How do Numerical Analysis and numerical methods more generally fit into this view of Computer Science? Numerical methods are widely used in many practical applications, but AFAIK generally aren't viewed as part of Computer Science. I don't know where Numerical Analysis fits in the academic universe now Applied Math departments are disappearing from universities.
Sometimes they are part of computer science! Berkeley had a long tradition of hosting numerical analysis and scientific computing as core to CS. I do worry that this might not continue after Jim Demmel retires.
Computer Science Departments were each shaped by the predilections of their founders. This is why optimization research remains core at the University of Wisconsin, and numerical analysis remains core at UC Berkeley. Computer Scientists, like all other academics, tend to hire people who work on things similar to them.
Numerical analysis belongs to the Applied math department. They are about the math of turning continuous math into discrete computer representation, though the math part is mostly about Taylor expansion and Fourier analysis. Numerical methods are algorithms which can be part of stat/cs or mechanical engineering etc. but I wish it belongs to computer science which means more funding
Really enjoyed this historical perspective. I'm sure you've read it, but just in case, "How data happened" (https://wwnorton.com/books/how-data-happened) also gives a very interesting history mid-to-late 20th century computer science. In particular, there's a very interesting discussion about why machine learning and data science developed in CS departments and not in statistics departments (because the former was worried about practical uses, aka engineering stuff, while the latter was more interested in theoretical stuff).
It's a great book. I have a complementary take, which I've written in pieces here and in full in a forthcoming book: modern ML was developed before computer science departments existed, and much of it was done in government and industrial labs.
Thanks for posting. I did not know about Louis Fein at all!
But one of the posts I wrote that got a lot of clicks back in the day was this alternative history of Turing (based on articles by historians Haigh and Daylight in the ACM Communications): https://blog.castac.org/2015/03/how-influential-was-turing/
As they tell it, the people who worked with computers were often psychologists or mathematicians and they did not want to make their field about computers; they wanted it to be about everything and so it became a science that was about information and algorithms (which are obviously in the computer but are also in our heads and in society) rather than about computers per se. Thus, the elevation of Turing Machines in the elementary computer science texts even though Turing Machines had very little to do with the computers being built (although Turing did, later on, build and work with computers at Manchester).
This is a great post! Thanks for sharing. (For whatever it's worth, it also strongly aligns with my own reading of the history of Turing's role in the first US computers.)
I guess I'd slightly quibble with the notion that theory wasn't needed for the first computers. There were very mathematical people involved! But let me read all of the references in your post before opining too much further.
Aren't Computer Science departments really Computer Engineering rather than science? There AFAICS there is precious little that can be called science and lot more that is about engineering as a discipline. Computer science should really be another part of engineering, like mechanical engineering or electrical engineering.
It's helpful to compare with library science and management science. These departments are about the associated branch of knowledge of a practical skill set.
But knowledge can be acquired in many ways, e.g., from tradition, which is not how science works. I cannot think of a single piece of knowledge applied to my MBA training acquired by what I understand as science from my biology degrees. Economists like to argue that they use science, but they really don't, except for those doing behavioral experiments in economics.
Computation isn't something that is discovered, but rather a discipline that uses math (sometimes) and design, which is not science.
> No matter what op-eds published in the New York Times say, computer science is as political and ideological as economics.
Could you please elaborate?
Our coursework was things like Data Structures, Operating Systems, Networking, PL theory. Didn't get the impression any of it was political or ideological. What's political about Red Black trees :)
How do Numerical Analysis and numerical methods more generally fit into this view of Computer Science? Numerical methods are widely used in many practical applications, but AFAIK generally aren't viewed as part of Computer Science. I don't know where Numerical Analysis fits in the academic universe now Applied Math departments are disappearing from universities.
Sometimes they are part of computer science! Berkeley had a long tradition of hosting numerical analysis and scientific computing as core to CS. I do worry that this might not continue after Jim Demmel retires.
Computer Science Departments were each shaped by the predilections of their founders. This is why optimization research remains core at the University of Wisconsin, and numerical analysis remains core at UC Berkeley. Computer Scientists, like all other academics, tend to hire people who work on things similar to them.
Numerical analysis belongs to the Applied math department. They are about the math of turning continuous math into discrete computer representation, though the math part is mostly about Taylor expansion and Fourier analysis. Numerical methods are algorithms which can be part of stat/cs or mechanical engineering etc. but I wish it belongs to computer science which means more funding
Really enjoyed this historical perspective. I'm sure you've read it, but just in case, "How data happened" (https://wwnorton.com/books/how-data-happened) also gives a very interesting history mid-to-late 20th century computer science. In particular, there's a very interesting discussion about why machine learning and data science developed in CS departments and not in statistics departments (because the former was worried about practical uses, aka engineering stuff, while the latter was more interested in theoretical stuff).
It's a great book. I have a complementary take, which I've written in pieces here and in full in a forthcoming book: modern ML was developed before computer science departments existed, and much of it was done in government and industrial labs.
Thanks for posting. I did not know about Louis Fein at all!
But one of the posts I wrote that got a lot of clicks back in the day was this alternative history of Turing (based on articles by historians Haigh and Daylight in the ACM Communications): https://blog.castac.org/2015/03/how-influential-was-turing/
As they tell it, the people who worked with computers were often psychologists or mathematicians and they did not want to make their field about computers; they wanted it to be about everything and so it became a science that was about information and algorithms (which are obviously in the computer but are also in our heads and in society) rather than about computers per se. Thus, the elevation of Turing Machines in the elementary computer science texts even though Turing Machines had very little to do with the computers being built (although Turing did, later on, build and work with computers at Manchester).
This is a great post! Thanks for sharing. (For whatever it's worth, it also strongly aligns with my own reading of the history of Turing's role in the first US computers.)
I guess I'd slightly quibble with the notion that theory wasn't needed for the first computers. There were very mathematical people involved! But let me read all of the references in your post before opining too much further.
Aren't Computer Science departments really Computer Engineering rather than science? There AFAICS there is precious little that can be called science and lot more that is about engineering as a discipline. Computer science should really be another part of engineering, like mechanical engineering or electrical engineering.
It's helpful to compare with library science and management science. These departments are about the associated branch of knowledge of a practical skill set.
But knowledge can be acquired in many ways, e.g., from tradition, which is not how science works. I cannot think of a single piece of knowledge applied to my MBA training acquired by what I understand as science from my biology degrees. Economists like to argue that they use science, but they really don't, except for those doing behavioral experiments in economics.
Computation isn't something that is discovered, but rather a discipline that uses math (sometimes) and design, which is not science.
> No matter what op-eds published in the New York Times say, computer science is as political and ideological as economics.
Could you please elaborate?
Our coursework was things like Data Structures, Operating Systems, Networking, PL theory. Didn't get the impression any of it was political or ideological. What's political about Red Black trees :)