Scan barcode
A review by lpm100
The Half-life of Facts: Why Everything We Know Has an Expiration Date by Samuel Arbesman
informative
fast-paced
3.0
Book Review
The Half Life of Facts
3/5 stars
"It just didn't feel like there was that much value added."
*******
Of the book:
-211 pages of prose over 10 chapters= 21 pps/chapter
-208 point citations, ≈1 per page.
This seems to be a decent book, but somehow I've read many of these things before in other places.
Toward the end, it seems like the author was straining to fill up some type of page quota. If there was a connection between the thesis of the book and the things that he was talking about, it didn't come through for me.
Of course, it never hurts to revisit/recapitulate sundry aspects of Epistemology--and this author is an applied mathematician and so he should have been the best choice to write a book like this. (But, I have read one applied mathematician that is done a bit better.)
I think the whole book could be read in a couple of afternoons. (211 pages of fast flowing prose.)
A bit from each chapter:
1. Author marks out his territory in describing "mesofacts." This is midway between things that are well known and impossible to overturn (such as the fact that there are five fingers on every hand) versus things that change every single day (the weather).
2. A Derek Price in this book notices around 1947 the exponential growth of knowledge in journals. Author lets us know that fields have unique doubling times. The number of entries in a dictionary of national biography doubles once every hundred years. But the number of asteroids known doubles every 10 years. Medicine and hygiene has a doubling time of 87 years whereas Genetics has a doubling time of 32 years. (Stylistic problem: only English people know the Euler number as the Napier constant.)
3. Knowledge increases exponentially, but it also decays exponentially. The author took the specific field of medical knowledge about cirrhosis or hepatitis. He calculated that it took about 45 years for half of it to be disproven or become out of date. Papers also decay in the amount that they are cited by other authors. (Economics has a half life of 9.38 years.)
4. A lot of technology approximates Moore's Law. (Doubling every 18 months.) If not directly that, it is exponential with some other unspecified doubling time. Tightly linked logistic curves can, when combined, yield a smooth curve over time. (But then, we don't know if this smooth curve will have logistic curve characteristics as well.) Population growth is limited by technical progress, but then technological progress is helped by population density above a certain level.
5. Re-exposition of network theory. (Six degrees of separation, strong versus weak ties, short versus long ties, et al.) "A lie can travel halfway around the world while the truth is still putting on its shoes."
6. Undiscovered knowledge. Multiple independent discoveries. Meta analyses. Even medieval publications can be fit onto a logistic curve.
7. Analogizing of changes of state of knowledge to physical phase transitions (it's an interesting analogy, but a little bit strained --phase transitions are reversible with changing conditions). Knowledge may increase gradually (knowing one new species one at a time that doesn't lead to anything earth-shattering) or, it may occur in leaps and bounds (once we had an understanding of the atom, then we could do GPS).
8. Things learned in learning how to measure certain things (mountains, trees, India). This chapter is a synopsis/recapitulation of many things learned in Simon Winchester's "The Perfectionists." Furthercomments about how poor the reproducibility of certain experiments is.
9. Technology can be defined as "anything that was invented after you were born." Arthur spends some time discussing types of reasoning and error and why errors stay around for such a long time. And why scientific progress proceeds one funeral at a time.
10. The author talks a little bit about the limits to our knowledge. (He doesn't actually make any predictions about the consequence of having so much knowledge; that was more aptly and more accurately done by Nate Silver in the book that I will reference below: Silver suggests that the political polarization that we see today is people's response to dealing with too much knowledge.)
Verdict: Weak recommendation. I would not pay more than $5 for this book. Also, in the event that you have a choice between this book and Nate silver's "The Signal and the Noise," I would recommend that you pay for the Silver book instead.
Other books that I've read that cover: knowledge from this book in a more interesting fashion:
1. Rigor Mortis, Richard Harris (talks about reproducibility issues in biological experiments.)
2. The Perfectionists, Simon Winchester (talks about the difference between accuracy and precision and about increasing accuracy in measurement).
3. Nate Silver, The Signal and The Noise (he takes a lot of specific case studies of things that are actually measured and shows why and how the measurements are / are not good. A variety of interesting topics.)
Vocabulary:
Lazarus taxa
S-curve theory
Logistic curves
Carrying capacity
zeta-
femto-
yotta-
yocto-
actuarial escape velocity
Hawthorne effect
preferential attachment (Matthew effect)
Godwin's Law (reductio ad Hitlerium)
citation decay
Meta-analysis
Cumulative meta-analysis
mesofact(s)
taxonomic bias
taxonomic chauvinism
Factoids:
1. (p.91) "Only about 20% of scientists who cite an article of actually read that paper."
2. The entire Popeye franchise was because of a decimal error in the content of spinach--which contains nowhere near as much iron as the cartoons would have you believe.
The Half Life of Facts
3/5 stars
"It just didn't feel like there was that much value added."
*******
Of the book:
-211 pages of prose over 10 chapters= 21 pps/chapter
-208 point citations, ≈1 per page.
This seems to be a decent book, but somehow I've read many of these things before in other places.
Toward the end, it seems like the author was straining to fill up some type of page quota. If there was a connection between the thesis of the book and the things that he was talking about, it didn't come through for me.
Of course, it never hurts to revisit/recapitulate sundry aspects of Epistemology--and this author is an applied mathematician and so he should have been the best choice to write a book like this. (But, I have read one applied mathematician that is done a bit better.)
I think the whole book could be read in a couple of afternoons. (211 pages of fast flowing prose.)
A bit from each chapter:
1. Author marks out his territory in describing "mesofacts." This is midway between things that are well known and impossible to overturn (such as the fact that there are five fingers on every hand) versus things that change every single day (the weather).
2. A Derek Price in this book notices around 1947 the exponential growth of knowledge in journals. Author lets us know that fields have unique doubling times. The number of entries in a dictionary of national biography doubles once every hundred years. But the number of asteroids known doubles every 10 years. Medicine and hygiene has a doubling time of 87 years whereas Genetics has a doubling time of 32 years. (Stylistic problem: only English people know the Euler number as the Napier constant.)
3. Knowledge increases exponentially, but it also decays exponentially. The author took the specific field of medical knowledge about cirrhosis or hepatitis. He calculated that it took about 45 years for half of it to be disproven or become out of date. Papers also decay in the amount that they are cited by other authors. (Economics has a half life of 9.38 years.)
4. A lot of technology approximates Moore's Law. (Doubling every 18 months.) If not directly that, it is exponential with some other unspecified doubling time. Tightly linked logistic curves can, when combined, yield a smooth curve over time. (But then, we don't know if this smooth curve will have logistic curve characteristics as well.) Population growth is limited by technical progress, but then technological progress is helped by population density above a certain level.
5. Re-exposition of network theory. (Six degrees of separation, strong versus weak ties, short versus long ties, et al.) "A lie can travel halfway around the world while the truth is still putting on its shoes."
6. Undiscovered knowledge. Multiple independent discoveries. Meta analyses. Even medieval publications can be fit onto a logistic curve.
7. Analogizing of changes of state of knowledge to physical phase transitions (it's an interesting analogy, but a little bit strained --phase transitions are reversible with changing conditions). Knowledge may increase gradually (knowing one new species one at a time that doesn't lead to anything earth-shattering) or, it may occur in leaps and bounds (once we had an understanding of the atom, then we could do GPS).
8. Things learned in learning how to measure certain things (mountains, trees, India). This chapter is a synopsis/recapitulation of many things learned in Simon Winchester's "The Perfectionists." Furthercomments about how poor the reproducibility of certain experiments is.
9. Technology can be defined as "anything that was invented after you were born." Arthur spends some time discussing types of reasoning and error and why errors stay around for such a long time. And why scientific progress proceeds one funeral at a time.
10. The author talks a little bit about the limits to our knowledge. (He doesn't actually make any predictions about the consequence of having so much knowledge; that was more aptly and more accurately done by Nate Silver in the book that I will reference below: Silver suggests that the political polarization that we see today is people's response to dealing with too much knowledge.)
Verdict: Weak recommendation. I would not pay more than $5 for this book. Also, in the event that you have a choice between this book and Nate silver's "The Signal and the Noise," I would recommend that you pay for the Silver book instead.
Other books that I've read that cover: knowledge from this book in a more interesting fashion:
1. Rigor Mortis, Richard Harris (talks about reproducibility issues in biological experiments.)
2. The Perfectionists, Simon Winchester (talks about the difference between accuracy and precision and about increasing accuracy in measurement).
3. Nate Silver, The Signal and The Noise (he takes a lot of specific case studies of things that are actually measured and shows why and how the measurements are / are not good. A variety of interesting topics.)
Vocabulary:
Lazarus taxa
S-curve theory
Logistic curves
Carrying capacity
zeta-
femto-
yotta-
yocto-
actuarial escape velocity
Hawthorne effect
preferential attachment (Matthew effect)
Godwin's Law (reductio ad Hitlerium)
citation decay
Meta-analysis
Cumulative meta-analysis
mesofact(s)
taxonomic bias
taxonomic chauvinism
Factoids:
1. (p.91) "Only about 20% of scientists who cite an article of actually read that paper."
2. The entire Popeye franchise was because of a decimal error in the content of spinach--which contains nowhere near as much iron as the cartoons would have you believe.