orbiting teapot. If 'ontic' means pertaining to beings, and 'ontological' means pertaining to the Being of beings (the esse of entia), then 'God exists' is both ontic and ontolological. It says that there is a being possessing such-and-such divine attributes, but it also says something about the Being of what is other than God, namely, that its Being is createdness, a form of continuous ontological dependency. 'An orbiting teapot exists,' however is merely an ontic claim. It says or implies nothing about the Being of anything distinct from it. Now this difference between an ontic-ontological claim and a merely ontic one strikes me as very important. It is a difference that throws a spanner into the works of such facile comparisons as Russell's.
Second, on some accounts necessarily existent abstracta are also dependent on God. If (Fregean) propositions are divine thoughts then they are dependent on God despite their metaphysical necessity. The exist necessarily, but they have their necessity not from themselves but from another. Not so for the teapots and the unicorns.
Third, God is not only the ultimate ground of all beings, both contingent and necessary (except himself); he is also the ultimate ground of the intelligibility of all beings, of their aptness to be understood by us or anyone, their aptness to be subjects of true predications. Propositional or sentential truth is made possible by ontic truth, the intelligibility of that which is veridically represented by true propositions. But I don't think one would want to say that an angry unicon on the far side of the moon is the ultimate ground of intelligibility.
Fourth, God is the ultimate source of all value.
Fifth, God is the all-pervasive One, immanent in each thing yet transcendent of all things. This is not true of lunar unicorns and celestial teapots. If there is a lunar unicorn, then this is just one more isolated fact about the universe. But if God exists, then everything is unified by this fact: everything has the ground of its being and its intelligibility and its value snd its unity in the creative activity of this one paradigmatic purely spiritual being, a being who does not have existence like a teapot but is its existence
So, on a sophisticated conception, God cannot be just one more being among beings. The Source of being is not just another thing sourced. The ground of intelligibility is not just another intelligible item. The Thinker behind every thought is not just another thought. The locus and source of all value is not just another valuable thing. The One is not just another member of the Many.
These differences between God classically conceived and outlandish specimens of space junk is connected with the fact that one can argue from general facts about the universe to the existence of God, but not from such facts to the existence of lunar unicorns and celestial teapots. Thus there are various sorts of cosmological argument that proceed a contingentia mundi to a ground of contingent beings. But there is no similar a posteriori argument to a celestial teapot. There are also arguments to God from truth, from consciousness, from apparent design, from desire, from morality, and others besides. But as far as I know there are no similar arguments to teapots or unicorns or flying spaghetti monsters.
The very existence of these arguments shows two things.
First, since they move from very general facts (the existence of contingent beings, the existence of truth) to the existence of a source of these general facts, they show that God is not a being among beings, not something merely in addition to what is ordinarily taken to exist. Affirming and denyng the existence of God is not simply a matter of adding to or subtracting from a pre-given ontological inventory. For God does not make a merely ontic difference, but an ontological one as well. The existence of God changes the ontology. For if God exists, then the Being of non-divine entities is createdness, hence different from what it would be were there no God. Socrates is a being whose existence/nonexistence makes no difference to the system of ontological categories, and no difference to the nature of existence, property-possession, etc. God, however, is a being whose existence/nonexistence does make such a difference.
Second, these arguments give positive reason for believing in the existence of God. Are they compelling? No, but then no argument for any substantive philosophical conclusion is compelling.
People like Russell, Dawkins, and Dennett who compare God to a celestial teapot betray by so doing a failure to understand, and engage, the very sense of the classical theist's assertions. To sum up. (i) God is not a gratuitous posit in that there are many detailed arguments for the existence of God; (ii) God is not a physical being; (iii) God is not a being who simply exists alongside other beings. In all three respects, God is quite unlike a celestial teapot, a lunar uncorn, an invisible hippopotamus, and suchlike concoctions.
God is a not a being among beings, but the very Being of beings. To deny God, then, is not like denying an orbiting teapot; it is more like denying Being itself, and with it, beings. Or it is more like denying truth itself as opposed to denying that a particular proposition is true.
One who appreciates this ought to find discussions about the antecedent probability of theism as compared to teapotism faintly absurd. The question of the antecedent probability of something like Russell's teapot makes sense and has an easy answer: very low! The question of the antecedent probability of there being truths has no clear sense. The probability of a proposition is the probability of its being true. Hence, that there is truth, or that there are truths, is a presupposition of any meaningful talk of probability. It is therefore senseless to ask about the antecedent probability of there being truths, and the following answer is clearly absurd: the antecedent probabilty of there being any true propositions is extremely low.
Now my point is that the God question is like the truth question, not like the teapot question.
Unfortunately, the line I have sketched here will be rejected both by all atheists, but also by many theists, those theists who think of God as a being among beings, uniquely qualified no doubt, but no different in his Being or in the way he has properties than any other being qua being. Or, in the quasi-Heideggerian jargon employed above, these theists will say that 'God exists' is an ontic, not an ontic-ontological claim, and as such no different than 'Socrates exists' or 'Russell's celestial teapot exists.'
And the widely-bruited 'death of God?' It is an 'event' of rather more significance than the discovery that there is no celestial teapot (or Santa Claus, or . . . ) after all. As Nietzsche observed, the death of God is the death of truth.
I issued the following challenge: translate 'Something exists' into standard first-order predicate logic with identity. This is the logic whose sources are Frege and Russell. So I call it Frege-Russell logic, or, to be cute, 'Fressellian' logic. My esteemed commenters don''t see much of a problem here. So let me first try to explain why I see a problem. I then consider David Brightly's proposal.
1. First of all, 'Something exists' cannot be rendered as 'For some x, x exists.' This is because 'exist(s)' is not an admissible first-level predicate in Frege-Russell logic. The whole point of the Fressellian approach is to make 'exist(s)' disappear into the machinery of quantification. There is no such propositional function as 'x exists.' 'For some x, x exists' is gibberish, syntactic nonsense in Frege-Russell logic.
2. But the following is not gibberish: 'For some x, x = x.' So one will be tempted to say that 'Something exists' can be rendered as 'For some x, x = x,' ('Something is self-identical') and 'Everything exists' as 'For all x, x = x' ('Everything is self-identical').
But this won't work either. It is true that everything that exists is self-identical, and vice versa. But it doesn't follow, nor is it true, that existence is self-identity. Here is one consideration. When I say of Tom that he exists, I am not saying that he is self-identical. Suppose I hear a false rumour to the effect that Tom is no more. But then I encounter him in the flesh. I exclaim, "You still exist!" Clearly, "You are still self-identical" does not mean the same. If I said that, Tom might retort, "What the hell, man, were you worried that I had become legion?" In some circumstances, that a man should continue in existence is surprising. But we are never surprised by a man's continuing in self-identity.
Furthermore, when Tom ceases to exist, he does not become self-diverse. Loss of existence is not loss of self-identity. To put the point in formal mode, after his demise 'Tom' continues to refer to one and the same individual, Tom. The bearer of the name is gone, but not the reference. Otherwise it could not be true that Tom is gone. There is also a modal consideration. Tom is a contingent being: he exists but he might not have existed. If existence is self-identity, then Tom's possible nonexistence is Tom's possible self-diversity -- which is absurd. It makes prima facie sense to say of an individual that it might not have existed or that it no longer exists; but it make no sense at all to say of an individual that it might not have been self-identical or that it is no longer self-identical. If Tom might not have existed, then it is Tom who might not have existed. But if Tom might not have been self-identical, then it is not Tom who might not have been self-identical.
So, even if everything that exists is self-identical and conversely, existence is not self-identity. When we say that something exists we are not saying that something is self-identical, and when we say that everything exists we are not saying that everything is self-identical. I conclude that 'Something exists' is not expressible in the terms of the Frege-Russell system. As for 'Everything exists,' it is surely a presupposition of the whole Frege-Russell approach: the approach presupposes that Meinong was wrong to speak of nonexistent objects. But this presupposition cannot be expressed, cannot be 'said,' in Fressellian terms.
We are in the following curious predicament. Something that must be true if if the Fresselian system is to be tenable -- that everything exists, that there are no nonexistent objects -- is not expressible within the system.
3. David Brightly accepts my challenge to give a Frege-Russell translation of 'Something exists.' He writes:
And as a Fressellian I accept the challenge. That property is Individual aka Object, the concept at the root of the Porphyrean tree. We can say 'Something exists' with ∃x.Object(x), ie, there is at least one object. Likewise ∀x.Object(x) (which is always true, even when the box is empty) says 'Everything exists' and its negation (which is always false) says 'Some thing is not an object'. But both these last are unenlightening---because always true and always false, respectively, they convey no information, make no distinction, are powerless to change us.
I asked: which property is it whose instantiation is the existence of something? David's answer is that it is the property or concept Individual or Object. And so I take David to be saying something like the following. "Just as the existence of cats is the being-instantiated of the concept cat, the existence of something is the being-instantiated of the concept Object."
David mentions the tree of Porphyry:
David speaks of the 'root' of the tree where I speak of its apex. No matter. However we visualize it, upside down or right side up, David's suggestion is that Object or Substance (as above) is a summum genus, a supreme genus. It is a concept superordinate to every concept, a concept under which everything falls.
Operating with a scheme like this, we can, in the spirit of Frege's dialogue with the illustrious Puenjer, reduce every existential proposition (or at least every general existential proposition) to a predication by climbing Porphyry's tree. Thus:
Cats exist --> Some mammal is a cat Mammals exist --> Some animal is a mammal Animals exist --> Some living thing is an animal Living things exist --> Some body is a living thing Bodies exist --> Some substance is a body Substances exist --> Some Objects are substances.
The point of these translations is to dispense with 'existst(s)' by showing how propositions of the form Fs exist can be replaced salva veritate with propositions of the form Some G is a F, where G is superordinate to F. This amounts to the elimination of existence in favor of the logical quantity, someness.
We have now climbed to the tippy-top of the tree of Porphyry. We have ascended to a concept superordinate to every concept (except itself) a genus generalissimum, a most general genus. And what concept might that be? Such a concept must have maximal extension and so will have minimal intension. It will be devoid of all content, abstracting as it does from all differences. Frege in his dialog with Puenjer suggests something identical with itself as the maximally superordinate concept. 'There are men' and 'Men exist' thus get rendered as 'Something identical with itself is a man.' (63) Something identical with itself is equivalent to Brightly's Object.
4. Now why can't I accept the Frege-Brightly view? Well, I've already shown that 'Everything exists' cannot be translated as 'Everything is self-identical.' But this is tantamount to having established that the concept whose instantiation is the existence of everything cannot be the concept self-identical something or the concept Object.
Another way to see this is by considering two individuals at the very bottom of the Porphyrean tree. So consider my cats, Max and Manny. In respect of being cats, mammals, beasts, animals, living things, material substances, and self-identical somethings, they do not differ. They do not differ quidditatively. But they do differ: they differ in their very existence. Each has his own existence. Max is not Manny, and Manny is not Max. That is not a mere numerical difference; it is a numerical-existential difference. Since each cat has its own existence, the existence of either cannot be the being-instantiated of any quidditative concept. All such concepts abstract from existence. The same goes for all individuals. Individuals exist. But the existence of individuals is not the being-instantiated of any concept. If you want, you can think of existent (self-identical something) as a highest genus, but Existence -- that in virtue of which things exist and are not nothing -- is not a highest genus. And it is Existence that is the topic. There are no instances of Existence. Existing things are not a kind of thing.
The Frege-Russell theory fails utterly as a theory of Existence.
As sure as I am sitting here, I am sure that I will not convince the Londonistas. That fact is more grist for the (meta)philosophical mill.
2. But can this presupposition be expressed (said) in this logic? Here is a little challenge for you Fressellians: translate 'Something exists' into standard logical notion. You will discover that it cannot be done. Briefly, if existence is instantiation, which property is it whose instantiation is the existence of something? Same problem with 'Nothing exists.' If existence is instantiation, which property is it whose non-instantiation is the nonexistence of anything? Similarly with 'Everthing exists' and 'Something does not exist.'
But couldn't we translate those expressions this way (assuming we have only two properties: a, b)? 1. "something exists" -> "there is an x that instantiates either a or b or ab" 2. "everything exists" -> "there is an x that instantiates a and there is a y that instantiates b and there is a z that instantiates ab" 3. "nothing exists" -> 1 is false 4. "something doesn't exist" -> 2 is false
I am afraid that doesn't work. We need focus only on on 'Some individual exists.' The reader's proposal could be put as follows. Given the properties F-ness and G-ness,
What 'Some individual exists' says is exactly what 'Either F-ness is instantiated or G-ness is instantiated' says.
I would insist however that they do not say the same thing, i.e., do not have the same meaning. The expression on the left says that some individual or other, nature unspecified, exists. The expression on the right, however, makes specific reference to the 'natures' F-ness and G-ness. Surely, 'Some individual exists' could be true even if there are are no individuals that are either Fs or Gs.
Note that it is not a matter of logic what properties there are. This is an extralogical question.
On the Frege-Russell treatment of existence, 'exist(s)' is a second-level predicate, a predicate of concepts, properties, propositional functions and cognate items. It is never an admissible predicate of individuals. Thus in this logic every affirmation of existence must say of some specified concept or property that it is instantiated, and every denial of existence must say of some specified concept or property that it fails of instantiation.
This approach runs into trouble when it comes to the perfectly meaningful and true 'Something exists' and 'Some individual exists.' For in these instances no concept or property can be specified whose instantiation is the existence of things or the existence of individuals. To head off an objection: self-identity won't work.
That there are individuals is a necessary presupposition of the Frege-Russell logic in that without it one cannot validly move from 'F-ness is instantiated' to 'Fs exist.' But it is a necessary presupposition that cannot be stated in the terms of the system. This fact, I believe, is one of the motivations for Wittgenstein's distinction between the sayable and the showable. What cannot be said, e.g., that there are individuals, is shown by the use of such individual variables as 'x.'
The paradox, I take it, is obvious. One cannot say that 'There are individuals' is inexpressible without saying 'There are individuals.' When Wittgenstein assures us that there is the Inexpressible, das Unaussprechliche, he leaves himself open to the retort: What is inexpressible? If he replies, 'That there are individuals,' then he is hoist by his own petard.
Surely it is true that there are individuals and therefore expressible, because just now expressed.
"The suicide of a thesis," says Peter Geach (Logic Matters, p. 265), "might be called Ludwig's self-mate . . . . " Here we may have an instance of it.
Ludwig Wittgenstein sometimes shot his mouth off in summary judgment of men of very high caliber. He once remarked to M. O'C. Drury, "Russell's books should be bound in two colours: those dealing with mathematical logic in red -- and all students of philosophy should read them; those dealing with ethics and politics in blue -- and no one should be allowed to read them." (Recollections of Wittgenstein,* ed. R. Rhees, Oxford 1984, p. 112.)
Here is a passage from Russell's The Conquest of Happiness (Liveright 1930, p. 24) whose urbanity, wit, and superficiality might well have irritated the self-tormenting Wittgenstein:
I do not myself think that there is any superior rationality in being unhappy. The wise man will be as happy as circumstances permit, and if he finds the contemplation of the universe painful beyond a point, he will contemplate something else instead.
*This title is delightfully ambiguous. Read as an objective genitive, it refers to recollections about Wittgenstein, while read as a subjective genitive, it denotes Wittgenstein's recollections. The book, consisting as it does of both, is well-titled.
Bertrand Russell's (1872-1970) The Analysis of Mind first appeared in 1921. Lecture I contains a discussion of Brentano, Meinong, and mental acts. He quotes the famous Brentano passage from the 1874 Psychology From an Empirical Standpoint, and then confesses that until very lately he believed "that mental phenomena have essential reference to objects . . .'" but that he no longer believes this. (p. 5) One of Russell's arguments against acts is contained in the following passage:
. . . the act seems unnecessary and fictious. [. . .] Empirically, I cannot discover anything corresponding to the supposed act; and theoretically I cannot see that that it is indispensable. We say: "I think so-and-so," and this word "I" suggests that thinking is the act of a person. Meinong's "act" is the ghost of the subject, or what once was the full-blooded soul. It is supposed that thoughts cannot just come and go, but need a person to think them. (p. 6)
Russell is making three claims. The first is phenomenological: acts are not given to introspection. The second is dialectical: there are no arguments or considerations that make plausible the positing of acts. The third is genetic: the reason some believe that there are acts is because they have been bamboozled by the surface grammar of sentences like 'I want a unicorn' or 'I see at tree' into the view that when thinking takes place there is an agent who performs an act upon an object.
The Phenomenology of the Situation
What is involved in the awareness of the lamp on my desk? Phenomenologically, as it seems to me, there is awareness of (i) the lamp and of (ii) being aware of the lamp. At a bare minimum, then, we need to distinguish between the object of awareness and the awareness of the object. Both items are phenomenologically accessible. There is straightforward awareness of the lamp if it is seen or imagined or remembered, whereas the awareness of the lamp is given to introspection. Of course, the awareness does not appear alongside the lamp as a separate object. Being aware of the awareness of a lamp is not like being aware of a lamp being next to a clock. And yet, phenomenologically, there is awareness of the lamp and awareness of the awareness of the lamp. Notice that I didn't smuggle in any ego or subject of awareness in my description. So far, then, we are on solid phenomenological ground: there are objects of awareness, there is awareness of objects, and there is awareness that the two are different. This is the phenomenological bare minimum.
But of course this does not show that there are mental acts. For the bit of phenomenology that I have just done is consistent with the subjectlessness of awareness. If awareness is subjectless, as Sartre et al. have maintained against Husserl et al., then it cannot be articulated into individual acts of awareness unless some individuating/differentiating factor can be specified. But there seems to be no phenomenological evidence of such a factor.
Well, let's see. There is awareness of the lamp; there is awareness of the clock; there is awareness of the books piled up on the desk, etc. But awareness appears 'diaphanous,' to borrow a word from G. E. Moore's 1903 "The Refutation of Idealism." The diaphanousness of awareness is a phenomenological feature of it. This being so, there is no phenomenological evidence of any act-articulation on the side of awareness. All the articulation and differentiation appear on the side of the object. But aren't there differences among seeing a lamp, imagining a lamp, and remembering a lamp? No doubt, but why must they be act-differences? It is consistent with the phenomenology of the situation that these differences too fall on the side of the object. Instead of saying that there are acts of imagination and acts of memory, one could say that there are imaginal objects and memorative objects.
The point, then, is that phenomenology alone cannot justify the positing of mental acts. So Russell does have a point with respect to his first claim. Phenomenology needs dialectical supplementation.
The Dialectics of the Situation
Being aware of a centaur and being aware of a mermaid are of course different. This difference is phenomenologically evident. But what differentiates them if there are no mental acts? Not the objects, since they don't exist, and not the awarenesses since they are one and not two on the assumption that there are no mental acts. And if there are no mental acts, then there are no subjects of mental acts. And yet there must be something that accounts for the difference between awareness of a centaur and awareness of a unicorn. The denier of acts seems at this point forced to embrace a Meinongian theory of beingless items. He could say that the centaur-awareness and the mermaid-awareness are numerically different in virtue of the fact that a centaur and a mermaid are distinct denizens of Meinong's realm of Aussersein.
To this I respond that there are no beingless items. The realm of Aussersein is empty. (The arguments cannot be trotted out here.) Hence there is no Meinongian way out. I conclude that we are justified in positing mental acts to account for the difference. I gave this argument already in more detail in my recent reconstruction of an argument from Laird Addis for mental acts.
I conclude that Russell is wrong in his second claim. If the argument I gave is sound, then acts are theoretically indispensable.
Russell's Genetic Claim
This is fairly weak inasmuch as Russell seems not to appreciate the distinction between a mental act and a mental action. An action is the action of an agent who performs the action. But a mental act is merely an occurrent episode of intentional awareness. As such, it needn't be anchored in a substantial self. One could reject substance ontologies as Bergmann does while admitting mental acts. There is nothing in the notion of a mental act that requires that the subject of the act be a substance that exists self-same over time.
The following passage is from Bertrand Russell, A History of Western Philosophy (New York: Simon & Shuster, 1945), p. 427. I found it here, but without a link and without a reference. So, exploiting the resources of my well-stocked library, I located the passage, and verified that it had been properly transcribed. Whether Russell is being entirely fair to the Arabs is a further question.
Arabic philosophy is not important as original thought. Men like Avicenna and Averroes are essentially commentators. Speaking generally, the views of the more scientific philosophers come from Aristotle and the Neoplatonists in logic and metaphysics, from Galen in medicine, from Greek and Indian sources in mathematics and astronomy, and among mystics religious philosophy has also an admixture of old Persian beliefs. Writers in Arabic showed some originality in mathematics and in chemistry; in the latter case, as an incidental result of alchemical researches. Mohammedan civilization in its great days was admirable in the arts and in many technical ways, but it showed no capacity for independent speculation in theoretical matters. Its importance, which must not be underrated, is as a transmitter. Between ancient and modern European civilization, the dark ages intervened. The Mohammedans and the Byzantines, while lacking the intellectual energy required for innovation, preserved the apparatus of civilization, books, and learned leisure. Both stimulated the West when it emerged from barbarism; the Mohammedans chiefly in the thirteenth century, the Byzantines chiefly in the fifteenth. In each case the stimulus produced new thought better than that produced by the transmitters -- in the one case scholasticism, in the other the Renaissance (which however had other causes also).
Many orthodox people speak as though it were the business of sceptics to disprove received dogmas rather than of dogmatists to prove them. This is, of course, a mistake. If I were to suggest that between the Earth and Mars there is a china teapot revolving about the sun in an elliptical orbit, nobody would be able to disprove my assertion provided I were careful to add that the teapot is too small to be revealed even by our most powerful telescopes. But if I were to go on to say that, since my assertion cannot be disproved, it is intolerable presumption on the part of human reason to doubt it, I should rightly be thought to be talking nonsense. If, however, the existence of such a teapot were affirmed in ancient books, taught as the sacred truth every Sunday, and instilled into the minds of children at school, hesitation to believe in its existence would become a mark of eccentricity and entitle the doubter to the attentions of the psychiatrist in an enlightened age or of the Inquisitor in an earlier time.
One thing Russell is doing in this passage is making an unexceptionable point about burden of proof and/or the ad ignorantiam fallacy. If the existence of X has not been disproven, it does not follow that X exists, or even that it is reasonable to believe that X exists. So if anyone were to affirm the existence of something like Russell's celestial teapot or Edward Abbey's angry unicorn on the dark side of the moon, then the onus probandi would be on him to support his outlandish claims. The burden of proof would not rest on those who deny or dismiss such claims.
So far, so good. Russell is of course doing more than underscoring a couple of obvious points in the theory of argumentation. He is applying his points of logic to the God question. Here too I have no complaint. If the existence of God has not been disproven, it does not follow that God exists or even that it is reasonable to believe that God exists.
But the real appeal to atheists and agnostics of the Teapot passage rests on a third move Russell makes. He is clearly suggesting that belief in God (i.e., belief that God exists) is epistemically on a par with believing in a celestial teapot. Just as we have no reason to believe in celestial teapots, irate lunar unicorns (lunicorns?), flying spaghetti monsters, and the like, we have no reason to believe in God. But perhaps we should distinguish between a strong and a weak reading of Russell's suggestion:
S. Just as we cannot have any reason to believe that an empirically undetectable celestial teapot exists, we cannot have any reason to believe that God exists.
W. Just as we do not have any reason to believe that a celestial teapot exists, we do not have any reason to believe that God exists.
Now it seems to me that both (S) and (W) are plainly false: we have all sorts of reasons for believing that God exists. Here Alvin Plantinga sketches about two dozen theistic arguments. Atheists will not find them compelling, of course, but that is irrelevant. The issue is whether a reasoned case can be made for theism, and the answer is in the affirmative. Belief in God and in Russell's teapot are therefore not on a par since there are no empirical or theoretical reasons for believing in his teapot.
Another suggestion embedded in the Russell passage is the notion that if God existed, he would be just another physical thing in the physical universe. But of course this has nothing to do with anything maintained by any sophisticated theist. God is a purely spiritual being.
Another problem with the teapot analogy is that God as traditionally conceived in the West is not an isolani — to use a chess expression. He is not like an isolated pawn, unsupported and unsupporting. For if God exists, then God is the cause of the existence of every contingent being, and indeed, of every being distinct from himself. This is not true of lunar unicorns and celestial teapots. If there is a lunar unicorn, then this is just one more isolated fact about the universe. But if God exists, then everything is unified by this fact: everything has the ground of its being and its intelligibility in the creative activity of this one paradigmatic being.
This is connected with the fact that one can argue from general facts about the universe to the existence of God, but not from such facts to the existence of lunar unicorns and celestial teapots. Thus there are various sorts of cosmological argument that proceed a contingentia mundi to a ground of contingent beings. But there is no similar a posteriori argument to a celestial teapot. There are also arguments from truth, from consciousness, from apparent design, from desire, from morality, and others besides.
The very existence of these arguments shows two things. First, since they move from very general facts (the existence of contingent beings, the existence of truth) to the existence of a source of these general facts, they show that God is not a being among beings, not something in addition to what is ordinarily taken to exist. Second, these arguments give positive reason for believing in the existence of God. Are they compelling? No, but then no argument for any substantive philosophical conclusion is compelling.
People like Russell, Dawkins, and Dennett who compare God to a celestial teapot betray by so doing a failure to understand, and engage, the very sense of the theist's assertions. To sum up. (i) God is not a gratuitous posit in that there are many detailed arguments for the existence of God; (ii) God is not a physical being; (iii) God is not a being who simply exists alongside other beings. In all three respects, God is quite unlike a celestial teapot, a lunar uncorn, an invisible hippopotamus, and suchlike concoctions.
I am quite at a loss to explain why anyone should think the Teapot analogy any good. It leaks like a sieve.