Theist here. The difference between the God of the Gaps argument and any other scientific theory that attempts to fill in gaps is that the scientific theories are inherently fallible. The scientists advancing scientific theories will change their theories as new conflicting evidence comes to light. Theists who invoke God as an explanation for evolution/the creation of the universe/whatever else don't typically allow for fallibilism, so the God of the Gaps retort is valid. The God of the gaps charge is not just a general objection to using God; it is a general objection to not relying on a fallibilist epistemology, which is a valid objection and the foundational key to scientific inquiry.
How do you use the God Hypothesis to generate observations? You have to know some things about God and what it wants/how it behaves.
Atheists say “If God exists, then it would save innocents from harm. Innocents are harmed all the time, therefore the hypothesis generates the wrong observations.”
Theists often respond, “God is wiser than you, so it is folly to predict anything about it.”
This is basically admitting to the God Hypothesis being unusable, and an unusable hypothesis gets Occam’s Razored.
The “Dogs make pee” hypothesis is usable. It predicts that sometimes you find pee on the floor when dogs are around. If you come home and find a puddle of crude oil or custard, then you won’t say “Dog kidneys work in mysterious ways, it is folly to assume I know how they work, therefore the dog did it.”
More abstractly, the challenge is to make a God Hypothesis that is a “shorter description” for our observations than the observations themselves. The baseline worst “theory” would just be a list of the observations themselves.
Observations: [“Physical events and psychic events match.”, “The universal constants allow for life.” …]
God Hypothesis: [“What God wants happens.”, “God wants” + “, and”.join(Observations)]
The God Hypothesis is strictly longer, and therefore worse.
This unneeded increase in complexity is what the God of the Gaps is trying to indicate.
The God hypothesis posits more stuff but simplicity is about the fundamental stuff. Simplicity doesn't give you a reason to, for instance, think that nebula disappear when we don't see them because then there would be less stuff.
That’s not the measure of simplicity. Nebulas disappearing would require an extra nebula-disappearing-law, making it more complicated.
If there is some metric by which this universe maximizes that value, I would be interested to hear it! Calling it “perfection” is probably unhelpful because of all of the baggage that brings (and the clear counter evidence that must be hand waved away.)
This universe already obeys the Principle of Least Action, so we already have an analogue. A theory that can prove that the physical constants maximize some value would be a huge breakthrough.
The reason it's not simple to posit disappearing nebulas is that even though there'd be less stuff there would be more *fundamental* stuff.
The fundamental thing, on theism, is a single being of unlimited goodness, which is a simple hypothesis.
Solomonoff induction being the right way to calculate priors is this weird LessWrong belief with no reason supporting it that they trot out as established fact because Eliezer wrote smugly about its correctness. Why in the world would we think that it's the right way to calculate priors?
3. How do you change the weighting in response to new evidence?
Partial answers:
2. Occam’s razor: But what does it mean for one hypothesis to be simpler than another?
3. Bayes theorem: But how do you know how likely the evidence is according to the hypothesis?
Solomonoff Induction answers these questions in a way I consider satisfying and intuitive. It gives me a set of standards I’ve looked at, reflected upon, and decided “yes, this looks like a good guide to avoiding becoming completely confused when it comes to weighing hypotheses.”
If you have some arguments for rejecting Solomonoff Induction, or if you’ll play around in the Solomonoff Induction Microtheory, I look forward to what you have to say (you might even support theism with it?!). But until then, I think I’ll skip your theology posts.
3. By applying bayes theorem. You use your best guess, relying on intuition.
If the claim is just that Solomonoff induction matches your intuitions, then that's fine, but it doesn't match mine (I don't think you can compute qualities like perfection or possible worlds). And I think your intuitions are corrupted by spending time around rationalists who get their priors based on solomonoff induction.
Intuition isn’t magic. It has some structure and function we would do well to understand. And intuition can fail. Just as we use our minds to identify and avoid illusions and biases identified by our minds, we ought to use improve our intuition, using our intuitive notions as a starting place.
I’m unimpressed by your unwillingness to formalize “explain” or “simplicity”. These are important concepts, and in order to avoid confusion we ought to ground them.
How can intuition get corrupted? By what metric?
Please point me to an essay explaining how Solomonoff Induction fails for perfection and possible worlds. In interested in learning more.
I don't know how much of John Lennox you have encountered, but his work seems to be tailor-made to the point you are making here. He explains that "God of the gaps" presupposes a theory of God that most theists don't hold to: that God exists ONLY to explain gaps in our knowledge. Rather, God is the agent explanation, while science provides the mechanical explanations; thus the two are not inherently in conflict.
I would argue that the god of the gaps "fallacy" or whatever you would call it isn't specificly about using god to fill a gap, but the way how nicely god works to specifically fill gaps. When someone infers the existence of genetic code to explain the poor shape of ape thumbs, the gap-filling "genetic code" eventually unwraps into the equivalent of millions of pages of scientific reasoning with practical implications over the centuries, whereas "God is a good explanation for the fine fine-tuning of the universe" almost never gets expanded into more than that very sentence.
God's liquid-like conformity to the shape of the explanatory gap helps to explain why the ocean he filled yesterday has shrunk to a puddle today without inflicting much damage to the concept. If the same speaker who contends God is needed to explain X, is willing to revise when a scientific explanation is found and say well no, but maybe they're needed to explain this other phenomenon Y, and we iterate this process several dozen times, then we might be concerned they are taking some liberties with the concepts of "God" and "need". If the gap is supposed to stand as evidence for God in itself, rather than God's ability to patch it being a pleasant side effect of his otherwise manifest nature, then wouldn't we like some constraints on "the kinds of gaps a God-shaped concept can and should be expected to plug"? Otherwise we can confidently claim that he'll be stuffed into every available explanatory chasm and the assertion that "we need God to fill /this/ particular gap," insofar as it reflects an ability to hazard other explanations, has little value. And if it's claimed that no constraints can be put on such a concept, then no weight can be given to what is advanced for evidence.
It doesn’t need to. Intent to create life, ability to create life (necessary power to create it) and necessary knowledge to create it is all that is necessary. And God with his classical attributes is (and has traditionally been) shorthand for these attributes.
Except, this doesn't have any explanatory value if your goal is to understand anything about these concepts, because the explanation of God rules out fundamental understanding per definition. If your goal is to prove the existence of God, sure, that's good, but I would rather like to learn something about ape-thumbs or fine-tuning of the universe itself. To me personally, "god is the reason" is absolutely no improvement over "I don't know". If explanations don't lead to agency down the road, what's the point? I can accept lack of agency by itself, but I'm not going to actively welcome it by entertaining this mode of thinking.
Because of how range the god hypothesis answers questions for (from moral knowledge to things like abiogenesis), it seems very hard to create an established credence for what actually makes it more or less probable. It seems not to decrease people’s credence as much if you explain a phenomena without god than if our lack of scientific or philosophical knowledge leads to an unexplained phenomena that people claim god is the cause of. I think an underratedly important question to be asking in these debates is, under naturalism, how much would we expect not to be able to understand given our progress.
The point of the "God of the gaps" objection is that you can (historically and synchronically) predict that theists will invoke their deity for any profound question you as a nonbeliever express serious uncertainty about. Other explanations don't tend to work this way at all. If I tell an economics professor that I find it mysterious how my cousin recovered from cancer after a terminal diagnosis, she's not going to say, "See, invisible hand of the market!"
The problem with the argument from incredulity is not so much with the incredulity itself (skepticism is usually considered a good thing in rationalist and atheist circles), but with drawing positive conclusions based on the incredulity, and with the false binaries it often sets up.
“I don’t see how X can be true, therefore X is false.” No, even if you reject evidence that X is true, that only gets you to X is unknown. It makes it no more likely that X is false without additional evidence. Evidence such as chocolate all over Timmy’s face, or familiarity with basic physics.
But theists often, implicitly or explicitly, take it one step further: “I don’t see how X can be true. Therefore X is false. Therefore Y is true.”
“I don’t believe evolution can be true considering the platypus…”
Interesting. Tell me more.
“Therefore my belief that new animal species spontaneously come into being when you sacrifice a goat as the full moon rises due East of the secret tomb of Odin is justified.”
What? No, that doesn’t follow at all.
Even if you positively disprove evolution (as opposed to simply not being convinced by the evidence in favor of it), that tells us virtually nothing about what actually is true.
Theist here. The difference between the God of the Gaps argument and any other scientific theory that attempts to fill in gaps is that the scientific theories are inherently fallible. The scientists advancing scientific theories will change their theories as new conflicting evidence comes to light. Theists who invoke God as an explanation for evolution/the creation of the universe/whatever else don't typically allow for fallibilism, so the God of the Gaps retort is valid. The God of the gaps charge is not just a general objection to using God; it is a general objection to not relying on a fallibilist epistemology, which is a valid objection and the foundational key to scientific inquiry.
How do you use the God Hypothesis to generate observations? You have to know some things about God and what it wants/how it behaves.
Atheists say “If God exists, then it would save innocents from harm. Innocents are harmed all the time, therefore the hypothesis generates the wrong observations.”
Theists often respond, “God is wiser than you, so it is folly to predict anything about it.”
This is basically admitting to the God Hypothesis being unusable, and an unusable hypothesis gets Occam’s Razored.
The “Dogs make pee” hypothesis is usable. It predicts that sometimes you find pee on the floor when dogs are around. If you come home and find a puddle of crude oil or custard, then you won’t say “Dog kidneys work in mysterious ways, it is folly to assume I know how they work, therefore the dog did it.”
More abstractly, the challenge is to make a God Hypothesis that is a “shorter description” for our observations than the observations themselves. The baseline worst “theory” would just be a list of the observations themselves.
Observations: [“Physical events and psychic events match.”, “The universal constants allow for life.” …]
God Hypothesis: [“What God wants happens.”, “God wants” + “, and”.join(Observations)]
The God Hypothesis is strictly longer, and therefore worse.
This unneeded increase in complexity is what the God of the Gaps is trying to indicate.
You reason about what a perfect being would do.
The God hypothesis posits more stuff but simplicity is about the fundamental stuff. Simplicity doesn't give you a reason to, for instance, think that nebula disappear when we don't see them because then there would be less stuff.
That’s not the measure of simplicity. Nebulas disappearing would require an extra nebula-disappearing-law, making it more complicated.
If there is some metric by which this universe maximizes that value, I would be interested to hear it! Calling it “perfection” is probably unhelpful because of all of the baggage that brings (and the clear counter evidence that must be hand waved away.)
This universe already obeys the Principle of Least Action, so we already have an analogue. A theory that can prove that the physical constants maximize some value would be a huge breakthrough.
Dude. Look into Solomonoff Induction.
The reason it's not simple to posit disappearing nebulas is that even though there'd be less stuff there would be more *fundamental* stuff.
The fundamental thing, on theism, is a single being of unlimited goodness, which is a simple hypothesis.
Solomonoff induction being the right way to calculate priors is this weird LessWrong belief with no reason supporting it that they trot out as established fact because Eliezer wrote smugly about its correctness. Why in the world would we think that it's the right way to calculate priors?
Some questions:
1. What is a hypothesis?
2. How do you weigh how a priori likely it is?
3. How do you change the weighting in response to new evidence?
Partial answers:
2. Occam’s razor: But what does it mean for one hypothesis to be simpler than another?
3. Bayes theorem: But how do you know how likely the evidence is according to the hypothesis?
Solomonoff Induction answers these questions in a way I consider satisfying and intuitive. It gives me a set of standards I’ve looked at, reflected upon, and decided “yes, this looks like a good guide to avoiding becoming completely confused when it comes to weighing hypotheses.”
If you have some arguments for rejecting Solomonoff Induction, or if you’ll play around in the Solomonoff Induction Microtheory, I look forward to what you have to say (you might even support theism with it?!). But until then, I think I’ll skip your theology posts.
1. A view that might explain things.
2. You look at virtues like simplicity.
3. By applying bayes theorem. You use your best guess, relying on intuition.
If the claim is just that Solomonoff induction matches your intuitions, then that's fine, but it doesn't match mine (I don't think you can compute qualities like perfection or possible worlds). And I think your intuitions are corrupted by spending time around rationalists who get their priors based on solomonoff induction.
Intuition isn’t magic. It has some structure and function we would do well to understand. And intuition can fail. Just as we use our minds to identify and avoid illusions and biases identified by our minds, we ought to use improve our intuition, using our intuitive notions as a starting place.
I’m unimpressed by your unwillingness to formalize “explain” or “simplicity”. These are important concepts, and in order to avoid confusion we ought to ground them.
How can intuition get corrupted? By what metric?
Please point me to an essay explaining how Solomonoff Induction fails for perfection and possible worlds. In interested in learning more.
I don't know how much of John Lennox you have encountered, but his work seems to be tailor-made to the point you are making here. He explains that "God of the gaps" presupposes a theory of God that most theists don't hold to: that God exists ONLY to explain gaps in our knowledge. Rather, God is the agent explanation, while science provides the mechanical explanations; thus the two are not inherently in conflict.
Aristoteles wins again! Mechanistic (material causes) and Purpose (teleological causes) are not incompatible!
I would argue that the god of the gaps "fallacy" or whatever you would call it isn't specificly about using god to fill a gap, but the way how nicely god works to specifically fill gaps. When someone infers the existence of genetic code to explain the poor shape of ape thumbs, the gap-filling "genetic code" eventually unwraps into the equivalent of millions of pages of scientific reasoning with practical implications over the centuries, whereas "God is a good explanation for the fine fine-tuning of the universe" almost never gets expanded into more than that very sentence.
God's liquid-like conformity to the shape of the explanatory gap helps to explain why the ocean he filled yesterday has shrunk to a puddle today without inflicting much damage to the concept. If the same speaker who contends God is needed to explain X, is willing to revise when a scientific explanation is found and say well no, but maybe they're needed to explain this other phenomenon Y, and we iterate this process several dozen times, then we might be concerned they are taking some liberties with the concepts of "God" and "need". If the gap is supposed to stand as evidence for God in itself, rather than God's ability to patch it being a pleasant side effect of his otherwise manifest nature, then wouldn't we like some constraints on "the kinds of gaps a God-shaped concept can and should be expected to plug"? Otherwise we can confidently claim that he'll be stuffed into every available explanatory chasm and the assertion that "we need God to fill /this/ particular gap," insofar as it reflects an ability to hazard other explanations, has little value. And if it's claimed that no constraints can be put on such a concept, then no weight can be given to what is advanced for evidence.
It doesn’t need to. Intent to create life, ability to create life (necessary power to create it) and necessary knowledge to create it is all that is necessary. And God with his classical attributes is (and has traditionally been) shorthand for these attributes.
Except, this doesn't have any explanatory value if your goal is to understand anything about these concepts, because the explanation of God rules out fundamental understanding per definition. If your goal is to prove the existence of God, sure, that's good, but I would rather like to learn something about ape-thumbs or fine-tuning of the universe itself. To me personally, "god is the reason" is absolutely no improvement over "I don't know". If explanations don't lead to agency down the road, what's the point? I can accept lack of agency by itself, but I'm not going to actively welcome it by entertaining this mode of thinking.
Because of how range the god hypothesis answers questions for (from moral knowledge to things like abiogenesis), it seems very hard to create an established credence for what actually makes it more or less probable. It seems not to decrease people’s credence as much if you explain a phenomena without god than if our lack of scientific or philosophical knowledge leads to an unexplained phenomena that people claim god is the cause of. I think an underratedly important question to be asking in these debates is, under naturalism, how much would we expect not to be able to understand given our progress.
The point of the "God of the gaps" objection is that you can (historically and synchronically) predict that theists will invoke their deity for any profound question you as a nonbeliever express serious uncertainty about. Other explanations don't tend to work this way at all. If I tell an economics professor that I find it mysterious how my cousin recovered from cancer after a terminal diagnosis, she's not going to say, "See, invisible hand of the market!"
The problem with the argument from incredulity is not so much with the incredulity itself (skepticism is usually considered a good thing in rationalist and atheist circles), but with drawing positive conclusions based on the incredulity, and with the false binaries it often sets up.
“I don’t see how X can be true, therefore X is false.” No, even if you reject evidence that X is true, that only gets you to X is unknown. It makes it no more likely that X is false without additional evidence. Evidence such as chocolate all over Timmy’s face, or familiarity with basic physics.
But theists often, implicitly or explicitly, take it one step further: “I don’t see how X can be true. Therefore X is false. Therefore Y is true.”
“I don’t believe evolution can be true considering the platypus…”
Interesting. Tell me more.
“Therefore my belief that new animal species spontaneously come into being when you sacrifice a goat as the full moon rises due East of the secret tomb of Odin is justified.”
What? No, that doesn’t follow at all.
Even if you positively disprove evolution (as opposed to simply not being convinced by the evidence in favor of it), that tells us virtually nothing about what actually is true.
Great post Matthew! I enjoyed it a lot because I see this arguments all the time.