A eulogy for What.cd, the greatest music collection in the history of the world—until it vanished

The end of an era.
The end of an era.
Image: What.cd
By
We may earn a commission from links on this page.

In 2007, I became obsessed with a musician named Lizzy Mercier Descloux, a nearly forgotten French eccentric who created a singular style of punk-rock deeply influenced by African music and funk. I read incessantly about her music, but I could not find a way to actually listen to it. Amazon’s catalog was limited, so I couldn’t buy a CD. Spotify didn’t exist yet. And even illicit file-sharing sites came up dry on searches for Lizzy.

So I decided to dive into a deep corner of the internet, attempting to gain access to a mythical website that was said to have all the music you’d ever heard of, and all the music you hadn’t. It was called What.cd, and it promised a lot. Its Twitter profile used to read, “Beyond here is something like a utopia—beyond here is What.CD.”

It lived up to this utopian promise, until the site was shut down on Nov. 17 after a raid by French authorities. It now reads:

Due to some recent events, What.CD is shutting down. We are not likely to return any time soon in our current form. All site and user data has been destroyed.

This is a tragedy for music fans, and for humanity: What.CD was an unprecedented cultural repository. It offered something close to the sum-total of humanity’s recorded musical output, organized and classified to near-perfection.

It was a musical Library of Alexandria, and now has suffered the same utter destruction as its ancient predecessor.

What exactly was What.cd?

In file-sharing jargon, it was a “torrent tracker,” which meant that the site did not actually host audio files itself. And not just anybody could join the community. (More on that below.)

As a tracker, What.cd provided downloadable “.torrent” files that connected users who already had a set of files—say a rare bootleg mp3—with other users who wanted to download them. That’s the essence of peer-to-peer file sharing. The role of the site was facilitating the sharing of music, and making music discoverable.

It also developed a rigorous self-policing culture, enforcing its draconian rules on audio quality, and aggressively removing users who did not follow these rules. In the end, it lasted for nearly ten years.

Beyond Spotify

The scope of music collected on What.cd was almost incomprehensibly vast: More than a million distinct “releases” of songs, albums, and bootlegs.

Bach cello suites. Obscure Chinese indie rock. Nigerian hip-hop. Thai psych-funk from the 1970s. Every release of Miles Davis’ Kind of Blue, including vinyl rips and remasters. UK techno tracks that were pressed on vinyl in the 90s, with only a few hundred copies made, and uploaded by dedicated crate-diggers.

The collections of Spotify and Apple Music may seem infinite, but What.cd had thousands of albums that were not available anywhere else—and now, are not available anywhere at all. The site had about 800,000 artists as of early 2016.

“I did sound design for a show about Ceaușescu’s Romania, and was able to pull together all of this 70s dissident prog-rock and stuff that has never been released on CD, let alone outside of Romania,” one former What user told me.

Perhaps most importantly, all of this music was organized by a large community collectively obsessed with musical metadata. Album, track, and artist names were meticulously edited, organized, and collated.

A What.cd artist page.
A What.cd artist page.
Image: What.cd via The Verge

In your iTunes (if you still use iTunes), you might have some files by Beyoncé (accent on the e) and others by Beyonce (no accent). That kind of error was swiftly eradicated on What. There were deep taxonomic debates in the forums about whether some newfangled genre deserved its own tag, or whether adding it would be confusing.

This problem of tagging is distinct to a vanishing era of digital audio files. Once upon a time, people owned physical media: records, tapes, CDs. Then Napster brought us into the age of digital. Listeners stored mp3s (some legal, some not) on their hard drives, transferred them to iPods, burned them onto CDs. What.cd was the last great institution of that era.

Now, in the age of Spotify, YouTube, and Apple Music, having mp3s on your computer seems as quaint as owning 8-tracks in 2007.

Invite-only

Gaining access to utopia is never easy, and What was no exception: New users could only gain admission through a referral from an existing user, or via a famously difficult interview process. There are endless threads online with people fretting about interviews, discussing failed interviews, and ranting about the excessively demanding process.

In 2007, I studied for days before submitting an interview application on What’s IRC channel. I waited several hours, staring impatiently at my computer, until a What moderator got back to me. I was then subjected to a 20-minute grilling, in which the moderator lodged obscure audio encoding questions at me: What is the difference between constant bitrate and variable bitrate mp3s? Look at these two spectrograms—which of them belongs to an mp3 that was encoded poorly? How can you tell? If you encode a lossless file at 320 kilobytes per second, and then again at 256 kilobytes per second, is this the same as encoding at 256 directly? And more.

Reader, I passed the test.

Once in, users were subjected to a complicated system of incentives to make sure everyone contributed as much as they consumed.

If you downloaded several albums of music upon arrival, without uploading files to other users, you were banned. If your download-to-upload ratio fell out of whack, even after years of contributions, your download privileges would be taken away. If you uploaded an album that was not encoded to the site’s strict standards, you’d get a warning. If you absentmindedly mislabeled a Justin Bieber album as a Taylor Swift album, you might be banned.

Nobody whom I invited managed to stay very long.

These policies may sound excessive, but they facilitated what was arguably the world’s greatest crowdsourcing effort, maintained by a community of dedicated, diverse, and knowledgeable music lovers. A team of volunteer coders created a custom CMS for the site, called Gazelle, that harnessed all of this musical knowledge.

“Collages” were one of What’s best features. Users arranged lists of albums on the site into useful categories like “Intro to free jazz” or “Bands with a male and female singer.” These were indispensable sources of musical discovery.

A collage on the site What.cd
A What.cd collage page
Image: What.cd via Gizmodo

There were also the magical “staff picks.” These select albums were offered, for a short time, at no download cost, encouraging you to listen to music you never would have otherwise. Then there were the highly active forums, where music nerds of all stripes shared obscure recommendations and reviews.

Outside the law

Of course, What.cd was illegal from the beginning. (Most of it, anyway—many users were musicians who willingly added their own music to the What.cd archive.) The entire enterprise was a rejection of the notion of music copyrights, and a middle finger to the music industry’s army of lawyers.

The strict policing of users—plus the site’s low profile, and the fact that it did not actually host audio files—allowed What to deny culpability and exist in a questionable legal grey area. But only for so long.

There’s an important caveat to this issue of legality, though: The site offered much that is unavailable via legal channels, even to those willing to pay. There were the albums that weren’t available anywhere else.

Plus, most of the albums on What were available in pristine lossless formats, meaning that no audio quality had been lost in the transfer from the original recording to your computer. This may seem like audiophile snobbery, since most people can’t tell the difference between even low-quality mp3s and lossless files. But lossless files are important for other reasons, like archiving. Right now, mp3 is the dominant audio format. If mp3 gets superseded by something better, though, there is no way to convert your old mp3s to a new format without losing further audio quality. And if you have to convert again, even more is lost. With a lossless format, there is always an unvarnished original.

There is also the matter of ownership. Spotify is ubiquitous, but it is a tech company, and tech companies fail. If Spotify goes away after you’ve spent hundreds of dollars on an account for multiple years, you are left with nothing but memories.

What’s more, most streaming services use proprietary formats that make it nearly impossible, for example, to turn a Spotify playlist into an Apple Music playlist. With open formats, like mp3, FLAC, and m3u for playlists, users control everything.

Yet there is no major digital music marketplace that will even sell you open, lossless files. Itunes files are “lossy,” meaning they have already lost quality from the original recording. Same for Amazon and Google Play. Some sites, like HDTracks, offer downloads of lossless files, but the selection is limited. (There is no Lizzy, for example, a deal-breaker for me.) This means that the only way to get most albums into a lossless format that you can control is to literally buy a physical CD and rip it yourself. Or, until today, you could use What.cd.

What.cd's user-generated "Similar Artist Map."
Image: What.cd

The end of What is the end of the most complete musical database ever created by humanity. It opened my ears to sounds I could not have imagined nine years ago, at the time of my quest for Lizzy Mercier Descloux. On What, I encountered so many life-changing musical favorites, old and new: the Japanese avant-reggae band Fishmans, the Panocha Quartet’s renditions of Dvořák, Carly Rae Jepsen’s Emotion, the Blade Runner soundtrack, Swiss funk-house producer Kalabrese, and lots of weird jazz, to name very few.

For me, the end of What is a final reiteration of the ultimate lesson of 2016: Take nothing for granted. No set of values, no repository of human knowledge, no political system, no passionate community, is guaranteed safety from outside forces and from the passage of time. Nothing is too great to fail.

What now says that “all site and user data has been destroyed.” So a decade of meticulous collection and curation work has vanished into the ether. It is not unlike the whole of Wikipedia disappearing overnight.

But there is still some hope. All of the rare and unreleased music that What organized still exists somewhere, on people’s computers. And What users were usually paranoid about losing their collections—I recall reading that one user stored his on an external hard drive kept in a fire- and water-proof safe in his basement.

The treasure is still out there—we have merely lost the map.