Hacker News Reader

Comment by Koshkin | Hacker News Reader

To be honest, Category Theory is the last thing a software developer ("working hacker") should worry about learning. It is indeed a fun subject, it has grown into a large branch of algebra, it has connections with functional programming (lambda calculus) etc. But if one wants to be a good programmer, there is so much of really useful and important (and difficult) stuff to learn before you can even come to appreciate what, if anything, knowing Category Theory can do to improve your software engineering skills, that I would dare to characterize it just as an obscure subject that in all likelihood is not worth spending your time on.

View on HN

Have you actually watched the presentation? If not, I encourage you to do so. Category theory and abstract algebra help us understand what it means to build composable and reusable software components with well defined semantics. There is already ample focus on algorithms and discreet math at university courses, but very little of the above. This is despite the fact that modern software development is becoming more about composing existing code and less about reimplementing classic algorithms. So given the already ample focus on the topics you suggest, I think we can forgive Phil Wadler for giving a very brief glimpse into a promising subject that most hackers would dismiss as obscure/irreverent.

I thing that if you are the sort of person that really likes design patterns, the sort of applied category-theory that the likes of Erik Meijer, Bartosz Milewski, Rúnar Bjarnason or Gabriel Gonzalez talk about is actually a good next step after the GOF Desing Patterns :-)

Some basic notions of folds/unfolds, algebraic data types and recursion schemes are great to have, I think.

I'm curious, what should come before as a far as a mathematical subject ? Graph theory ? Number theory ?

The purely mathematical subjects that are deemed important for programmers are parts of the so-called Discrete Mathematics. The Wikipedia article on it lists what is included. It is a vast collection of topics, but discerning what would be more important (or interesting) for you should not be hard.

Depends on what you do. I work in robotics, so probability and linear algebra are the two most indispensable subjects I know. Graph theory might be a little more generally applicable.

At the end of the day, you don't need a lot of math to write CRUD apps. The more math you know, though, the more approaches you'll have to formulate and solve problems.

A lot of graph theory is implemented as linear algebra, using things like adjacency matrices.

A lot of linear algebra is implemented using graph theory, using things like sparse matrices.

There isn't a specific branch of mathematics you need to know in order to learn Category Theory. Its main role is in providing a way of describing other parts of mathematics (similar to the way set theory and certain logics are commonly used to formulate definitions and proofs). In other words, it's a kind of 'foundation,' so there isn't really anything under it. (This is my understanding anyway—please correct me if I'm wrong here.)

That said, it will probably be rather difficult to comprehend if you haven't been exposed to a good amount of other work in pure mathematics first. If you wanted one other more introductory subject, I'd recommend checking out Abstract Algebra (lots of other mathematics are expressed in its terms, also; so independent of Category Theory it would probably be time well spent. https://en.wikipedia.org/wiki/Abstract_algebra).

Category theory doesn't make someone a better Java applications developer, which is EXACTLY why they should be learning it

I agree it isn't directly useful but I think that is good common knowledge to have and a little exposure can only help.

Reply on news.ycombinator.com