The Stratigraphy of Taste: A 20-Year Audio Excavation (2005-2026)

If you cut open a tree, you see rings. If you cut open a hard drive, you see... well, shattered glass. But if you cut open a Last.fm account with 20 years of continuous data, you see geological strata of the soul.

Charles has been tracking his listening habits since 2005. That's 514,754 individual plays ("scrobbles") logged across 22 years. It is a dataset that survives platform deaths (RIP Rdio), format shifts (MP3 to streaming), and life changes (college to fatherhood).

The data reveals five distinct eras. Not just "phases" like "I liked ska for a month," but deep, tectonic shifts in personality. We can pinpoint the exact year Charles stopped being a pop-punk kid and started listening to Frank Sinatra (2012). We can see the precise moment the "Toddler Algorithm" took over his account (2022).

This is an excavation of those layers.

Era 1: The High Energy Phase (2005-2007)

The dataset begins in 2005. The vibe is frantic, loud, and incredibly specific to the mid-2000s American experience.

2005 is dominated by Chris Tomlin (1,338 plays) and Green Day. It’s the sound of a youth group kid discovering distortion. 2006 amps up the energy with Yellowcard (3,626 plays!) taking the crown. This was the peak of pop-punk: high tempos, whiny vocals, and absolute conviction. By 2007, things start to fracture. Death Cab for Cutie takes the #1 spot, signaling the end of the sugar rush and the beginning of the moodiness.

Era 2: The Indie Introspection (2008-2011)

The high school energy crashes into the reality of young adulthood. The guitars get cleaner, the lyrics get sadder, and the synths start creeping in.

2008 is a transitional year—Aqua (#1) is a bizarre outlier, likely a nostalgic joke that went too far, but Death Cab remains strong. 2009 is the full pivot: Death Cab again, MSTRKRFT, and Muse. This is the "Search for Meaning" soundtrack. 2010 sees Yellowcard return (#1), but they are sharing space with Miike Snow and The Bloody Beetroots. It's a tug-of-war between the past and the electronic future. By 2011, Yellowcard and Miike Snow are neck-and-neck (4k+ plays each), representing the final battle between Pop Punk and Indie/Electronic.

Era 3: The Crooner Pivot (2012-2015)

In 2012, something drastic happens. The "Taste Drift" metric spikes to its all-time high of 0.86. Charles stops listening to whiny boys with guitars and starts listening to... Frank Sinatra.

Frank dominates 2012 with nearly 3,000 plays. It’s a hard reset. A maturation. 2013 is a glitch year—Icona Pop takes #1 (likely one song on repeat), but Sinatra is still #2. Then, the transformation completes in 2014 and 2015 with the ascent of Billy Joel. For two years, the Piano Man is the undisputed king. This era is about seeking timelessness over trends.

Era 4: The Folk/Lyrical Deep Dive (2016-2021)

As the world got weirder (2016 onward), the music got softer. The Crooner Era softened into full-blown Folk.

2016 is the year of John Denver. Comfort food for the ears. 2017 is a massive anomaly—Nightwish (#1) and RADWIMPS (#2). This is clearly a specific project or phase, perhaps coding to symphonic metal. But by 2018, Avicii takes the top spot (a tribute after his passing?), followed by a return to folk. 2019 sees Avicii again, but 2020 and 2021 are dominated by Neil Young. In the isolation of the pandemic, Charles retreated to Harvest and After the Gold Rush.

Era 5: The Algorithmic Takeover (2022-Present)

And then, the children arrived.

2022 is the last gasp of autonomy, led by Eydie Gormé (Latin Boleros). But look at 2023: "Canciones Para Niños" is #1. The wall has been breached. By 2024, it is a rout. "Daniel Tiger’s Neighborhood" clocks 6,755 plays. 2025 is even worse: Daniel Tiger hits 8,039 plays.

Charles no longer controls his own audio destiny. He is a vessel for the algorithm of a toddler.

How I Built This

This analysis wasn't manual. It’s built on a custom Node.js pipeline that scrapes, syncs, and analyzes the raw Last.fm data.

The Sync Engine: scripts/sync_lastfm.js handles the brutal task of fetching 20 years of history. It uses a "Monthly Sharding" strategy—saving each month as a separate JSON file (data/2012-03.json) to avoid memory issues and allow for incremental updates. It’s idempotent; it only fetches months that are missing or incomplete.

The Analysis Layer: scripts/analyze_music.js reads these thousands of files and aggregates them in memory. It calculates:

• Taste Drift: Uses the Jaccard Index to measure the similarity between the Top 50 artists of Year X vs Year X-1. A low score means high stability. A high score (like 2012's 0.86) means a total personality reboot.
• Discovery Rate: The percentage of "New Artists" in a given year that had never been played before.

All running in a local Node environment, because owning your data means you can query it whenever you want.

Data current as of March 2026. Daniel Tiger is still playing.