Slashdot Mirror
Forget Math to Become a Great Computer Scientist?
Coryoth writes "A new book is trying to claim that computer science is better off without maths. The author claims that early computing pioneers such as Von Neumann and Alan Turing imposed their pure mathematics background on the field, and that this has hobbled computer science ever since. He rejects the idea of algorithms as a good way to think about software. Can you really do computer science well without mathematics? And would you want to?"
Who needs math? Bogosort is a good a sort algorithm as any. Hey, without math, how would you be able to tell?
Maths IS needed for computer science. Just be sure not to confuse Computer Science with Software Engineering. Software engineering is only a part of the computer science sphere.
Do the lessons of VB6 teach us nothing?
COMPUTING IS HARD. You can't dumb it down just because it would be nice to do so. And I'm sorry but mathematics is just the way in which meaning is expressed for machines. There's no free lunch here. And he's wrong about algorithms too - since a non-terminating algorithm is always expressible by deconstruction into a series of terminating algorithms.
Taking CS without math is like taking engineering without any physics.
WTF is the the author smoking?.. There are of course parts of CS that are less involved in math, but it is still overall a fundamental part.
- These characters were randomly selected.
Without mathematics, there is no internet porn.
Good luck on doing a kernel, file system, network stack, crypto, image processing, window manager, animation or 3D without math or algorithms. I look forward to reviewing some of this guys code.
I attained a Computer Science BS in 1986. At the time everyone was getting Math minors. I opted for a communication minor instead. I've worked in high-tech engineering environments with real-time programming for many years. What I found is that I've never needed the intense mathematics attained by those with math minors. I needed to be able to implement equations that staff mathmaticians would develop. Though math is a fundamental of computer science, I believe the ability to logically assess a situation from multiple perspectives; communicate your approach with the customer; and then implement a maintainable solution is the key components required for computer scientists.
This guy just doesn't seem to understand what math is. Substituting theory of computation with his "theory of expressions" just shifts focus on another field of math.
Mainly, he claims to want to create a "comprehensive theory of process expression". Fair enough, but as soon as you want to extract usable, reliable results from your "comprehensive theory", you've really just created a branch of mathematics. Maths is not just numbers and calculus, but any systematic treatment of relations in a symbolic fashion - unless he plans a lot of fairly useless hand waving, "Oh, my process is expressed as *insert long winded ambiguous English description", he will be working within the remit of mathematics. Heck, one of my areas of study is the development of processes (studied through the use of process calculi) - a highly mathematical tool.
He also ignores the vast array of work on non-deterministic algorithms, stating that "Any program utilising random input to carry out its process, such...is not an algorithm". Sure, it's not a deterministic algorithm, but even if you artificially restrict your definition of algorithm to just be deterministic, it's a useful tool in analysing such problems.
Finally, statements such as "Computer science does not need a theory of computation" are just so bizarre as to be funny. I suggest he forgets all he knows about formal computational theory, and I'll contract "Theseus Research" to write me a program to determine the halting problem for an arbitrary program. I wonder what his bid will be, given that he doesn't need a theory of computation (that would tell him you can't do it, at least with our models of computation - and probably with any).
Now, all of this is not to say you can't make progress in computer science without the mathematics that's currently been developed - however, you will either spend a lot of time achieving unreliable results, be reinventing the wheel, or just be creating a new branch of mathematics.
....Abstraction.
And computer science, the software side, is really the science of abstraction physics.
http://threeseas.net/abstraction_physics.html
At some point in the higher levels of abstraction creation and use you use the lower mathematical level as more or less a carrier wave of the higher level abstraction, than for the purpose of performing a mathematical calculation. The analogy is that of using radio waves to carry the music you hear over the radio, but the carrier wave is discardedafter it has done it job. Likewise, the mathematics of computers boils down to binary flipping of transistor swiches upon which the higher level of mathematics is carried upon.
With a correct approach to the abstraction manipulation machine computers really are, we can accomplish a lot more, similar to the difference between using the limitation of roman numerals in math vs. the decimal system with its zero place holder.
This is something I've thought a lot about. There have been any number of times that math has helped me in my software development efforts. Things like trig to predict the path of a moving target in Robowars (back when I was in high school) to various vector and angle related maths in CustomTF for Quake 1 (www.customtf.com) to partial derivatives to calculate the slope on a surface. I've also needed math for various economics related things over the years, and probability and statistics have also been exceptionally useful to me. Currently I'm having to decipher a guy's code which is all eigenmath, so my linear algebra course is saving me from having to hire someone just to explain all the math to me.
But the kicker is that you can't just tell a student that they should "study vector math" because one day they'll write a Quake Mod, because, truth be told, they probably won't. It's the trouble with all examples you give when students ask how math will be useful -- I could pull any number of examples from my life, but the problem is, they probably won't happen in a student's life. Instead, they'll have their own trials. The best you can tell someone is to study all the math they can, because some day it *might* be useful, and they'll want to have that tool in their toolkit.
And that's just not a very satisfying answer to students who want to make sure that they'll be damn well using what you're teaching in the future.
But believe me, I thought I'd never have an application for eigenvectors, and now not only do I have to clean out my brain on the topic, but I have to parse someone else's code (PhD thesis code no less) and add functionality to it. Two other friends of mine got stuck on legacy Fortran apps which are essentially mathematical solvers (one for differential equations, the other for huge linear algebra problems), and both of them are extremely happy they paid attention in their respective math classes.
So, yeah. To CSE students out there: take math. Pay attention. It could very well save your neck some day at a job, and if it doesn't, at least try to make it interesting to yourself to think of applications where you might use them. All math through the first two years in college can find applications for it quite easily.
The term itself is a product of the academic environment, similar to the equally dubious "Library Science" and "Management Science". For what it's worth, the European term "informatics" would have been better, but never caught on.
That said, I believe there's a useful set of relationships well understood in other fields:
Science = The search for fundamental knowledge and predictive models;
Engineering = The creative application of the results of science;
Technology = The routine application of the results of engineering.
giving us, for example:
Science: Physics
Engineering: Electrical engineering
Technology: TV Repair, Cable TV Installation
The punch line is that application of this model to computing works as follows:
Science: Mathematics
Engineering: Programming, Informatics, "Computer Science"
Technology: Coding, Computer Installation, Home Computer Repair, etc.
Mathematics IS the science in "Computer Science".
Anyone who has studied advanced Mathematics knows that Math is not about numbers; think of mathematical logic, Boolean algebra, abstract algebra, set theory, topology, category theory, etc. ad infinitum. Dijkstra defined Mathematics as "the art of precise reasoning". In the same sense, "computation" doesn't mean "number crunching", but more generally the automated manipulation of information.
It is true that there are legitimate concerns in today's computational landscape (networking, concurrency, etc.) which didn't figure in the mathematical/engineering world view of the 1940s, but that's simply a sign that the field has grown up (i.e. grown beyond the limited perspectives of its founders). That's also true in many other applications of Mathematics. For example, early research in differential equations paid much more attention to linear differential equations (because they were more tractable). However, we now know that most "interesting" systems in the real world involve non-linearity.
Science, Engineering, and Technology share with living systems an important rule: "Grow or die!" Fortunately, the field of computing has grown.
I believe the author's point isn't that you don't need to know any mathematics, or that it doesn't have an important role to play in CS. He's simply arguing that some of the main issues in computer science are not fundamentally mathematical problems (even if they require some mathematics).
If you buy that argument, then treating CS as if it were merely simply another branch of mathematics will not help solve those problems.
Of course, this also takes us into the perennial debate between where to draw the line between "computer science" and "software engineering". One could certainly define away the author's problem by saying that his examples are software engineering issues rather than computer science issues. And it's true that it's software engineering has been driving a lot of the theory with respect to expressiveness (design patterns and the like). But that view also seems to really impoverish computer science - if all you leave the field of computer science is the stereotypical mathematics, why not just become an applied mathematics major?
Let me make this clear: your ability to write code in no way makes you a computer scientist. It's like saying that the ability to operate a forklift makes you a structural engineer. Stop it already.
That said, I'm sure you're good at what you do. I bet you can write good code in VB, as well as many other languages. This isn't a personal insult. VB, PHP, and other brutish languages are equally bad in my eyes.
These languages are brutish because they oversimplify key concepts. That oversimplification also makes them attractive to new programmers, and new programmers universally write terrible code. The languages themselves aren't bad, the coders are. That said, more experienced coders will generally choose more capable languages, so most of the time, a program written in a brutish language will be a bad one.
We need fewer programmers, not more. Maybe professional certification would help somewhat.
(Incidentally, we were lucky that Javascript became the de-facto client-side web language. We could have done far, far worse, and although we can change server languages, we can't change a user's web browser!)
The author really sucks at math but heard that there's big bucks in the computer stuff, right?
Computers are (by their very definition as well as by the word used to describe them) mathematical machines. A computer can essentially do NOTHING BUT calculate. It can in its core add, subtract, shift and move data around. How is this supposed to work without math?
We used to have a Bill of Rights. Now, with the rights gone, all we have left is the bill.
It's lucky Jobs went to his calligraphy classes; if he hadn't we'd all still be using monochrome terminals. (A pretty arrogant thing to imply)
// MD_Update(&m,buf,j);
What you have written is 100% nonsense and puts you in exactly the same crank camp as Fant. It is always interesting to hear people that don't understand computer science describe what is wrong with it. The model of interaction that you (and he) describe is normally called Reactive software, and it is true to say that it cannot be modelled by a Turing Machine as it performs interaction continuously rather than at the beginning and end of the computation.
From here you've both made a giant leap to assume that programs can't be described by an algorithm. You haven't understood that the difference between a "computation" and "reactive software" is actually a technical triviality that is easily overcome. Indeed it is so trivial that most languages simply ignore it and have stateful operations for input/output. Reactive programs are normally modelled as a sequence of algorithmic steps, everything that the program does apart from sending / receiving data is modelled by an algorithm. So we can either consider this "non-algorithm" to be a sequence of algorithms or consider the program as an algorithm operating over a larger state that includes the environment. The input/output actions become alrgorithmic state transitions over the program/environment state. Look at the way programs in CSP/CCS or other process algebra are written to how this works. To see how the theory of algorithms can be applied to reactive systems take a look at multi-headed Turing Machines.
Finally, if you're going to lob a technical term into a discussion then you should understand what it means. Automaton is a well defined term in CS, and it doesn't mean what you think. In particular what you are describing is not a decision problem and so there is not a problem of language recognition to be solved. I vaguely remembering reading the crank research that you are pointing before, and would like to ask you a simple question. Name one problem that you believe can be computed by a UBM, but not by a UTM?
Slashdot: where don knuth is an idiot because he cant grasp the awesome power of php
"The notion of the algorithm," he concludes "simply does not provide conceptual enlightenment for the questions that most computer scientists are concerned with."
The assertion that computer science is not math is similar to the assertion made in the book "The World is Flat" saying the world is now "flatter" than it used to be. In the case of the flat world, Friedman (the author of "The World is Flat") claims the world is flat to create a sense of shock that he can then use to get his message about globalization across. In the case of "computer science is not math" Fant here is trying first to shock as a method of capturing attention...
Most Americans use math in the singular. The Brits say maths. That is because there are multiple branches of mathematics. What we are discovering is that the tie between arithmetic and calculus and computer science is falsely reinforced. The fact is there are other branches of mathematics that are more important to computer science. There are also many new branches of mathematics that need to be developed in order to solve the new kinds of problems we are trying to solve in modern computer science.
I am really bothered by programmers who, when I interview them, say they have been writing software for years and can't remember ever having to use math.
I know they can't possibly mean that... or they don't know what math is...
I know that in several years of programming you must have at least been tempted to write an if statement or at least one loop of some kind.
The if statement uses a form of algebra called boolean algebra. It was named after George Boole who was very much a mathematician. I know that there are many programmers today who use the if statement and this form of mathematics makes up a large part of many programmer's jobs. I guess it must be falling out of fashion.
I know how to perform boolean algebraic operations on a white board and I have many times been confronted with a gigantic morass of if and else if statements and using simple truth tables and a little boolean math have reduced enormous sets of ifs down to just a few.
The new computer science needs to focus on solving problems involving processes. Processes are like algorithms in that they have a set of instructions but they are unlike algorithms in that they also have many temporal components and may exhibit parallelism, asynchronous invocations, and may not have a finite product. These are the types of problems addressed in newer mathematic disciplines that are trying to see information processes not as tied to computing machinery but as tied to the natural world.
Computer Science may point to a new kind of science that describes an underlying natural computational order of the universe. We are starting to observe computational processes everywhere, in the brains of animals, to the interactions of ecosystems, to quantum mechanics. We may lack the right mathematics to describe these things and we may have to invent new kinds of math but that doesn't mean that math becomes unimportant. An understanding of math can help when studying logic and so too would it help in studying any new disciplines that we may need to invent.
New kinds of math are invented every day to describe new kinds of problems. To say you don't need math to study any formal science let alone computer science is just silly. It is just something shocking to say that grabs attention... and the article nearly contradicts itself by the end... and it's only 7 paragraphs. The distinction Fant makes is nearly academic. Just as the distinction between a Statistician, a Geometer ( a mathematician who studies geometry ), and a Logician is academic. Yet that is not what the readers of the headline will read... Fant is arguing to make computer science a new kind of science much as Wolfram has. Yet it would be sil
[signature]
Oh I see. All this time I was lead to believe that Donald Knuth created TeX to satisfy the desperate need for a half decent digital typography tool and after all it must have been due to some class that steve jobs took when he dropped out of college. Knowing that TeX remains to this day the best typesetting system and knowing a bit about Adobe and the history of PostScript, I guess that that half baked assertion makes sense and must be true.
...or maybe not.
Please. Steve Jobs doesn't walk over water, nor is he behind every single thing which can be accounted as progress in the computer world. This whole jobs-worshiping thing is starting to become ridiculous.
Slashdot, fix your code or at least hire someone who is competent at it to do it for you.
is the same as writing litterature with a programming language.
The reason computer science is so heavily influenced by math is the binary architecture that every piece of hardware is designed around. Every real world problem, right down to choosing the color of a font, has to be translated into the digital world by algorithmic approximation - a lot of math! The problem is that it is this very abstraction that makes computers so "flexible" in what they can do. Analog computers existed many years ago but they could only ever be built for a single purpose.
Unfortunately(?) it is much easier to design and mass produce something which is based on a finite lowest common denominator (bits) than it is to do so based on the continuum that a non-digital solution would require.
That said, who's to say that a beautiful painting rendered in Gimp/PhotoShop isn't a program of sorts? Certainly it has input, (from the original creator), and output, (its effect on us), and the "code" can be modified to change both!
Yes, but I wrote a new book that claims that debate is better off without logic. Early debating pioneers such as Kant and Aristotle imposed their logic background on the field, and this has hobbled debate ever since. I reject the idea of convincing arguments as a good way to resolve any conflict.
Slashdotter, ID #101. UIDs are in binary, right?