Lacking a word for lack
October 1, 2007 12:18 PM Subscribe
Is there a scientific name for the information gap between an event and the traces that we have of this event?
For example: the gap between the information accessible in a fossil and the sum of all biological information relative to the original; or the gap between the particles' tracks detected in a cyclotron and all the information relative to the real event; in fact, I realize that it may apply to all kind of events (the information gap between a photography and an event), but I am guessing the right word has more chances to be useful in science.
For example: the gap between the information accessible in a fossil and the sum of all biological information relative to the original; or the gap between the particles' tracks detected in a cyclotron and all the information relative to the real event; in fact, I realize that it may apply to all kind of events (the information gap between a photography and an event), but I am guessing the right word has more chances to be useful in science.
Yeah, I think that Pierce (and semiotics in general) would be helpful here. Google for the "semiotic triangle" (here's a good example). Signs, such as photographs and the traces from a particle physics experiment, are such that they stand in a particular relationship to the things that they represent.
To answer your particular question, however, I don't know of a good word for each of the legs of the triangle. I've seen some with labels, such as the one I'm showing, but the idea of a "lack" is not quite fitting with semiotics as I understand it. A "lack" makes it sound as if there is some perfect knowledge (the denominator) and a given representation captures some portion of it (the numerator). But that's not most philosophers of science think of scientific truths these days.
Note: I'm not a semiotician or philosopher of science. However, I've used some of these theories in my work (in computer science / HCI).
posted by zpousman at 1:30 PM on October 1, 2007
To answer your particular question, however, I don't know of a good word for each of the legs of the triangle. I've seen some with labels, such as the one I'm showing, but the idea of a "lack" is not quite fitting with semiotics as I understand it. A "lack" makes it sound as if there is some perfect knowledge (the denominator) and a given representation captures some portion of it (the numerator). But that's not most philosophers of science think of scientific truths these days.
Note: I'm not a semiotician or philosopher of science. However, I've used some of these theories in my work (in computer science / HCI).
posted by zpousman at 1:30 PM on October 1, 2007
Response by poster: zpousman: A "lack" makes it sound as if there is some perfect knowledge (the denominator) and a given representation captures some portion of it (the numerator).
Sure. That's why I restricted the total information about a fossil to "biological". I understand that "behavioral" or "social" knowledge are types of information that can't be accounted for in a fossil.
But there are measurable gaps: fragments of Neanderthal DNA vs complete Neanderthal DNA, for example. Or fragments of Lucy's skeleton vs complete Lucy's skeleton. This is a measurable quantity.
Or the difference between the information collected from coelacanths fossils vs coelacanths discovered alive in the wild. Previously, scientists knew only a specific percentage of coelacanths' bilology. Now they can know it all.
posted by bru at 2:31 PM on October 1, 2007
Sure. That's why I restricted the total information about a fossil to "biological". I understand that "behavioral" or "social" knowledge are types of information that can't be accounted for in a fossil.
But there are measurable gaps: fragments of Neanderthal DNA vs complete Neanderthal DNA, for example. Or fragments of Lucy's skeleton vs complete Lucy's skeleton. This is a measurable quantity.
Or the difference between the information collected from coelacanths fossils vs coelacanths discovered alive in the wild. Previously, scientists knew only a specific percentage of coelacanths' bilology. Now they can know it all.
posted by bru at 2:31 PM on October 1, 2007
This is kind of a version of the problem of induction. (one among many related philosophy of science/epistemological issues) Roughly: Our observations of the world will always be incomplete; how can we have confidence when we generalize from them?
posted by LobsterMitten at 2:50 PM on October 1, 2007
posted by LobsterMitten at 2:50 PM on October 1, 2007
Response by poster: I appreciate your answers but I am not looking for theoretical or philosophical terms.
Painquale, from your second link: "(Undetermination) is only a problem for those who suppose that there is nothing more to science and scientific theories than an appeal to data."
I understand that. And it is not what I am looking for, but it helps to narrow my question. I am looking for a name for the gap of missing data in a dataset. The number of bones in a skeleton is a finite quantity, as is a the number of base pairs in a complete genome.
posted by bru at 4:57 PM on October 1, 2007
Painquale, from your second link: "(Undetermination) is only a problem for those who suppose that there is nothing more to science and scientific theories than an appeal to data."
I understand that. And it is not what I am looking for, but it helps to narrow my question. I am looking for a name for the gap of missing data in a dataset. The number of bones in a skeleton is a finite quantity, as is a the number of base pairs in a complete genome.
posted by bru at 4:57 PM on October 1, 2007
bru, are you thinking of this gap as something that can be filled by collecting enough data, and you want a name for the extra data that we still have to collect?
Or are you thinking of it more as: there will always be more "information" about a given object than we can know?
posted by LobsterMitten at 5:05 PM on October 1, 2007
Or are you thinking of it more as: there will always be more "information" about a given object than we can know?
posted by LobsterMitten at 5:05 PM on October 1, 2007
Also , you say I am not looking for theoretical or philosophical terms, but I don't see why.
Philosophy of science and epistemology study questions exactly like this. (Exactly like both of the ones that I mention above.) These are fields that would have the term you're looking for.
posted by LobsterMitten at 5:09 PM on October 1, 2007
Philosophy of science and epistemology study questions exactly like this. (Exactly like both of the ones that I mention above.) These are fields that would have the term you're looking for.
posted by LobsterMitten at 5:09 PM on October 1, 2007
Some people have already touched on this, but all science is about creating a model of the universe that explains our observations. If a model makes sense and makes useful predictions, we keep it until something better comes along. But this is, to some extent, like describing how a clock works by seeing the hands. The question "what really happened" is answered if we have a good model that everybody agrees on, but this is just by convention so we don't have to admit "we don't know" every time. See the following examples to see how conventions change.
I have this cloud chamber photograph--what really happened?
A: I don't know; lets call that particle "Strange"
B: You're looking at a strong force interaction; the exchange particle is a muon
C: You're looking at a strong force interaction; it's all about quarks and gluons.
D: The standard model is incompatible with any theory of gravity so far; I'm going to explain this with quarks and gluons, but we really don't know what's going on.
I saw the sun rise--what's really going on?
A: Apollo in his chariot drives the sun around the earth every day.
B: The earth actually goes around the sun in a circle
C: No, actually it's an ellipse.
D: Gravity does it: the same force that makes apples fall keeps the planets in orbit. There's a gravitational field, and an inverse-square law. Problem: Action at a distance is "spooky".
E: We're still calling it gravity, but it's not a force; matter or energy affects the curvature of space-time.
F: If you have a lot of mass in a really small space you have to use quantum mechanics too, and we don't know what's "really" going on with gravity.
Any of these explanations would have been generally accepted as "the truth" at various times in history, where the other explanations might have caused people to doubt your sanity. We improve the models to explain new things, but as for what's really going on, we can only guess. Maybe angels push the planets around the sun; we just don't use this model because we can't test it or make predictions with it.
If you want to know what's really going on, particle physics is weird. Quantum mechanics says that electrons are only there when we observe them. When we're not observing them, we say there are probability waves that interfere with each other and describe where we are likely to observe these electrons. As well as those probability waves describe things, you want the electrons to be "really" doing something? Tough luck! Bell's Theorem gives you 2 options:
1: There's not "really" anything happening, and the magical act of measuring something causes the wave functions to collapse and the electrons to appear.
2: Something even weirder! You can choose a non-local theory (requires faster than light communication and abandons traditional notions of cause and effect, in which case why have science at all), or you can choose the Many-Worlds option (favorite of all the "alternate universe" sci-fi dramas).
And that's just quantum! Thermal Physics will tell you that all electrons are indistinguishible. They're not like marbles where you can paint one red and the other green; they're all fundamentally the same. Yup, that means that an electron in a star and another electron in my right hand could instantaneously (faster than light) trade places, and we couldn't tell. "But that probably doesn't happen--it's unphysical." "So, realistically, I can tell the difference between an electron in a different star and one in my hand." "Gee, I'd really feel better telling you that, but if they're not completely indistinguishable, then all the math breaks down and we can't explain Bose-Einstein condensates, or anything that uses Fermi-Dirac statistics (like most of electronics)."
In summary: The universe is a really weird place, and there's no word for the gap between observation and reality because we don't know what reality is. The word that takes you from an observation to a scientific model is "Hypothesis," "Theory," or "Law," depending on how much you respect the model you're using. Only observations are real; the rest is useful, but ultimately just our best guess at explaining the universe.
posted by rossmik at 5:30 PM on October 1, 2007 [1 favorite]
I have this cloud chamber photograph--what really happened?
A: I don't know; lets call that particle "Strange"
B: You're looking at a strong force interaction; the exchange particle is a muon
C: You're looking at a strong force interaction; it's all about quarks and gluons.
D: The standard model is incompatible with any theory of gravity so far; I'm going to explain this with quarks and gluons, but we really don't know what's going on.
I saw the sun rise--what's really going on?
A: Apollo in his chariot drives the sun around the earth every day.
B: The earth actually goes around the sun in a circle
C: No, actually it's an ellipse.
D: Gravity does it: the same force that makes apples fall keeps the planets in orbit. There's a gravitational field, and an inverse-square law. Problem: Action at a distance is "spooky".
E: We're still calling it gravity, but it's not a force; matter or energy affects the curvature of space-time.
F: If you have a lot of mass in a really small space you have to use quantum mechanics too, and we don't know what's "really" going on with gravity.
Any of these explanations would have been generally accepted as "the truth" at various times in history, where the other explanations might have caused people to doubt your sanity. We improve the models to explain new things, but as for what's really going on, we can only guess. Maybe angels push the planets around the sun; we just don't use this model because we can't test it or make predictions with it.
If you want to know what's really going on, particle physics is weird. Quantum mechanics says that electrons are only there when we observe them. When we're not observing them, we say there are probability waves that interfere with each other and describe where we are likely to observe these electrons. As well as those probability waves describe things, you want the electrons to be "really" doing something? Tough luck! Bell's Theorem gives you 2 options:
1: There's not "really" anything happening, and the magical act of measuring something causes the wave functions to collapse and the electrons to appear.
2: Something even weirder! You can choose a non-local theory (requires faster than light communication and abandons traditional notions of cause and effect, in which case why have science at all), or you can choose the Many-Worlds option (favorite of all the "alternate universe" sci-fi dramas).
And that's just quantum! Thermal Physics will tell you that all electrons are indistinguishible. They're not like marbles where you can paint one red and the other green; they're all fundamentally the same. Yup, that means that an electron in a star and another electron in my right hand could instantaneously (faster than light) trade places, and we couldn't tell. "But that probably doesn't happen--it's unphysical." "So, realistically, I can tell the difference between an electron in a different star and one in my hand." "Gee, I'd really feel better telling you that, but if they're not completely indistinguishable, then all the math breaks down and we can't explain Bose-Einstein condensates, or anything that uses Fermi-Dirac statistics (like most of electronics)."
In summary: The universe is a really weird place, and there's no word for the gap between observation and reality because we don't know what reality is. The word that takes you from an observation to a scientific model is "Hypothesis," "Theory," or "Law," depending on how much you respect the model you're using. Only observations are real; the rest is useful, but ultimately just our best guess at explaining the universe.
posted by rossmik at 5:30 PM on October 1, 2007 [1 favorite]
Hm, maybe those links weren't so good. The links I included make it sound like underdetermination necessarily balloons outward into indeterminacy, relativity, or skepticism, such that underdetermination a purely philosophical term that describes a purely philosophical problem. You don't need to think of underdetermination that way. If you know that there are a finite number of base pairs in a genome and you know that the number is between N and M, it's still perfectly acceptable to say that the number of base pairs is underdetermined by the evidence. So it does sound like it describes the phenomenon that you're interested in.
But it sounds like you're not asking for a name for the phenomenon; you're asking for a name for the missing information, or maybe a name for the hole in our knowledge that needs to be filled in by the missing information (note that these are different - I'm not sure which of these you want). I don't think there's a common scientific term for either of these.
posted by painquale at 5:37 PM on October 1, 2007
But it sounds like you're not asking for a name for the phenomenon; you're asking for a name for the missing information, or maybe a name for the hole in our knowledge that needs to be filled in by the missing information (note that these are different - I'm not sure which of these you want). I don't think there's a common scientific term for either of these.
posted by painquale at 5:37 PM on October 1, 2007
Yes, you can't tell sociological information from a fossil, but you can't even tell a lot of biological information, which isn't "missing" any more than the sociology. For example, what color was their skin? Was there one species of dinosaur with this skeleton, or were there 2 non-interbreeding species with different skin and internal organs, but similar skeletons?
I understand that. And it is not what I am looking for, but it helps to narrow my question. I am looking for a name for the gap of missing data in a dataset. The number of bones in a skeleton is a finite quantity, as is a the number of base pairs in a complete genome.
"Stuff we didn't/couldn't/would like to/are going to/can't be bothered to measure?"
Seriously, even these questions can't be answered well. The number of bones in a dinosaur skeleton is finite, but perhaps not known. If you say "we're missing one," you're sticking to one model of reality. But if you find enough skeletons missing the same bone, you might want to propose a new model of reality: that there were two species of dinosaurs with similar skeletons.
Similarly, if you're talking about the human genome, exactly how many base pairs are there? Even if you're talking about my particular genome, the number of base pairs may differ from cell to cell; why should DNA be perfectly transcribed and unmutated every time? Even if you're talking about the DNA from one cell, can we even make repeatable measurements on that, or does looking at that DNA mutate it? (it must require some interesting chemistry or some high energy microscopy.) We're pretty sure we know how many genes there are in the human genome, but we seem to be finding that "one gene, one protein" isn't true, and that the "junk" dna actually does stuff too.
"A name for the gap of missing data in a dataset":
If you are missing a few points on a graph, you might call those "gaps in the data." If you have a graph of the middle of something, but not the edges, you might call that "an insufficient range of measurement." If you have a graph of z=f(x,y) versus x and you held y constant you might call that "a slice of the data" or an iso-whatever curve; if you didn't hold y constant and it seemed to have a large effect, you might call it "useless."
And this comes back to the original problem; you don't know what you should be looking for to fill in the gaps in your data. Different scientific models suggest different sorts of gaps. Underdetermination means you don't have enough data to know which hypothesis is best; I'm going to go a step further and say that even if you know which hypothesis is best, you still can't know what hypothesis is true.
We moved from Newton's law of Gravity to Einstein's General Relativity because it was a better model; it explained everything Newton did, it reduces to Newtons laws in simple cases, it explained things Newton didn't such as the precession of Mercury's orbit, and it predicted things that were later observed like gravitational lensing. And even though the math to do anything with Relativity gets pretty hairy, the law itself is perfect simplicity: Energy (including mass) equals curvature. So it's a better model, but is it what's really going on? We really suspect it isn't, so much so that if you suggest a better model than GR, you'll probably get a Nobel Prize. And even then, is that what's really going on? Who knows.
posted by rossmik at 6:29 PM on October 1, 2007
I understand that. And it is not what I am looking for, but it helps to narrow my question. I am looking for a name for the gap of missing data in a dataset. The number of bones in a skeleton is a finite quantity, as is a the number of base pairs in a complete genome.
"Stuff we didn't/couldn't/would like to/are going to/can't be bothered to measure?"
Seriously, even these questions can't be answered well. The number of bones in a dinosaur skeleton is finite, but perhaps not known. If you say "we're missing one," you're sticking to one model of reality. But if you find enough skeletons missing the same bone, you might want to propose a new model of reality: that there were two species of dinosaurs with similar skeletons.
Similarly, if you're talking about the human genome, exactly how many base pairs are there? Even if you're talking about my particular genome, the number of base pairs may differ from cell to cell; why should DNA be perfectly transcribed and unmutated every time? Even if you're talking about the DNA from one cell, can we even make repeatable measurements on that, or does looking at that DNA mutate it? (it must require some interesting chemistry or some high energy microscopy.) We're pretty sure we know how many genes there are in the human genome, but we seem to be finding that "one gene, one protein" isn't true, and that the "junk" dna actually does stuff too.
"A name for the gap of missing data in a dataset":
If you are missing a few points on a graph, you might call those "gaps in the data." If you have a graph of the middle of something, but not the edges, you might call that "an insufficient range of measurement." If you have a graph of z=f(x,y) versus x and you held y constant you might call that "a slice of the data" or an iso-whatever curve; if you didn't hold y constant and it seemed to have a large effect, you might call it "useless."
And this comes back to the original problem; you don't know what you should be looking for to fill in the gaps in your data. Different scientific models suggest different sorts of gaps. Underdetermination means you don't have enough data to know which hypothesis is best; I'm going to go a step further and say that even if you know which hypothesis is best, you still can't know what hypothesis is true.
We moved from Newton's law of Gravity to Einstein's General Relativity because it was a better model; it explained everything Newton did, it reduces to Newtons laws in simple cases, it explained things Newton didn't such as the precession of Mercury's orbit, and it predicted things that were later observed like gravitational lensing. And even though the math to do anything with Relativity gets pretty hairy, the law itself is perfect simplicity: Energy (including mass) equals curvature. So it's a better model, but is it what's really going on? We really suspect it isn't, so much so that if you suggest a better model than GR, you'll probably get a Nobel Prize. And even then, is that what's really going on? Who knows.
posted by rossmik at 6:29 PM on October 1, 2007
I believe you're looking for a data palimpsest.
This is not a scientific name. I just made it up. But I think it's groovy as hell.
posted by YamwotIam at 7:09 PM on October 1, 2007
This is not a scientific name. I just made it up. But I think it's groovy as hell.
posted by YamwotIam at 7:09 PM on October 1, 2007
Okay, so I think there are two sides to what you are asking for..
[more, it would seem. -ed]
First, models can never be equivalent to reality, no matter how precise (ya, we'll get an existential debate here, sorry about that..). There is also mathematical underdetermination, as mentioned above. Observability characterizes whether a system is mathematically underdetermined or not.
I guess there could be a couple of other characteristics.. Information that was once available, but has been lost, and information that we don't have a way of measuring yet, but may one day be measurable.
Singularity. A singularity can obliterate information, making it inobservable. The gimbal lock singularity, for example. When two axes are aligned, you simply can't measure all degrees of freedom, the third direction has been locked out. But, just getting close to the singularity severely impinges your ability to measure some degrees of freedom.
Ya, those are self links.
posted by Chuckles at 7:12 PM on October 1, 2007
[more, it would seem. -ed]
First, models can never be equivalent to reality, no matter how precise (ya, we'll get an existential debate here, sorry about that..). There is also mathematical underdetermination, as mentioned above. Observability characterizes whether a system is mathematically underdetermined or not.
I guess there could be a couple of other characteristics.. Information that was once available, but has been lost, and information that we don't have a way of measuring yet, but may one day be measurable.
Singularity. A singularity can obliterate information, making it inobservable. The gimbal lock singularity, for example. When two axes are aligned, you simply can't measure all degrees of freedom, the third direction has been locked out. But, just getting close to the singularity severely impinges your ability to measure some degrees of freedom.
Ya, those are self links.
posted by Chuckles at 7:12 PM on October 1, 2007
Response by poster: LobsterMitten: are you thinking of this gap as something that can be filled by collecting enough data, and you want a name for the extra data that we still have to collect?
Exactly.
Philosophy of science and epistemology study questions exactly like this.
Point taken.
rossmik: love your overview of the state of our knowledge. You don't really answer my question but you touch all the bases around it in a beautiful way. I wish you were my physics teacher. I flagged your comment as "fantastic", because it is.
painquale: sorry, I read too fast and saw "un-determination" instead of "underdetermination". Now I get it, thanks.
you're asking for (...) a name for the hole in our knowledge that needs to be filled in by the missing information
Yes.
rossmik again: now you wander into Sisyphus territory. Even if we can't reach absolute knowledge, there is something like "manageable" knowledge, like the coelacanth story: maybe the fossil record gave us about 20% of the biology of coelacanths whereas live coelacanths allow us to reach 99%, even if the remaining 1% stays out of reach. So the gap has been closed by 79% (not an absolute value: a ballpark one, useful to have a measure of our knowledge in a specific domain).
YamwotIam: data palimpsest. Love it.
Chuckles: Information that was once available, but has been lost.
I like that. It covers the Neanderthal DNA question, for example. It makes me realize that rossmik's last point is about observable knowledge vs absolute and/or future knowledge, whereas my question is more about "if a given observation capacity could be applied retroactively." Hence the palimpsest. It's a kind of hypothetical memory gap.
This is another example where the quest is more enjoyable than the goal. Thank you all for guiding me into thinking better.
posted by bru at 8:06 PM on October 1, 2007
Exactly.
Philosophy of science and epistemology study questions exactly like this.
Point taken.
rossmik: love your overview of the state of our knowledge. You don't really answer my question but you touch all the bases around it in a beautiful way. I wish you were my physics teacher. I flagged your comment as "fantastic", because it is.
painquale: sorry, I read too fast and saw "un-determination" instead of "underdetermination". Now I get it, thanks.
you're asking for (...) a name for the hole in our knowledge that needs to be filled in by the missing information
Yes.
rossmik again: now you wander into Sisyphus territory. Even if we can't reach absolute knowledge, there is something like "manageable" knowledge, like the coelacanth story: maybe the fossil record gave us about 20% of the biology of coelacanths whereas live coelacanths allow us to reach 99%, even if the remaining 1% stays out of reach. So the gap has been closed by 79% (not an absolute value: a ballpark one, useful to have a measure of our knowledge in a specific domain).
YamwotIam: data palimpsest. Love it.
Chuckles: Information that was once available, but has been lost.
I like that. It covers the Neanderthal DNA question, for example. It makes me realize that rossmik's last point is about observable knowledge vs absolute and/or future knowledge, whereas my question is more about "if a given observation capacity could be applied retroactively." Hence the palimpsest. It's a kind of hypothetical memory gap.
This is another example where the quest is more enjoyable than the goal. Thank you all for guiding me into thinking better.
posted by bru at 8:06 PM on October 1, 2007
You know, there might be a reason that there isn't a word for what you describe. I remember speaking to a philosopher of science once who told me that the common conception of what scientists do is take measurements to catalogue unknown properties of things. But although scientists do that, it's a very small part of what they do. Usually, they're making experiments to test competing theories, and when they do perform measurements, it's in areas that look "promising" in some way or other -- ones that might lead to some sort of conceptual breakthrough.
It sounds like you're asking the question: given that we know there is a property X, but we don't know what that property X is, what's the term for that ignorance about X? Scientists very rarely will say the "given that we know that there exists a property X" part. At best, they'll suggest that it might be the case and then test it. To commit yourself to a solution space and hence commit yourself to a "known unknown" (to quote Rumsfeld) is to overstep your responsibility. In science, as in all other walks of life, once you've formulated the question such that you know what the answer will look like, you pretty much have the answer right at hand.
posted by painquale at 8:20 PM on October 1, 2007
It sounds like you're asking the question: given that we know there is a property X, but we don't know what that property X is, what's the term for that ignorance about X? Scientists very rarely will say the "given that we know that there exists a property X" part. At best, they'll suggest that it might be the case and then test it. To commit yourself to a solution space and hence commit yourself to a "known unknown" (to quote Rumsfeld) is to overstep your responsibility. In science, as in all other walks of life, once you've formulated the question such that you know what the answer will look like, you pretty much have the answer right at hand.
posted by painquale at 8:20 PM on October 1, 2007
Kernel (matrix, algebra).
"... the degree to which a homomorphism fails to be injective."
posted by Chuckles at 8:35 PM on October 1, 2007
"... the degree to which a homomorphism fails to be injective."
posted by Chuckles at 8:35 PM on October 1, 2007
This thread is closed to new comments.
People who know what they're talking about are welcome to jump in at this point.
posted by eritain at 12:42 PM on October 1, 2007