Hacker News Reader

Comment by pizza | Hacker News Reader

Question for music theory experts who are also coders: does it feel like understanding a programming language, or something else? Admittedly I've never really made any serious independent effort to learn it, or to retain whatever I tried to learn, but it is very difficult for me to even grasp a mental model of how understanding it is supposed to work, or feel, in my head.

I can read a piece of code and know how the bits affect one another. When I read basically anything on music theory it feels like a string of non-sequiturs basically. Maybe there's just some skill floor where that all starts to feel organized?

View on HN

I think the only "universal" part of music theory is the integer resonance stuff, "a perfect fifth is a 2:3 frequency ratio", because integer resonances do seem to have biological significance for us. The rest is culturally dependent: it's not describing some uniquely necessary mathematical object, but more like describing the spoken language of a certain region and time period. The best way to understand it is to play music from that period, notice some common "moves" (like the move from dominant to tonic), and realize "hey, music theory has a name for that".

You wouldn't use music theory to write music, same as you wouldn't use a grammar book to write a novel. You write music by intuiting the sequence of moves you want to make, guided by your ear and listening experience, and maybe even coming up with some new ones. Then if you're successful, theorists will make theories about it.

The canonical video on the subject: https://www.youtube.com/watch?v=Kr3quGh7pJA

Some people will naively say "All music has pentatonic! Universal!" but ignore vital cultural context. One example in the video is ragas: they can look a lot like Western scales and modes, but they have a time and context.

Western music also has a time and context, but it doesn't stand out as much to us unless someone goes full ham breaking that context like with the Imperial March in a Major mode: https://www.youtube.com/watch?v=B9MShtCg4fk

It changes the mood from "this is the bad guys!" to something more positive and triumphant.

There are also several traditions that have scales/modes where certain note sequences (typical pairs) are only "allowed" in one direction (i.e you can play #3 after #1 if you're ascending from #1 to #3, but not the other way around).

This adds additional confusion to comparing scales/modes between cultures.

We do have at least one such mode/scale in the western european classical tradition, but it is not widely used.

So it's not biological significance... You are born loving the sound “shush,” white noise. This is also the great irony of people complaining about distortion and thick bassy kick beats, we actually like it because it's warm and “takes us back”: indeed even to the noisy slushy womb and the heartbeat there.

Rather it's that things that hit one tone either also tickle the hairs in your organ of Corti at 2x, 3x, 4x, 5x the frequency (strings) or at 3x, 5x, 7x, 9x the frequency (half-open acoustic resonators—guitar bodies, panflutes), and your brain feels comfortable when those tickle together as they are familiar.

I would disagree with the second point. Music theory has definitely changed how I make music. The initial idea is typically still conceived intuitively, but I definitely use music theory to develop it into a whole piece. Before knowing music theory, developing an idea further felt a lot like applying a brute force algorithm. Now, I can recognize what my initial idea is doing and I know where I could take it - both by following conventions and by breaking them.

They have totally different goals.

Music theory is descriptive, not prescriptive. Oodles of people make music without knowing a lick of it, and people who learn it in academic environments still typically make music starting from other concepts. "Listen to records and vibe with what they are doing" is typically a much more effective way of learning to improvise seriously than "learn the theory rules", for example.

"Music Theory" also typically is very tightly bound to harmonic analysis modes that are focused on about two centuries of western european music. This is necessarily going to limit its effectiveness when applied in other spaces (even here, when applied to jazz).

This is exactly what I got wrong when I started learning jazz piano as an adult. My wife is a lifelong musician and I got to the annoying point where I'd be playing and she'd walk by and go "that third chord sounds wrong." I'd ask why and she'd be unable to elaborate until she sat down and played the better version intuitively. Then she'd retroactively analyze what she did to explain it to me. I understood the theory, she understood the sounds. It was always humbling and helped me understand that intellectualizing comes second.

A bit. Music is a bit like programming but I think there are 2 distinct ways to practise music: performance and composition. Theory helps you with both. I think that there's less of a "performance" aspect to programming, it's more "composition-like". Chess is an area where you have a combination of theory and skill that, to me, is more music-like (because you can play chess in a "performance way" and also in a "composition way").

If your goal is to be able to improvise then the purpose of theory is really ear training: it allows you to develop an instinct about what to do with your body to produce a desired result. Improvisation is a bit like "composition in the moment", like speed chess.

There are contrived situations where people treat programming as a performance activity but it's not the way most programmers spend their time.

This was my favorite reply to the question. Music theory is great, and it really can be helpful with composition. But I've also found that what I'm really looking for when trying to improvise, or perform, is having studied and worked on the theory - specifically on the ear training and the process of connecting that to my fingers - to the point where I don't have to think about it any more.

The composition vs performance framing is very useful I think. Some musicians might just want to do composition, and can spend a little more time between the notes figuring out what ought to come next. (Although even in this case, it's easy to get bogged down in analysis and forget about what feels right). But a lot of musicians want to be able to improvise - and the difference is like being able to write down a compelling argument vs being able to make a compelling "speech" off the cuff.

Yeah, in both cases, you're going to need a good understanding of grammar and vocabulary - but when you're trying to make that speech (or improvise a solo), you're just not going to have the time to think about "the sound I want to hear next is the sound of the 5th of the chord the band is implying." All our practice and theory homework is in service of the fingers just doing what's in our heads automatically - and sometimes, even surprising ourselves with what comes out.

> But I've also found that what I'm really looking for when trying to improvise, or perform, is having studied and worked on the theory - specifically on the ear training and the process of connecting that to my fingers - to the point where I don't have to think about it any more

"The purpose of all theory is ear training" is actually a Warne Marsh quote, although one that I can't verify. My dad (who is/was a jazz musician) read it in an interview in Coda Magazine in 1976 and while I can find references to the interview I can't find a copy of it online.

My dad used the quote "The purpose of all theory is ear training" as the inscription on his self-published basic and advanced keyboard harmony books. It's certainly a quote I've heard him repeat many times over the years.

A quote from a jazz theory book comes to mind. Goes something like this:

Jazz is 99% theory 1% magic. The difference between the greats is that they’ve forgotten all the theory

That makes no sense. They've internalized the theory so much that they can focus on the magic, and each of them has their own kind of magic, but they absolutely know the theory.

There’s a bunch of theory that goes into understanding why a tritone sub works for a Dom7 for the 5 chord, but it’s not something you think about when you’re using it. You learn how it sounds and you start using it instinctively.

Other people can probably figure out better analogies, but I’d say it’s something like the difference reading an article about generics or monads vs actually using them every day. To an outsider these seem like a ton to think about, but when you use them every day they’re just part of your toolbox.

I fit the bill. There are obviously similarities: music notation is a formal language like programming languages. When you read notation, you can pretty quickly grasp certain relationships and substructures within the piece. With jazz notation especially you can quickly figure out what the general strategy of the piece is and once you’ve got an ear for how it works, you can theorize or write little melodies or complementary bits just from sight alone.

I would also say that the obvious impact of the ear and/or the tools you’re using to play music make it very different than programming. Without the knowing how you would actually play a piece of music, it can be difficult to visualize or interpret notation, whereas code is pretty straightforward in how it’s made or what it actually does. The “feel” is really abstract and interacting with the piece is more akin to observing a running system than it is to touching code.

Not in my experience.

Sure, music theory involves learning the "syntax" of music notation, the "semantics" of e.g. key and genre, and the "programming pearls" of common musical idioms like tritone substitutions. There are analogs.

But what distinguishes music theory is:

1. The language and notation is hundreds of years old. It's been with us in one form or another 200 times longer than COBOL has. Maybe longer.

2. You don't have the peculiar challenge of having to speak it in the very precise form that a computer requires of a programming language. The terminology and communication between musicians is much fuzzier than that, much more like a human language.

(More like 20 times. But what's an order of magnitude to a musician accustomed to dB anyway.)

Western Music theory is an attempt to formalize certain things about music. But it's better with proper context. Like, say, a song is not about the music theory you put in it. A good song starts with a good story--and that's another craft altogether--which you make into good lyrics. Good Lyrics combine story, language quirks and phonetics. The lyrics are a heavy ingredient that needs spacing and intonation which are good-sounding and practical to a human singer, and that needs to mesh with melody and rhythm. Then comes harmony and the doses of emotion you can put into it, and that's what this article is about.

You can certainly code lots of ingredients. For example, you can use embeddings to find better rhymes faster when composing a lyrics, or even LLM help to check if your lyrics make sense (If ChatGPT can guess what your lyrics are about, a human may be able too). You can use a host of computer utilities to find harmonies, and even use voice synthesizers to sketch the singing. But at the end, what makes a piece of music interesting is how well it meshes with its listener...you can model that with statistics, if you are pedantic enough, but at that point the maws of hell open and suck your soul out.

Haha this is the second time my degree has come in handy after getting me a job in a record store after college.

I’d say “music theory” as a general concept is closer to understanding computer science principles like managing complexity or the memory hierarchy. Individual disciplines, such as orchestrating for a specific kind of ensemble (string quartet, big band…) or writing in specific forms (fugue, AABA popular song…) are closer to mastering practical programming tasks like working fluently in a given programming language.

[I am not a music theory expert but just finished two semesters of music theory at community college and have several decades of experience as an adult]

Functional harmony is usually what people mean by music theory.

Functional is the key to my understanding. Functional Harmony is what has been done harmonically and melodically in styles of music derived from The Period of Common Practice. Functional Harmony describes what some people did. Using it makes music that sounds like what those people did.

From the street where I live, I see cultural institutions that ascribe moral value to the conventions of The Period of Common Practice. There’s an industry built up around The Period of Common Practice and the flow of money means there are jobs for those of jobbing age and careers for those of careering age.

It is a disjointed mess like any jobs program.

Learning music theory is more about expanding your taste vocabulary and giving names to each new thing.

To me it's one of those midtwit memes where on the left and right side, it's like "they're so similar" and in the middle "they're totally different".

On the left, they're similar because on the surface level, you're basically executing scripts and there's a lot of jargon to describe higher levels of abstraction.

In the middle, they're different because, as others have said, music theory only loosely maps to how people write music, whereas software development seems much more algorithmic.

But as I became more experienced, I realized software is way more of an art than a science, and that all the best practices are contextual.

As someone who has come back to jazz improvisation after many years, and being a coder, the performance aspect is a little like playing a video game.

But this game has 1) very short levels (1-2 seconds), 2) repeated patterns in the levels and sequence of levels, 3) strong incentives to solve levels in novel and interesting ways, but 4) there are hidden mines everywhere which can damage you, but 5) you get points for hitting the mines in the right way so that you avoid damage. The mines are “wrong notes.”

Music theory gives you power ups to see the mines, quickly plot paths, and treat sequences of levels as part of a meta-level. But you have to be FAST.

(I have been known to stretch an analogy too far..)

Yes - well, almost. As coders, theory scratches the itch of understanding what's happening "under the hood." We might say that "Bach's music is brilliant", but why? Theory helps make the music, and the things that make it sound interesting, explainable.

That said, making music is not merely a technical exercise - there could be multiple theoretical ways to describe a harmony. You still have to listen to it and ask, "what makes the most sense? Is this functioning as a resolution to some tension?" It's easy in a classroom to lose sight of the fact that the theory was invented after the fact.

If you're programmer-brained, lots of things will feel like understanding a programming language to you. There's no particular programming-ish qualities to music theory.

I am not an expert, but to me music notation feels more HTML or json than code.

Reply on news.ycombinator.com