Multi-Disciplinary Reading – Book Reviews

How music got free, Stephen Witt, 2015 – This was a trip down memory lane for me as music is very dear to me and I have seen this transition of IRC channels and Winamp/xmms on PC/mp3 players to iPod to BitTorrent and ThePirateBay to Apple Music in my music listening journey. Having also worked on a mpeg decoder back in the day, the psychoacoustic bit and huffman coding and filter bank and AAC were nostalgia. The book also talks about the origin of piracy through two colourful characters Dockery and Glover working at Polygram plant that enabled almost all pirated music we enjoyed in the 2000s. Most interesting of course is the business transformation of the creation and distribution of music over the last 3 decades along with the plethora of value migration that happened.

My notes –

• Psychoacoustic – the scientific study of the way humans perceive sound. Psychoacoustic masking illusions show how the sound we hear is fiction (anatomical imperfections + our brain approximating it)

• Psychoacoustic experiments (by Zwicker) showed that it might be possible to record high-fidelity music with possibly small amounts of data, without losing much detail (compression). Most of the data on a compact disc could thus be discarded as our ears were already doing it.

• CD audio used 1.4 million bits to store a single second of stereo sound. Early pioneers of compression (Brandenburg/Seitzer) wanted to do it in 128k – 12:1 compression (128 kbps mp3 if you remember the 2000s)

• Zwicker’s 4 psychoacoustic tricks – 1. human hearing is best at certain range of pitch frequencies. you can assign fewer bits outside this range 2. Tones closer in pitch tended to cancel each other out (lower overrides the higher) 3. Auditory system cancels noise after a loud sound – like post a cymbal crash, fewer bits can be used 4. Auditory system also canceled out sounds prior to a loud click (weirdest) – this is because it took the ear few milliseconds to process and during that time it could be interrupted and drop what it was processing by a loud crash – so fewer bits before loud sounds

• You could do multiple passes using the 4 tricks above and each time audio will be compressed further but also progressively losing more information

• Huffman coding – on some tracks, there’s nowhere to hide (like a violin solo) and the psychoacoustic tricks dont work as well. So another algorithm was needed to compress repeating patterns using information theory. This is referred to as huffman codin. Combination of psychoacoustic tricks + huffman coding gives best compression

• Brandenburg’s compression could be used to stream and store music

• CD manufacturing – a digital master tape transported from studio under heavy security and cloned in a clean room and then replicated on virgin discs with bit-perfect copies

• The debut of MTV in 1981 marked the end of album-oriented rock and the resurgence of single-oriented pop

• NHL was the first to use mp3 compression for streaming. The codec was custom tuned for it to save thousands of dollars on satellite transmission costs

• Commercial usage of the codec for transmission was still a small market. The breakthrough for mp3 came when pentium chips were powerful enough to run the mp3 decoder (L3Enc) without stalling. Level 3 encoder (L3Enc) was distributed on the internet for free by Brandenburg to help its adoption (biggest turning point in music piracy)

• First mp3 player was conceived in 1995 with decoding chips that could be used for portable personal music consumption while PC was used for home consumption with floppy disks for distribution

• Brandenburg’s version of the codec without using MUSICAM’s filter bank (AAC codec) was adopted by Sony, AT&T and Dolby (there was a tussle with Philips’s mp2). The naming scheme of mp3 implied it was succesor of mp2 which was unintentional

• Piracy of music originated first on IRC (Internet Relay Chat) alongside other “warez”. The L3Enc was used to encode stolen CDs from manufacturing facilities of PolyGram. “Until it Sleeps” by Metallica was the first officially pirated mp3 (fondly remember hanging out on Undernet partaking in said deeds back in 2001 :-))

• Late 1990s the recording industry enjoyed record profits. Cost of production of a disc was less than a dollar and was retailing for $16.98. Economy was in boom and Americans were spending record sums on music

• Seagram purchased PolyGram records from Philips for $10b at the peak of the boom after selling Tropicana division to Pepsi. The home CD burner, broadband and piracy pretty much destroyed this capital allocation decision as freshmen in college filled their harddrives to the brim with pirated mp3s.

• Macromedia licensed mp3 for Flash player, Microsoft for Windows media player and WorldSpace for its satellite radio broadcast. The format just took off

• For studio guys and musicians sound was described in “tone” and “warmth” but for acoustic researcher it was a physical property of the universe. When acoustic researcher argued with a record producer, the debate wasn’t even in the same language

• Shawn Fanning’s Napster in ‘99 took piracy mainstream from IRC channels into more accessible territory for the layman. Within few months Napster had 20 million users and 14000 songs were being downloaded every minute. Napster wasn’t just a file-sharing service – it was an infinite digital jukebox, and it was free. It was thievery at an unprecedented scale

• Recoding industry sued to mp3 device makers and Napster for copyright infringement even as Napster peaked at 60 million users

• AOL-Time Warner merger is perhaps the stupidest transaction in history of organised capitalism. Time warner valued AOL at 200x earnings and got paid in inflated AOL stock

• The number of Limp Bizkit albums Universal could sell one year would vary a lot from the next year. It had to keep inventing new products unlike say orange juice which would sell consistently. But the back catalog, say of Led Zeppelin would sell consistently year after year and made up 30% of its revenue stream

• In RIAA vs Diamond lawsuit, RIAA won against Napster (shutdown in July 2001) but lost against Diamond (The mp3 player company). The music industry perhaps won the wrong lawsuit because the files that people already had pirated on their drives could still be played – and on better mp3 players – like the iPod!

• Windows media audio (.wma) was introduced by Microsoft to counter Fraunhofer’s mp3 format but it was too late (first mover advantage). Fraunhofer was paid millions in royalty from mp3 player manufactured and Microsoft for licensing mp3

• Steve Jobs pushed for adoption of AAC (the version of mp3 without the filter bank) so aggressively that people that Apple invested AAC

• Jobs wanted to buy Universal for its back catalog (after CD sales plunged 30% from its peak due to piracy) and wanted to sell songs for 99 cents apiece through iTunes store. Jobs wanted AAC to diminish the portability of existing base on pirated mp3s onto the iPod. But he ended up introducing iPod with both codecs and iPod acted like a money laundering system washing pirated mp3 to the legitimacy of the iPod

• Universal signed up with Apple at 70cents out of 99 cents per song in 2003. The iTunes store was an instant hit with sales of over 70 million songs in the first year

• Bram Cohen’s BitTorrent protocol enabled better peer-to-peer file sharing which was completely decentralised in sharing (but not the discovery, which is where sites like thepiratebay came in)

• Payola scandal – where DJs in radio stations were tempted with few hundred bucks of merchandise for creating demand for a song (or through “astraturfing” where people hired by the company would request songs creating artificial demand for it, making it sound like a “hit” song)

• Album oriented rock died in the 80s (victim of MTV and Walkman) and music became a hits-first business. A hit song and bunch of fillers made up albums and naturally people didn’t want to buy entire albums (making the case for iTunes)

• By 2007, CD sales had dropped 50% from their 2000s peak

• On Vevo, music videos saw resurgence after decades as it did initially with MTV. These become economic assets of their own in some cases earning more than even the complete albums did (became YouTube’s most popular channel)

• Fans who saved money on pirating music now spent more on concerts – many musicians earned more by touring than recording

• With the money made from Thriller, Jackson had snatched the Beatles catalog rights for $47m. Over next 25 yrs, this would appreciate 20x to be worth more than $1b

• Jobs wanted to start iTunes music label that would pay artists an unprecedented 50% royalty but with zero advance (Studios paid generous advances but smaller royalty). Apple took less risk and broadened the platform for publishing. In a world of digital abundance, it would become hard to earn a profit

• Spotify didn’t use mp3, it used Ogg (better quality open-source format)

• 2011 for first time, Americans spent more money on live music than recorded for first time since the phonograph. In 2012, sales of digital music surpassed CDs for the first time (value migration!)

• New Spotify customers and Vevo users (5 billion views a month and growing 50% in 2013) almost completely stopped pirating music and also stopped buying albums

Fundamentally lot of things have changed even since the book was written. Music has become very cheap with Spotify/Apple Music subscription but it has also eroded the quality of music as users get lot more choice and artists barely get anything. Music production as well has become cheaper with DACs and DAWs and home production setups and even worse, with AI. This also means the back catalogs of The Beatles, Led Zeppelin, Pink Floyd etc. are worth a lot more now. As we mindlessly listen to music chosen by the algorithm, the platforms prioritise what’s on bulk discount and also push AI generated music to the unsuspecting. Its one of those classic cases where capitalism and our own greed has destroyed art but I am digressing. The book is worth a read if you like music and technology and business. 9/10