Reductionism is wrong only in cases where it doesn’t reduce far enough, or when computation limits one’s ability to work the very large number of factors that make up our macro experiences. For these reasons, some amount of higher-level modeling is needed, to gloss over the underlying math.
This does NOT mean that the big thing is anything other than the sum of small things. It just means there are too many small things to compute over, so it’s easier, and gives good-enough results to treat them as a big thing.
Your examples are not of reductionist failing, but of a poor choice of modeling level. Evolutionary explanations that include inclusive genetic fitness and population ecology is already non-reductionist explanation. Evolutionary explanations that start with individual cell biology, random combination and mutation, and 10^15 or so specific deaths are probably reductionist enough for most purposes.
At least this tells me I didn’t make a silly mistake in my post. Thank you for the feedback.
As for your objections,
All models are wrong, some models are useful.
exactly captures my conceit. Reductionism is correct in the sense that is, in some sense, closer to reality than anti- or contra-reductionism. Likely in a similar sense that machine code is closer to the reality of a physical computation than a .cpp file, though the analogy isn’t exact, for reasons that should become clear.
I’m typing this on a laptop, which is a intricate amalgam of various kinds of atoms. Hypothetically, you could explain the positioning of the atoms in terms of dense quantum mechanical computations (or a more accurate physical theory, which would exist ex hypothesi), and/or we could explain it in terms of economics, computer science and the vagaries of my life. The former strictly contains more information than the latter, and subsumes the latter to the extend that it represents reality and contradicts it to the extend it’s misleading.
At an objective level, then, the strictly reductionist theory wins on merit.
Reductionism functions neatly to explain reality-in-general, and even to explain certain orderly systems that submit to a reductionist analysis. If you want completeness, reductionism will give you completeness, at the limit. But sometimes, a simple explanation is nice. It’d be convenient to compress, to explain evolution in abstract terms.
The compression will be lossy, because we don’t actually have access to reality’s dataset. But lossy data is okay, and more okay to more casual the ends. Pop science books are very lossy, and are sufficient for delivering a certain type of entertainment. A full reprinting of a paper’s collected data is about as lossless as we tend to get.
A lossless explanation is reductionist, and centribus paribus, we ought to go with the reductionist explanation. Given a choice between a less lossy, very complex explanation and a lossy, but simple explanation, you should probably go gather more data. But failing that, you should go with one that suits your purposes. A job where every significant bit digit of accuracy matters chooses the first, as an example.
There are two things you could mean when you say ‘reductionism is right’. That reality is reductionist in the “big thing = small thing + small thing” sense, or that reductionist explanations are better by fiat.
Reality is probably reductionist. I won’t assign perfect certainty, but reductionist reality is simpler than magical reality.
As it currently stands, we don’t have a complete theory of reality, so the only criteria we can judge theories is that they 1) are accurate, 2) are simple.
I am not arguing about the rightness or wrongness of reductionism. Reductionism and contra-reductionism are containers, and they contain certain classes of explanations. Contra-reductionism conatins historical explanations, explaining the state of things by the interactions with outside forces, and reductionism contains predictive explanations, explaining the future behavior in terms of internal forces.
All models are wrong, some models are useful.
Reductionism is wrong only in cases where it doesn’t reduce far enough, or when computation limits one’s ability to work the very large number of factors that make up our macro experiences. For these reasons, some amount of higher-level modeling is needed, to gloss over the underlying math.
This does NOT mean that the big thing is anything other than the sum of small things. It just means there are too many small things to compute over, so it’s easier, and gives good-enough results to treat them as a big thing.
Your examples are not of reductionist failing, but of a poor choice of modeling level. Evolutionary explanations that include inclusive genetic fitness and population ecology is already non-reductionist explanation. Evolutionary explanations that start with individual cell biology, random combination and mutation, and 10^15 or so specific deaths are probably reductionist enough for most purposes.
At least this tells me I didn’t make a silly mistake in my post. Thank you for the feedback.
As for your objections,
exactly captures my conceit. Reductionism is correct in the sense that is, in some sense, closer to reality than anti- or contra-reductionism. Likely in a similar sense that machine code is closer to the reality of a physical computation than a .cpp file, though the analogy isn’t exact, for reasons that should become clear.
I’m typing this on a laptop, which is a intricate amalgam of various kinds of atoms. Hypothetically, you could explain the positioning of the atoms in terms of dense quantum mechanical computations (or a more accurate physical theory, which would exist ex hypothesi), and/or we could explain it in terms of economics, computer science and the vagaries of my life. The former strictly contains more information than the latter, and subsumes the latter to the extend that it represents reality and contradicts it to the extend it’s misleading.
At an objective level, then, the strictly reductionist theory wins on merit.
Reductionism functions neatly to explain reality-in-general, and even to explain certain orderly systems that submit to a reductionist analysis. If you want completeness, reductionism will give you completeness, at the limit. But sometimes, a simple explanation is nice. It’d be convenient to compress, to explain evolution in abstract terms.
The compression will be lossy, because we don’t actually have access to reality’s dataset. But lossy data is okay, and more okay to more casual the ends. Pop science books are very lossy, and are sufficient for delivering a certain type of entertainment. A full reprinting of a paper’s collected data is about as lossless as we tend to get.
A lossless explanation is reductionist, and centribus paribus, we ought to go with the reductionist explanation. Given a choice between a less lossy, very complex explanation and a lossy, but simple explanation, you should probably go gather more data. But failing that, you should go with one that suits your purposes. A job where every significant bit digit of accuracy matters chooses the first, as an example.
Isn’t that what people mean when they say reductionism is right?
There are two things you could mean when you say ‘reductionism is right’. That reality is reductionist in the “big thing = small thing + small thing” sense, or that reductionist explanations are better by fiat.
Reality is probably reductionist. I won’t assign perfect certainty, but reductionist reality is simpler than magical reality.
As it currently stands, we don’t have a complete theory of reality, so the only criteria we can judge theories is that they 1) are accurate, 2) are simple.
I am not arguing about the rightness or wrongness of reductionism. Reductionism and contra-reductionism are containers, and they contain certain classes of explanations. Contra-reductionism conatins historical explanations, explaining the state of things by the interactions with outside forces, and reductionism contains predictive explanations, explaining the future behavior in terms of internal forces.