Notes for 1/30/2026

 1/30/2026
Could a perfect designer produce an imperfect design?

[Philosophy Club every Monday, 4-5 pm, in the Buchtel College of Arts and Sciences room 436 ("The Cave")]
[Challenge for today: Try to think of (and possibly ask) at least one question.]
 
Most real-world arguments, even those that are deductive in structure, are really probabilistic. 

This is because the premises in most arguments are derived from experience or testimony.

Example:
1.    All ducks quack.
2.    Daffy is a duck.
3.    Therefore, Daffy quacks.

The 1st claim here is called a “universal generalization.”
Universal generalizations are justified either by definition (conceptual analysis) or by empirical observation (inductive reasoning). The reason to think all ducks quack is that all ducks observed so far quack.

The first-cause argument rests on two premises that appear to be empirical rather than definitional:
Everything that exists must have had a cause.
Nothing can cause itself.

That is, both of these are based on generalizing from past experiences: We have never observed anything coming into existence without a cause or causing itself.

Why this matters: An argument doesn’t have to guarantee its conclusion in order to be reasonable. This applies to some arguments about God.

Elliot Sober’s ‘likelihood principle’:
Observation O supports hypothesis H1 more than it supports hypothesis H2 if and only if O would be more likely given H1 than it would be given H2.

Alternatively (expressed in terms of evidence):
Evidence E supports H1 more than H2 if and only if E increases the likelihood of H1 being true more than it increases the likelihood of H2 being true.

Or, expressed in terms of expectation:
Evidence E supports H1 more than H2 if and only if E would be more expected under H1 than under H2.

 
The standard way of representing comparative probability judgments is as follows:
Pr(H1/O) > Pr(H2/O)
This is read, “The probability of H1 given O is greater than the probability of H2 given O.”

The basic question is what possible explanations for some observation best fit (are most probable given) our observations or evidence.

We make judgments like this all the time. 
Example:
  

O: Eric’s jacket is lying on the floor.
H1: Eric lazily dropped his jacket on the floor.
H2: Eric hung his jacket up, but a tornado swept through the house and blew the jacket off the hook onto the floor.



 
Let’s consider two design arguments employing the likelihood principle:

Let:
Naturalism = Only natural laws and forces operate in the universe.
Theism = There is a supernatural being (God) operating in the universe in addition to natural laws and forces. 

Probabilistic design argument

1.    Living structures exhibit functional complexity (FC).
2.    FC is more expected under theism than under naturalism.
3.    Therefore, probably, theism is true. (God probably exists.)
 

Fine Tuning design argument:

1.    We live in a universe with Life Permitting Conditions (LPC). 
2.    LPC are more expected under theism than under naturalism.
3.    Therefore, probably, theism is true. (God probably exists)



Both of these accept the following comparative probability/likelihood judgment:

The probability/likelihood of FC/LPC is low under naturalism but is not low (or is at least higher) under supernaturalism.
Before evaluating this claim, here is another analysis:

An extremely common type of inductive argument is what I call “Arguments from defeated expectations” (these are basically the same as applications of Sober’s likelihood principle):

Version1:
1. If hypothesis H were true, then we would expect to find evidence E.
2. It is not the case that we find E.
3. Therefore, it is not the case that H.

Example: If my daughter is home, I would expect to find her shoes by the door, but I don’t see them, so she must be out.

Version2:
1. If hypothesis H were true, then we would not expect to find evidence E.
2. But we do find E.
3. Therefore, it is not the case that H.

Example: If Higgens was murdered, we would not expect to find all the doors and windows locked. But they all are, so he wasn’t murdered.

There are also “Arguments from confirmed expectations”:

Version1:
1. If hypothesis H is true, then we would expect to find evidence E.
2. We do find E.
3. Therefore, probably, H is true.

Example: I ask my son to put away the dishes. If he did, I expect to find the dishwasher empty, and I do.

Version2:
1. In most previous cases where we have found E, H has been true.
2. We find E in the present case.
3. Probably, H is true.

Example: In most previous cases where blood spatter has been found on the ceiling, the murder was committed by blunt force. Since there is blood spatter on the ceiling, the murder must have been committed by blunt force.

What do these have to do with design arguments?

In all these cases, the BASIS of the expectations is prior experience.

The likelihood of FC/LPC is low because prior experience with models of naturalistic processes shows that non-FC/non-LPC results of these processes occur with greater frequency than FC/LPC results.

That is, the probability judgments in these cases are (primarily) frequentist or otherwise empirically based.

But this is certainly NOT how judgments regarding the likelihood of FC/LPC under theism, because there are exactly zero cases of known theist causation.

Instead, the likelihood of FC/LPC under theism is Bayesian (probability is a measure of subjective confidence – what seems likely to someone).

But how, exactly, are Bayesian probabilities justified?

Usually, this involves an appeal to pre-existing beliefs and assumptions.