What this project is - Introduction

 

One Song a Day: What This Project Is

I listen to a lot of music, but this project really started a long time ago.

I'm a statistician by trade, and I've been fascinated by the idea of ranking popular music since I first discovered American Top 40 in the early 1980s. I used to listen to Casey Kasem religiously every Sunday morning. Eventually, just hearing the countdown wasn't enough. I started writing the entire Top 40 down on notebook paper every week and keeping the weekly charts in a binder.

And because apparently maintaining somebody else's chart wasn't quite enough work, I started calculating my own running chart of the biggest hits of the year.

Every week.

On paper.

So I suppose what I'm doing here is not an entirely new obsession.

Music as History

There's another side of this that interests me just as much as the numbers.

I like to play amateur historian.

Popular music is a surprisingly good window into the time in which it was made. Songs don't become hits in a vacuum. Tastes change. Culture changes. Technology changes. Events happen. Ideas that seem exciting or provocative in one era can seem completely ordinary in another. Sounds that dominate popular music for a few years can almost completely disappear.

And sometimes a song becomes enormously popular at one particular moment when it's difficult to imagine the exact same record succeeding ten years earlier or ten years later.

I find that fascinating.

So while part of this project is simply asking, "Do I like this song?", another part is asking, "Why did people like this song then?"

What was happening in popular music when it appeared? Where did the artist fit into that world? Was the record following an existing trend or helping create one? Does it sound dated today? If it does, is that necessarily a bad thing? And what, if anything, does its popularity tell us about the people who were listening to it?

I'm not approaching those questions as a professional music historian. I'm approaching them as someone who likes music, statistics and history and enjoys figuring out how those things intersect.

The Experiment

The basic idea is simple:

Every day, I randomly select one song from the history of American popular music, listen to it deliberately, learn about it, and decide what I think of it.

Then I'll document the experience here.

The randomness is important.

I don't want to make a list of songs I think I should hear. I don't want a critic's list of the 1,000 greatest songs ever recorded. And I definitely don't want to quietly cherry-pick records that I already know will make interesting posts.

I want the dataset to decide what I hear.

That means I could draw one of the most famous records ever made. I could get a song I've loved for 40 years. I could discover something wonderful that I've somehow never heard before.

Or I could spend the day investigating a terrible record that everyone inexplicably bought in 1974.

That last possibility may actually be one of the more interesting parts of the project.

The Song Pool

The project isn't drawing from every song ever recorded. I wanted a pool large enough to cover a huge portion of American popular-music history while still having an objective reason for every song's inclusion.

The core dataset consists of songs that reached the upper portion of Billboard's major American pop charts. For the period before the modern Billboard Hot 100 began in 1958, I've extended the dataset backward using Billboard's predecessor charts.

That produces a large historical pool covering decades of changing artists, genres, sounds and tastes.

Once a song qualifies for the pool, however, its fame doesn't matter. The random selection doesn't care whether it's "Like a Rolling Stone" or something that peaked for two weeks and disappeared from public consciousness.

That's exactly what I want.

What Happens Each Day

Once the song is selected, my first job is simply to listen to it.

Whenever possible, I want that initial reaction before doing extensive research. I don't want to be told beforehand that a song is a masterpiece, historically important, critically despised or secretly about the singer's third divorce.

I want to hear the record first.

Then the investigation begins.

I'll look at its chart performance, where it occurred in the artist's career, who wrote and produced it, what else was happening in popular music at the time, and whatever historical or cultural context seems relevant.

When possible, I'll also watch the original music video. For older songs, I'll look for an authentic contemporary television performance, promotional film or other period footage. Seeing how a record was presented to its original audience can sometimes provide context that the recording alone doesn't.

I'll also follow whatever rabbit holes the song creates. There really isn't a predetermined endpoint.

Finally, I'll give the song a score out of 100.

That score represents one thing: how much I like the song.

It isn't my assessment of its objective musical importance. A historically significant masterpiece is perfectly capable of getting a lousy score from me. So is a Grammy winner. Meanwhile, some ridiculous forgotten novelty single may turn out to be something I love.

The historical investigation can change my opinion, too. Understanding what an artist was trying to do—or hearing a song in the context of its own time—may make me appreciate it more.

Or I may understand it perfectly and still hate it.

Where AI Comes In

I'm using AI—specifically ChatGPT—as an assistant throughout this experiment, and that's an intentional part of the project.

It isn't choosing my opinions or deciding how I should rate the songs.

I'm using it more like a research assistant, database manager and endlessly available person with whom I can argue about records.

It helps maintain the rules and data behind the project, researches chart and career information, finds useful historical sources and contemporary performances, and keeps track of what I've already heard and how I've rated it.

But the part I'm particularly interested in is the conversation after I've heard the song.

I can explain what struck me, what I liked, what annoyed me or what simply confused me, and then start asking questions.

Why does this 1979 song sound nothing like what I expected from 1979? Why was this singer suddenly having hits after years of obscurity? Was this production style normal at the time? Why did this become the hit instead of another song on the album? Am I hearing something that contemporary listeners would have interpreted completely differently?

AI is particularly useful for exploring those kinds of rabbit holes quickly.

It can also be wrong. That's an important limitation of the experiment. Factual claims—especially chart statistics, dates and historical information—need to be checked against reliable sources rather than accepted simply because an AI confidently said them.

The opinions, meanwhile, remain mine.

What I'm Hoping to Find

I don't really know.

That's probably the best reason to do it.

After enough songs, I'll have a growing dataset of my own reactions spanning different decades, artists and genres. Given my lifelong fascination with charts and rankings, I'm sure I'll eventually find entirely unnecessary statistical things to do with all of those ratings.

But I'm more interested in the discoveries.

I want to encounter artists I've ignored, songs I've forgotten and entire styles of music I've never seriously explored. I want to understand why records sounded the way they did and why people responded to them when they did.

And occasionally I expect to encounter a record that makes me stare at its chart position and wonder what on Earth everyone was thinking.

Those may be some of the best days.

This isn't an attempt to create a definitive ranking of popular music or tell anyone else what they should like.

In some ways, it's just the much larger and slightly more technologically sophisticated descendant of that binder full of handwritten Top 40 charts I kept as a kid.

One randomly selected song. One day at a time. And we'll see what happens.