Taking a type-token razor to Stannard’s God
I’ve been listening to the lectures in Open University’s Philosophy and the Human Situation podcast course series. They are pretty good. I particularly enjoyed Janet Radcliffe Richards’ “Taking Stock” talk and Ronald Hepburn’s talk “Wonder”.
While on the Tube, I listened to Russell Stannard’s talk “The New Argument from Design”. It presents the standard anthropic principle/fine tuning argument, but concludes at the end that you make the leap using faith alone from accepting the conclusions of a valid fine tuning argument to the God of faith. This is something I consider a major problem with many to-God-from-contingency arguments. However metaphysically raw and physical one goes just means one has a longer bridge to get to the eventual intended conclusion: fallin’ down on one’s knees and praisin’ the Lord before givin’ one’s soul to Baby Jesus!
Let us hear Stannard’s argument in his own words. From about 15m30s onwards: The alternative to many universes or the many different domains of the one universe is to accept that there is just the one universe - the laws of physics are the same everywhere in it - and that it’s a put up job. It was deliberately designed for life. And the designed is God. I, and others, have argued that this is the simplest hypothesis. Just one designer God rather than an infinite number of different kinds of universe or different domains of this universe. One of the criticisms of this is although this suggestion is simplest in the sense that it calls for one unknown rather than an infinite number of unknowns, it is not the simplest suggestion in the sense that a God is an entirely different kind of concept to the physical ones. We know that there are physical concepts so it is a relatively simple extension of that idea to postulate a wider variety of the sorts of things that are known to exist. God, on the other hand, is an entirely new and unverified concept - so it is claimed.
Stannard’s response to this? I can see the strength of that argument, at least how I would imagine it appearing to an atheist - someone for whom God is an unknown. That objection completely fails for someone like myself who already believes in God on other grounds. You see, I don’t see how anyone could ever be argued into belief in God.
Stannard then goes on to say that it is religious experience that is required to go from “creative power” to the God of faith.
What is wrong with this argument? I don’t think it gives the objection enough of a fair chance. The objection in this case is a purely Ockhamist one. What makes a multiverse (with ‘evolving universes’) theory attractive - beyond any empirical support that physicists may put forward - is that it reduces the number of types by dramatically increasing the number of instantiated tokens for the universe type. In ontology, we are concerned with what fundamental types exist. If we accept Ockham’s razor and a type-token distinction, then introducing a new type ought to be the costly thing, not instantiating new tokens for explanations.
At least, that’s my opinion and it’s held not particularly strongly. We must see a type-token distinction, but if we are to value either the types or the tokens to such a degree as to make them the swing vote on any particular metaphysical construction, we must have a few reasons why either reducing tokens is important at the risk of increasing types, or vice versa. And if one is even more serious about such divine accountancy, we must have some reasonable procedure by which we assign weight to each token and type posited. I think also that such an opinion of the comparative value of types and tokens can’t be solved with reference to either the types or the tokens themselves. That way lies great potential for special pleading (we see that with Stannard: it’s only because it’s God that one can make the intuitive appeal - if it was just “X”, one couldn’t say “Ah, but only for anti-X-ists is it mysterious - for pro-X-ists, X isn’t strange or unknown at all.”) and even question-begging. How does one resolve such a problem? A first guess might be to say that if we can’t use the nature of the types and tokens posited, we could use some kind of holistic analysis of the system as a whole.
That is how we do so in an area of practical reasoning - computer programming. When one uses high-level computer programming languages, one constructs tokens of various types. How one orders those types internally in the memory and processor of the computer is a difficult question. In the vast majority of high-level programming languages today, there is a distinction between a primitive type and a compound type. In Java, for instance, numbers and bytes are primitives rather than objects. We have a ‘bare’ number 6, but if you wish to represent something more complex like a database query, one has a Query object. Java, then, has at least eight fundamental things: byte, short, int, long, float, double, boolean and char. On top of that, we may add classes, objects, methods and some other stuff. On the other hand, Ruby doesn’t have these primitive data types: numbers (short, int, long, float), booleans and bytes/chars are instantiated as full-blown objects. Now, we ask, what benefits and what costs are there associated with the increase in simplicity? The answers are obvious for programmers: as these things are classes and instantiated objects, one can build upon them with ease: subclasses can be constructed, and it inherits methods from Object. For the programmer, whatever one can do with another class, they can do with these classes. There are costs though: primarily performance related. If you construct a boolean as a primitive, it takes up less memory and requires less processing than a boolean object that instantiates one of the two boolean classes. This is not necessarily a problem: a clever compiler or interpreter could see whether or not in a particular case the extra baggage of an object is required. Exactly what gets fed to the processor or the virtual machine need not necessarily concern the programmer. The point of this example is to give an example of how the kind of reasoning we are interested in is done. In computing, something like ontology is done very practically. Expressiveness, performance, extensibility and safety (with computers in cars, missiles, spaceships, aeroplanes, life-support systems and a whole load of other critical military and civilian systems, poor code means much more than lost sales or pissed off employees: it means dead bodies - see here and here).
Let’s look at another analogous situation in the physical world rather than the computational one: properties. It is an argument on it’s own for a nominalist account of properties, or indeed an account of properties that makes use of tropes (Stoutian particulars, abstract particulars etc. - they are not short of names!), that they reduce the number of types one needs to presume in one’s ontology.
I find Stannard’s response to that thought to be unsatisfying. Familiarity with a type has nothing to do with the Ockhamist objection to that type. To use a slightly absurd example: I am far more intuitively familiar with turtles than I am with quarks. I could give you a description that would help you identify with a reasonable degree of reliability whether or not something is a turtle or not - living, comparatively slow moving, with a hard shell, of such a size, found in certain placs etc. I’m sure a biologist could give you a description in both terms of species and genetics. But, I cannot, off the top of my head, explain to you what a quark is. But if someone presents to me a physical ontology, that quarks and similar entities would have a better chance of being more fundamental than turtles seems intuitively obvious. To switch it slightly, you can have atoms without turtles, but you can’t have turtles without atoms. If someone were to say that turtles have a fundamental ontological role (as one might say they would in the Discword novels), no amount of familiarity validates that.
To go back to our slightly more sensible analogy with properties: imagine if we had a defender of predicate nominalism stating, on Stannard-style grounds, that unlike the realist, he didn’t need to add this extra bloat to his ontology, but only acknowledge predicates. (And, well, no sensible person denies that predicates exist, right?) If you find predicate nominalism appealing, familiarity with predicates doesn’t make universals significantly less strange. And if you find predicate nominalism wanting, you do so in spite of the fact that predicates are familiar for you.
That’s not to say that the number of tokens is never relevant. I think nominalism is a perfect example: if one is a predicate nominalist, the problem is how do you construct a world where the number of predicates roughly matches the number of universals that a realist would want. One always bumps into the problem of arbitrariness. With a realist account, let’s say we’ve got ourselves an apple, and it’s red and juicy. The realist says “Okay, apple, you instantiate Appleness, Redness and Juicyness”. One particular, three universals (if we grant all three of them status as universals). The predicate nominalist says “No, no, no. We’ve got an object to which three predicates can be applied - “is an apple”, “is red” and “is juicy”. But says the realist, the number of properties can be multipled. The object can also satisfy “is a red apple”, “is a juicy apple”, “is a red, juicy thing” and even “is red but not furry”, “is red but not a chest of drawers”, “is not a packet of crisps” and so on. To satisfy the realist, the predicate nominalist must find a way of adequately slicing these back down to the bare, essential predicates or bite the bullet of there being an almost infinite number of predicates applicable. If we allow “is red and is red and is red” type predicates, well, the predicate nominalist is even more screwed. D.M. Armstrong in the chapter on ‘particularism’ (that is trope theory) Universals and Scientific Realism reckons there’s a similarly big problem for defenders of tropes.
How might one go about responding to the multiversist without getting into the complex physics of it - that is, what possible philosophical objections could we raise? Well, if I wanted to beat up multiverses, I’d go for probing away firstly at the scientific status of such a theory. Comparatively, that’s not that much of a problem. Perhaps I’d then move towards wondering about whether multiverses hold up conceptually. The idea certainly seems a strange one. I mean, the point about a universe is that it is everything. The idea that there could be an alternative everything is strange because surely one could just say “Ah, but there’s an everything-everything that contains the two different everythings”. Semantics, perhaps, but there might be something a bit spooky hidden behind the multiverse mask. You could even wonder about the way in which infinity is deployed here. One would then want to wonder about whether or not one can quantify over this infinite set of universes, and how one sorts such universes. To say there is a particular subset that match in some way might get you into counterpart/Twin Earth problem. Basically, you then get to riff on the sort of probelms that might face a modal realist like Lewis.
Familiarity with God is, I think, obviously a bad way to respond to Ockhamist objections. If we plan on resolving multiverse-vs.-God debates like Stannard has presented, we need to understand how we weigh up infinite tokens one type vs. two types with one token each. I’m not sure Ockham gets us that far.
