You have more patience than I do: if I was in such a debate and got muted like that, I would refuse to move on to any other point. Which probably isn't the most strategic move, but by gum my temper would not allow anything else. Good on you for having a cool head!
I once spoke with someone vaguely sympathetic to TJump on the matter of the importance of prediction for theories. They were not half as hardcore, so I asked for a paper describing their view on the matter. They sent me a paper by Roger White (03) which defended strong predictionism. In the paper, it said that strong predictionism is unfairly strawmanned as super-strong predictionism, the view that the fact that a theory predicted rather than accommodated a datum is evidence in itself for the theory. He said that “it is not clear that anyone has explicitly endorsed this thesis” (I figure Lakatos did though). After reading that part, I replied to the person who sent it to me that TJump was, in fact, a Super-Duper Strong Predictionist, which is two standard deviations more extreme than the strawman-which-no-one-has-ever-explicitly-endorsed strong predictionism. My interlocutor found that to be quite agreeable. I think the name should stick
Uh, many theists are strong predictionists (including Roger White), so I don’t think strong predictionism gives you justification to reject Theism. I also don’t think strong predictionism is itself very reasonable. But I wouldn’t say you’re a lunatic for being a strong predictionist or even a super strong predictionist. But Super-duper strong predictionism is past the threshold for crazy. So, I don’t know if this answers your question, but super strong < n < super duper strong
I think you are describing super-duper strong predictionism. Strong predictionism does NOT say that a hypothesis having predicted the data is direct evidence for the hypothesis: it is only indirect evidence. Super-strong predictionism says that it is direct evidence.
Neither give ANY reason to believe that the best hypothesis is going to be one which makes predictions. It only says that predictions can be some additional evidence for a hypothesis.
Yeah, pretty much. There are also plenty of cases where accommodation is vastly superior to prediction, such as when it comes to routing out data fraud. Obviously Alice isn’t doing as good of a job defending the FTA as someone like Jason Waller or Robin Collins, but if no other theory reasonably accommodates this data (which Waller argues that a multiverse has a good shot at), then it’s the best game in town. Few philosophers of science take seriously the idea that a theory which fully explains the data is in any way inferior to a theory which partially explains the data, but predicted it
Probably 1? Most Christians I know (myself included) have experienced things that are predicted by Christianity. Spiritual experiences, answers to prayer, the Christian life being a satisfying one, etc.
The great mathematical prodigy Srinivasa Ramanujan credited his discoveries to the goddess Namagiri revealing them to him in dreams. So I think we can safely lay atheism to rest at this point.
Someday, I am going to discover the true theory of everything, but I will claim that all of the laws of physics were caused to be true, and everything in the universe caused to exist, by a square circle in my pocket (they were created in a puff of logic via the principle of explosion). I will do it all just to mess with TJump.
Also, given that many early physicists were theists and incorporated God into their theories (e.g., believing that God created the natural laws or set in motion the initial conditions of the universe), isn't TJump obligated to believe in God under his epistemology?
It provides the opportunity to achieve synthesis between "only evidence based on predictions count" and "predictions don't matter for evidence". Both of these views are somewhat true in their appropriate context, but also crazy if taken outside of it.
If I tell you that I'm wearing socks right now, then it's some evidence that I indeed wear socks. Nevertheless, this lackluster standard of evidence is completely inappropriate for complex theories about the world. Not because probability theory works differently when talking about socks compared to talking about laws of the universe, but because humans are biased reasoners and fail to apply it correctly, smuggling in wishful thinking, counting the same evidence multiple times or even mistaking the direction the evidence is pointing to. And the more complex the matter, the more opportunities is to make mistakes and, therefore, the higher standard of evidence is required, to correct for such biases.
How do you test for the existence of a thing without specifying its properties? Properties that can be tested for. So that you can locate the thing in the real world via detection of a thing with those properties. So you specify a property if a god as perfection. Show me how you test for that.
We know that humans exist. Julius Ceasars is a human. But we don't know that gods exist even though there are in general things that exist. None of the things that we know exist are Gods or leprechauns. The analogy to Julius Ceasar fails.
Were there any novel testable predictions made on the assumption that dinosaurs once existed?
Did our discoveries about dinosaurs lead to any accurate predictions?
I'm genuinely curious. I don't recall reading something that said "because we learned about dinosaurs, we were able to accurately predict X", but maybe something like that is out there? If it is out there, I'd love to hear about it. If knowing about dinosaurs never enabled testable accurate predictions, that strikes me as a pretty conclusive counterargument to T Jump's epistemology.
You can do it epistemically by refusing to learn any modern biology, focusing on learning all about dinosaurs first, and then using dinosaurology to predict what current animals should look like, what genes they should have, ancestral paths, etc.
> Tom claimed in defense of it that it’s widely accepted in the sciences, but this isn’t true—science is methodologically naturalist in that it tends to only consider natural explanations of events, but scientists don’t have some well-worked out epistemology for examining supernatural explanations. It simply isn’t their field of expertise.
They do have a well-worked out epistemology for examining supernatural explanations, it's called science. They use this to rule out supernatural explanations because they are all extraordinarily bad at predicting or explaining anything. If scientists doing archaeology and paleontology had dug up the animals from Noah's ark and located the Garden of Eden, instead of what they actually did, this would be extremely strong scientific evidence for Judaism, just as digging up Pompeii and Roman-Egyptian tombs are very strong scientific evidence for the Roman Empire. If a double blind study on the power of prayer found that prayers specifically from Christians reduced cancer fatalities by 9%, this would be quite strong evidence for Christianity. Those things didn't happen, though, because those religions aren't true.
Sure, scientists can trivially rule out supernatural explanations that don’t fit the data. But that’s a bit different from what he’s discussing. You could conceivably have cases where a supernatural explanation seems to predict the evidence better than any natural explanation - for example, a miracle claim with tons of corroborating evidence. In those cases, you need to know something about the prior probability of a supernatural explanation to decide whether to accept it or not, given however much evidence there is. The standard methods of science don’t deal with the prior probabilities of supernatural explanations. So you need to do philosophy to get a handle on those cases.
> Second, as explained before, it would rule out believing in dark matter. Dark matter is a new kind of thing, not composed of existing stuff. It’s believed in because it explains existing data, not because it made advanced experimental predictions.
I think this is only true if you don't count new astronomical observations as "advanced experimental predictions", even though we didn't know what the results would be before taking the observation. There have certainly been new observations supporting dark matter theories taken after the formulation of the theory, in some cases explicitly taken to test the theory.
Particle dark matter could also interact with normal matter via the weak force (maybe the strong force), and might be detectable in Higgs boson style experiments.
You’re making a somewhat strong philosophical claim about the existence of dark matter. Is it consistent with the evidence? Sure. Would it be surprising if physicists found evidence against it? Not really. It’s definitely not as well supported as many other scientific theories.
But it’s not good to the same degree that natural selection or quantum mechanics is good. The main thing going for it that it explains inconsistencies in what they expected to find and the fact that it’s impervious to any measurement is questionable. I’m certainly not an expert but “undetectable particles that make our theories work” is the kind of thing you would expect if there was something fundamentally missing.
It's definitely not as well supported as other theories in physics, but the fact that it's considered very likely (even if not certain) means that you can use explanation of existing data to support the existence of a new kind of thing not made of previously known stuff. That's all that's needed to refute TJump's claim that you can't use that kind of argument at all.
Attempting to do things that one thinks has a chance of interacting with God is instrumentally rational, under Pascal's wager. Prayer plausibly falls into this category.
>"For all we know, our reality could've been utterly different, and our physics could've have been truly exotic/alien. Maybe it wouldn't even have close analogs of things like matter, particles, energy, force, etc."
Conceding this helps the fine-tuning argument. If the world could have been very different, but we are in a world that seems unusually well suited for us, then that would be evidence towards theories that account for that. You might argue that that's not conclusive, but it is evidence.
>"Ya, but what I'm saying is that space, time, matter, energy, etc, aren't reality "in and of itself". They are just elements of model we impose upon reality."
>"Perhaps even in our untweaked reality there are aliens that perceive the world in such a radically different way that their physics has utterly nothing familiar to us"
The models, though, seem uncommonly good at predicting reality, and much of it seems pretty fundamental to how reality works. At least, as best we know, the world behaves remarkably similar to how it would if those things existed. Now, I don't know quite what "exist" means, or how to identify what reality is in itself beyond what we can gather from empirical and rational data. Any future model will have to have the current models as limiting cases. The concepts are powerful and so we use them; it seems extremely unlikely. These tools are useful because they can collapse incredibly complex systems into a straightforward set of a few important factors. It seems like the importance of those factors is not arbitrary—there's no way that there will be anything that will represent the world with the same clarity without being able to talk about space, or light, or whatever.
>"But that doesn't mean that if reality were tweaked then reality would fall apart. Perhaps the tweaked reality would get along just fine, and the inhabitants would come up with models that didn't have anything like space, time, matter, energy, etc."
I read this as the assertion that life is common in the parameter space of physics, with randomly chosen values. Do you have reason to think this, when in many cases, under current models, we have reason to think that it wouldn't allow for complex lifelike structures to arise? I suppose the fine-tuning argument here is one to that effect, but then that's a concession that the argument works, just not that it works sufficiently to demonstrate that God exists. Which is fine, of course, just be aware of that.
You have more patience than I do: if I was in such a debate and got muted like that, I would refuse to move on to any other point. Which probably isn't the most strategic move, but by gum my temper would not allow anything else. Good on you for having a cool head!
I once spoke with someone vaguely sympathetic to TJump on the matter of the importance of prediction for theories. They were not half as hardcore, so I asked for a paper describing their view on the matter. They sent me a paper by Roger White (03) which defended strong predictionism. In the paper, it said that strong predictionism is unfairly strawmanned as super-strong predictionism, the view that the fact that a theory predicted rather than accommodated a datum is evidence in itself for the theory. He said that “it is not clear that anyone has explicitly endorsed this thesis” (I figure Lakatos did though). After reading that part, I replied to the person who sent it to me that TJump was, in fact, a Super-Duper Strong Predictionist, which is two standard deviations more extreme than the strawman-which-no-one-has-ever-explicitly-endorsed strong predictionism. My interlocutor found that to be quite agreeable. I think the name should stick
Uh, many theists are strong predictionists (including Roger White), so I don’t think strong predictionism gives you justification to reject Theism. I also don’t think strong predictionism is itself very reasonable. But I wouldn’t say you’re a lunatic for being a strong predictionist or even a super strong predictionist. But Super-duper strong predictionism is past the threshold for crazy. So, I don’t know if this answers your question, but super strong < n < super duper strong
I think you are describing super-duper strong predictionism. Strong predictionism does NOT say that a hypothesis having predicted the data is direct evidence for the hypothesis: it is only indirect evidence. Super-strong predictionism says that it is direct evidence.
Neither give ANY reason to believe that the best hypothesis is going to be one which makes predictions. It only says that predictions can be some additional evidence for a hypothesis.
Yeah, pretty much. There are also plenty of cases where accommodation is vastly superior to prediction, such as when it comes to routing out data fraud. Obviously Alice isn’t doing as good of a job defending the FTA as someone like Jason Waller or Robin Collins, but if no other theory reasonably accommodates this data (which Waller argues that a multiverse has a good shot at), then it’s the best game in town. Few philosophers of science take seriously the idea that a theory which fully explains the data is in any way inferior to a theory which partially explains the data, but predicted it
Probably 1? Most Christians I know (myself included) have experienced things that are predicted by Christianity. Spiritual experiences, answers to prayer, the Christian life being a satisfying one, etc.
I knew that he was a bit of a cliché internet atheist, but I didn't think it was that bad.
Muting you is particularly ridiculous.
The great mathematical prodigy Srinivasa Ramanujan credited his discoveries to the goddess Namagiri revealing them to him in dreams. So I think we can safely lay atheism to rest at this point.
Hardly means she's real. I've seen it written that everyone you encounter in dreams are aspects of yourself.
Sorry, I don't make the rules of epistemology.
Someday, I am going to discover the true theory of everything, but I will claim that all of the laws of physics were caused to be true, and everything in the universe caused to exist, by a square circle in my pocket (they were created in a puff of logic via the principle of explosion). I will do it all just to mess with TJump.
Also, given that many early physicists were theists and incorporated God into their theories (e.g., believing that God created the natural laws or set in motion the initial conditions of the universe), isn't TJump obligated to believe in God under his epistemology?
Sounds like some of the sophistry seen at UncommonDissent in the early oughts. I’m sure I contributed some myself.
Good times.
I'm leaving this link that may be helpful for those who want to clear their confusion about evidence and importance of predictions: https://www.lesswrong.com/s/zpCiuR4T343j9WkcK/p/fhojYBGGiYAFcryHZ
Do you think anything there conflicts with what I said?
In a matter of speaking.
It provides the opportunity to achieve synthesis between "only evidence based on predictions count" and "predictions don't matter for evidence". Both of these views are somewhat true in their appropriate context, but also crazy if taken outside of it.
If I tell you that I'm wearing socks right now, then it's some evidence that I indeed wear socks. Nevertheless, this lackluster standard of evidence is completely inappropriate for complex theories about the world. Not because probability theory works differently when talking about socks compared to talking about laws of the universe, but because humans are biased reasoners and fail to apply it correctly, smuggling in wishful thinking, counting the same evidence multiple times or even mistaking the direction the evidence is pointing to. And the more complex the matter, the more opportunities is to make mistakes and, therefore, the higher standard of evidence is required, to correct for such biases.
But I didn't say predictions don't matter for evidence.
How do you test for the existence of a thing without specifying its properties? Properties that can be tested for. So that you can locate the thing in the real world via detection of a thing with those properties. So you specify a property if a god as perfection. Show me how you test for that.
We know that humans exist. Julius Ceasars is a human. But we don't know that gods exist even though there are in general things that exist. None of the things that we know exist are Gods or leprechauns. The analogy to Julius Ceasar fails.
Were there any novel testable predictions made on the assumption that dinosaurs once existed?
Did our discoveries about dinosaurs lead to any accurate predictions?
I'm genuinely curious. I don't recall reading something that said "because we learned about dinosaurs, we were able to accurately predict X", but maybe something like that is out there? If it is out there, I'd love to hear about it. If knowing about dinosaurs never enabled testable accurate predictions, that strikes me as a pretty conclusive counterargument to T Jump's epistemology.
You can do it epistemically by refusing to learn any modern biology, focusing on learning all about dinosaurs first, and then using dinosaurology to predict what current animals should look like, what genes they should have, ancestral paths, etc.
> Tom claimed in defense of it that it’s widely accepted in the sciences, but this isn’t true—science is methodologically naturalist in that it tends to only consider natural explanations of events, but scientists don’t have some well-worked out epistemology for examining supernatural explanations. It simply isn’t their field of expertise.
They do have a well-worked out epistemology for examining supernatural explanations, it's called science. They use this to rule out supernatural explanations because they are all extraordinarily bad at predicting or explaining anything. If scientists doing archaeology and paleontology had dug up the animals from Noah's ark and located the Garden of Eden, instead of what they actually did, this would be extremely strong scientific evidence for Judaism, just as digging up Pompeii and Roman-Egyptian tombs are very strong scientific evidence for the Roman Empire. If a double blind study on the power of prayer found that prayers specifically from Christians reduced cancer fatalities by 9%, this would be quite strong evidence for Christianity. Those things didn't happen, though, because those religions aren't true.
Sure, scientists can trivially rule out supernatural explanations that don’t fit the data. But that’s a bit different from what he’s discussing. You could conceivably have cases where a supernatural explanation seems to predict the evidence better than any natural explanation - for example, a miracle claim with tons of corroborating evidence. In those cases, you need to know something about the prior probability of a supernatural explanation to decide whether to accept it or not, given however much evidence there is. The standard methods of science don’t deal with the prior probabilities of supernatural explanations. So you need to do philosophy to get a handle on those cases.
> Second, as explained before, it would rule out believing in dark matter. Dark matter is a new kind of thing, not composed of existing stuff. It’s believed in because it explains existing data, not because it made advanced experimental predictions.
I think this is only true if you don't count new astronomical observations as "advanced experimental predictions", even though we didn't know what the results would be before taking the observation. There have certainly been new observations supporting dark matter theories taken after the formulation of the theory, in some cases explicitly taken to test the theory.
Particle dark matter could also interact with normal matter via the weak force (maybe the strong force), and might be detectable in Higgs boson style experiments.
You’re making a somewhat strong philosophical claim about the existence of dark matter. Is it consistent with the evidence? Sure. Would it be surprising if physicists found evidence against it? Not really. It’s definitely not as well supported as many other scientific theories.
I think the evidence is good which is why it's widely accepted.
But it’s not good to the same degree that natural selection or quantum mechanics is good. The main thing going for it that it explains inconsistencies in what they expected to find and the fact that it’s impervious to any measurement is questionable. I’m certainly not an expert but “undetectable particles that make our theories work” is the kind of thing you would expect if there was something fundamentally missing.
That's true. But it's still a helpful example to illustrate the principled point.
It's definitely not as well supported as other theories in physics, but the fact that it's considered very likely (even if not certain) means that you can use explanation of existing data to support the existence of a new kind of thing not made of previously known stuff. That's all that's needed to refute TJump's claim that you can't use that kind of argument at all.
Attempting to do things that one thinks has a chance of interacting with God is instrumentally rational, under Pascal's wager. Prayer plausibly falls into this category.
>"For all we know, our reality could've been utterly different, and our physics could've have been truly exotic/alien. Maybe it wouldn't even have close analogs of things like matter, particles, energy, force, etc."
Conceding this helps the fine-tuning argument. If the world could have been very different, but we are in a world that seems unusually well suited for us, then that would be evidence towards theories that account for that. You might argue that that's not conclusive, but it is evidence.
>"Ya, but what I'm saying is that space, time, matter, energy, etc, aren't reality "in and of itself". They are just elements of model we impose upon reality."
>"Perhaps even in our untweaked reality there are aliens that perceive the world in such a radically different way that their physics has utterly nothing familiar to us"
The models, though, seem uncommonly good at predicting reality, and much of it seems pretty fundamental to how reality works. At least, as best we know, the world behaves remarkably similar to how it would if those things existed. Now, I don't know quite what "exist" means, or how to identify what reality is in itself beyond what we can gather from empirical and rational data. Any future model will have to have the current models as limiting cases. The concepts are powerful and so we use them; it seems extremely unlikely. These tools are useful because they can collapse incredibly complex systems into a straightforward set of a few important factors. It seems like the importance of those factors is not arbitrary—there's no way that there will be anything that will represent the world with the same clarity without being able to talk about space, or light, or whatever.
>"But that doesn't mean that if reality were tweaked then reality would fall apart. Perhaps the tweaked reality would get along just fine, and the inhabitants would come up with models that didn't have anything like space, time, matter, energy, etc."
I read this as the assertion that life is common in the parameter space of physics, with randomly chosen values. Do you have reason to think this, when in many cases, under current models, we have reason to think that it wouldn't allow for complex lifelike structures to arise? I suppose the fine-tuning argument here is one to that effect, but then that's a concession that the argument works, just not that it works sufficiently to demonstrate that God exists. Which is fine, of course, just be aware of that.