Write us a check, we'll get your brand poppin' (WUACWGYBP)⁰

I love reading a picture-perfect introduction. But I have to be so real with you.. I dont know how imma start this joint.¹

On one hand, I'd love to hit yall with something ultra poetic about the process of writing this and everything that went into building the accompanying application, because I really WISH I could give you this feeling ². I wish that you could experience the same depth of excitement and anticipation that I felt while putting everything in this project together.

But, on the other hand... thematically-- I feel like it would also be very appropriate to harken back to some of the iconic openings of our time.. think Feel It In The Air³, when Beanie Sigel was opening up the album with lines like this:

• "my spidey senses is tingling!!!!" 🕷️
• "something going on!!" 👁️
• "my nose twitchin!!" 👃🏿
• "i still close my eyes, i still see visions!!!" 🔮

But nah, let me start with something else... let's see ⁴

Okay here's something else. The origin of WUACWGYBP as a title.

Tana Talk 3, arguably the best work in Benny the Butcher's catalog, closes with a track titled All 70,⁵ and yoo.. when the beat comes in... ✨whewwweeeee✨ it sounds like something I imagine should've been playing @ Woodstock '94.

Then, with the beautifully painted sonic landscape being filled out by the sampled guitar riffs (from "Final" by Bernard Callais and La Troupe Du Théâtre De La Porte St Martin)⁶ the opening adlib sets the tone:

🗣️ "If this don't wake the streets up, I don't know what will."

What we get next is a true gem of a verse. Reproduced and lightly annotated below for peak clarity & enhanced enjoyment. 👇🏿👇🏿👇🏿👇🏿

While that magnificent work of art did inspire the title for the project, the content and framing for the project was conceived much earlier. From deep within &computers' archival materials, we're able to see that the initial thoughts that spawned this work were actually first recorded @ 12:20 AM on 08/24/2018. So, yes, its been roughly 7 years and 4 months from the moment we first thought about this to today. Unclear if we should be embarrassed or proud of that??? 😭

Damn, how much time is he spending on the intro??? ⁸

So with all that being said... Let's talk about what this actually is.

This work is a meditation on the ways in which music reflects the zeitgeist of our time.. Or doesn’t. It's an exploration into music as memory and as mirror. A computational interrogation of a criminally under-appreciated literary canon. Does music shape our reality or simply reflect the happenings of the world around us?

Very concretely, this project is the introduction of a brand-new dataset- one that I hope will spark original analysis and lead to contribute important context to thought provoking discussions.

How, Sway?

For the past few [redacted unit of time], the status of this project has shuttled between passive ideation and spurts of maniacal coding that have sometimes yielded progress, albeit hal--ting--ly. So now, after [redacted unit of time] of plodding along, trying to coax data out of an ether of fragmented datasets, paywalled websites, and non-existent algorithms.. there's finally something to share! 🥹

Walk with me..

The output of all that effort is a net-new, never-before-seen, and outrageously-unique data set that combines previously disparate, disorganized, and simply unqueryable data. Namely:

• Every Billboard charting Rap song from 1989-2024 📈
• The artist that created the song 🧑🏾‍🎤👩🏽‍🎤🕺🏿🎤✍🏿
• The region that the artist is from 🌍 🌎
• Every weekly chart date that the song charted 📆
• The lyrics of the song 📝

After stitching each of these components together from various databases and websites, we leveraged a few open-source machine learning models and wrote a multi-step post-processing data pipeline to takes the lyrics of each song and extract the meaningful canonical named entities that appear in that set of lyrics. We then took the output of those computational processes and bolted them on to the previously mentioned components.

Details aside, the end result of the work is a fully populated database that looks something like below.

While the normalized database design and efficiency of keeping relational data in a flat file for storage is incredibly exciting– if you're a dweeb (🙋🏿‍♂️) – for everyone else, the significance of all of this is that having the data structured in this way allows us to investigate a myriad of previously unknowable mysteries.

For instance, exactly when did Nike began to take up more of our collective mindshare than Adidas? We can also answer questions like: are there any regions in which musical references to Spelman, Howard, and Morehouse largely overshadow references to Princeton, Harvard, and Yale? Does rap music have a consensus on the eternally raging Jordan vs. Kobe vs. LeBron debate? How do the legacies of Obama, Clinton, and Reagan show up in our art? Is your city on the map? word?.. since when??

The following section is a smattering of explorations and thought-provoking observations based on some of the entities that are present in our dataset.

Checks Over Stripes: A Comparative Analysis of Brand Equity

Honestly, I blame Pusha-T for this, 100%. The Story of Adidon came out in 2018. & Adidas ain't been the same since! Looks causal to me 😌 (sarcasm)

HIGHER LEARNING

I know at least 1 person who has really strong opinions about that time Drake said "Sound so smart like you graduated college, like you went to Yale, but you probably went to Howard knowing you" – sir, what do you mean by that? 👁️

Nas has this line in Book of Rhymes where he says "My people be projects or jail never Harvard or Yale". That song didn't chart, so it's not represented in the data set, but I bet theres some interesting analysis about what those institutions represent to people in the culture and why. And perhaps more interestingly, how does that perception interact with what certain HBCUs represent?

Rap Really Loves Luxury

The luxury conglomerate LVMH has many more brands under it than just the 3 visualized in the chart, but we still thought it interesting to just see the eponymous ones. An open question that I'd love to hear an answer on from anyone– are there any other comparable conglomerates that would be interesting to see aggregated data for?

My presidents are scandalous (pejorative)

This is unexpected. I thought Obama was going to be far and away the most popular on an absolute scale. But Clinton had a generational run. I think his particular scandal was a really rich lyrical playground from which to draw references and I believe that's what is reflected in the data.

Shawn Carter v. Cristal (2006)

The well-documented beef between Jay-Z/Hip-Hop and Cristal took place in 2006. This data shows a gradual drop in the number of songs that mention it after 2006 which perhaps is to be expected after Jay called for a "boycott" of the brand. However, there's also a more dramatic drop after 2001, so it's hard to say if its just noise or a reflection of people buying into the anti-Cristal movement. It would be really interesting to see the breakdowns by specific artists (coming soon!).

You Can Explore The Data Directly.. Right Now!

The charts above were just some examples of initial questions and thoughts we had when we finally got our hands on the data. We've released the application that made all of these charts on the web for anyone to play around with and explore. Please give it a spin and reach out if you see anything interesting!

If you happen to use this for a project / publication, please cite:

@misc{juicecountyprodigy-write-us-a-check-well-get-your-brand-poppin,
  title        = {Write Us a Check We'll Get Your Brand Poppin'},
  author       = {Juice County Prodigy},
  year         = {2025},
  month        = {December},
  howpublished = {\url{https://andcomputers.i/write-us-a-check-well-get-your-brand-poppin}},
  note         = {Blog post}
}

Acknowledging Limitations

The initial goal of this project was to create a dataset that was a representative slice of named entities & their frequency of occurrences in rap music across time and space. This closing section is a non-exhaustive list of different sources of potential errors and biases that we know exist within the dataset. We hope to get feedback from everyone and continue improving our data cleaning processes and data sources.

General Complexity with Language & Computation

Some references were subtle and not detectable by the natural language processing (NLP) techniques used whether because of spelling, context, or model quality/domain fit.

Most entities in our dataset could be referred to in many different ways, whether that was by initials, regional slang, nicknames, alternative titles, etc. This was a known challenge going into the project and we designed for it by creating a multi-step pipeline for deduplication to ensure that entities like bron bron and lebron were combined and similarly that things like bape and bathing apes mapped to the same entity. While we tried our best to be thorough, the process is not perfect so there are likely a number of misses. Some types of misses that we anticipate you may encounter while exploring the data:

• The same entity name, referring to different entities. For instance Daytons (rims) vs Dayton, Ohio (city) or Apple (company) and Apple Bottom Jeans (brand)
  • While we were able to detect many of these types of overlaps by using context-aware entity recognition with part-of-speech tagging, some things undoubtedly slipped though the cracks.
• Different entity names that are referring to the same entity. For instance Eminem & Marshall Mathers both refer to the same underlying person. While we had steps to find the obvious examples of this, the dataset is large and so similarly, things may have fallen through the cracks. As another example, when we processed Drake's verse in in SICKO MODE, the line "checks over stripes" did not count trigger an increment in the number of occurrences for Nike and Adidas.
  • A subset of this issue might be names changing over time, i.e. Zaire to Democratic Republic of Congo.

Billboard Shenanigans

The dataset of songs is based on tracks that appeared in the Hot Rap Billboard 25 Chart, which is based on radio plays, streaming, and album purchases. Even though these sources cover many of the places people hear music today, there have been plenty of songs that were very popular, but never charted. This rings especially true for earlier periods of time in which digital music was not available. I've got no doubt that there are plenty of cuts that went quadruple platinum on burned CDs and bootleg cassettes, but don't appear on the Billboard charts.

Lastly, Billboard has a lightly controversial track record with how they categorize music, with some people contending that they view all Black music as Rap or R&B, even when it's not 🙄. As a result, the inclusion/exclusion of songs in this analysis is completely driven by the decision-making process of Billboard (for now).

Inconsistency, Accuracy, & Availability of Lyrics

The lyrics data we relied on for this project is varied and was sourced from a number of different places. Some of the lyrics that we obtained were found to have small inaccuracies. Some lyrics (even from the same datasource) might include verbatim adlibs, while others just completely omit them. Some lyrics had typos in the transcript or ended early. For a very small number of songs that appeared on the Billboard chart, we weren't able to find any lyrics at all (most common with the earlier periods, such as the songs that charted in 1989). In a few very rare cases our sources contained the wrong lyrics for a given song. In the cases where we found that to be the case, we removed the lyrics and unattached any entities that we had previously counted in the data set.

Entity Attribution Granularity for Multi-Artist Songs

For songs that have multiple artists associated with them as a result of features or co-credits or production every entity mentioned in the song gets attributed as a mention to every artists on that song regardless of the specific person whose verse the entity appears in. So even though Drake mentioned Howard and Yale in his verse on Make Me Proud, both him and Nicki Minaj will have those entities attributed to them in the dataset since they are both listed as artists for that song.

Conclusion

Sharing this work has been really fun and I'm excited for people to engage with it and share it with others who might find it interesting. If you aren't already subscribed please do, as there are a number of follow ups on this that I would love to share.

In the meantime please comment any thoughts! Or text me if we friends ☎️ Or both preferably.

Some things I'd love to hear:

• Is there anything you'd like to see in the future within the dataset?
• Suggestions for improvements after trying out the app?
• Thoughts about what might be interesting analysis to conduct with the data?
• Other form factors?
• Unacknowledged issues or biases that you see with the approach?

Anyways. There will be more, so stay tuned.

& as always, thanks for reading.

Footnotes

⁰ Almost had this alternatively titled as "I Like Girls That Wear Abercrombie & Fitch". Maybe you weren't hip, but it was important to the culture (🍊juice county shit!)

¹ Track 3 on Illmatic, NY State of Mind. Nas starts his verse with a non-chalant "I dont know how to start this shit.." then proceeds to record a legendary verse. A real moment.

² On the Corner, brothers robbin, killin, dyin, just to make a livin. "I wish I could give you this feeling" is really one of my favorite sentiments ever expressed. Applicable to many situations.

³ Nah, no words.. you just gotta listen to this one. The sample goes crazy.

⁴ Nas in Book of Rhymes tries 2 or 3 different introductions.. just a few bars for each one, but then just as he gets going, he stops and starts with a new one.. until he doesn't. A classic deep cut-- went triple platinum in this household!

⁵ Tana Talk 3 is maybe the best thing Benny has put out in the last decade. Langfield, 97 Hov, Intro: Babs, Rubber Bands & Weight, Scarface vs Sosa 2... its just.. top to bottom.. maybeee no skips. smh. Anyways, while I love the recorded studio version, this youtube clip of Griselda performing at SOBs in New York just hits different. "I know you feel it, but its realer in person!!" Plus extra points for the youtube video because Westside Gunn starts with Summer Slam 88... what are we talkin bout, hit play.

⁶ How did they even find this joint??? Daringer a real sicko (positive)!!

⁷ Josh Allen engaging in America's favorite pastime.

⁸ I hate that buddy has so many hits and like 4 different mentions in this article. But some shit is undeniable fr. Tuscan leather is one of those ones! 3 flips of the same sample on one track. sheeeesh.