Hi everyone, and happy Friday once again. Let’s get to it.
Generative Media, Mr. Scott’s metaverse, and more
It’s difficult to say whether Travis Scott or Reliance Jio has received more attention in this newsletter over the past few weeks. I’m also willing to bet that those two don’t often appear in the same sentence… Nonetheless, in case you’re curious to see the actual footage of Travis’s “concert” in Fortnite--which has now inspired much deep reflection from seemingly every corner--you can do so below.
I’m reminded of the similarities between this activation and 2Pac’s hologram performance at Coachella in 2012, and not just because of the similar outfits. That show inspired a lot of commentary about who controls an artist’s legacy, and whether we can truly appreciate a hologram’s performance the way we would an actual human’s. Some of the representational issues are different, though. For one thing, Travis Scott was intimately involved in the choreography and sequencing of his avatar’s performance. Other “virtual” artists--Gorillaz, Lil Miquela--perform what their creators program. 2Pac, long dead, did not decide to perform at Coachella, nor did he choose the way he appeared, the songs he performed, and so forth. While much of that hologram performance was ostensibly inspired by live footage of 2Pac--specifically his legendary House of Blues show--it strikes me as a somewhat different question of creative expression (and ownership).
Pitchfork's Marc Hogan has a great piece on JAY-Z’s copyright dispute over the use of his voice to power several audio deepfakes. I recommend reading the entire article, and checking out some of the examples they cite, particularly George W. Bush doing “In Da Club.” There are interesting questions at play around ownership (of one’s voice), and what constitutes original or derivative content. If I train a speech model on JAY-Z’s voice, but then I use it to recite original poetry that I’ve written, should that be allowed? What if I trained the model on JAY-Z’s voice and Kanye’s voice? Now it’s a synthesis of multiple human voices, which feels somewhat different in terms of ownership and infringement. What’s the threshold where we feel like a “robo-voice” is a sufficiently abstracted composite? I think each of us would likely say that something feels “off” about using someone else’s voice to perform something that they didn’t agree to perform, yet we tend not to have issues with comedians impersonating others (Caliendo-as-Madden). Perhaps that’s because we can clearly separate the performer from the object of imitation, and are clear that reality is being suspended, but all of this seems to be different points along the same spectrum.
Ryan Walsh wrote a piece on Stereogum about OpenAI Jukebox, which I referenced last week. It’s long but has a number of interesting points within, including a version of an idea I mentioned (to deafening silence) a few weeks back about “auto-scoring” podcasts and audiobooks. Ultimately, one could argue that imitation in music is far from a new phenomenon, as countless artists have built careers out of sounding like others (witness every rapper trying to sound like Young Thug in 2016-2018 or every rock band trying to sound like The Strokes in 2001-2005). For the moment, I’m personally less concerned with these “uncanny valley” moments created by AI musicians, but more focused on what type of economic reality this could portend for human creators in the future.
Luckin Out (!!!)
For some reason, I continue to find competition between coffee companies oddly captivating. Luckin is a Chinese coffee chain that experienced incredible growth over the past ~3.5 years since its founding, aggressively winning share from regional players and multinationals like Starbucks through a model built around a mobile app, order-ahead dynamics, incredibly efficient stores, deep discounting and incentives, and apparently fraudulent accounting. Don’t think that last one is part of the official playbook, but recent news reveals a “financial scandal” around over-reported sales (~$300M+). Both CEO and COO have been fired, but this feels like a more pervasive cultural problem and a likely outcome of the “growth at all costs” mentality that has become the norm in recent years as investors have poured unprecedented amounts of capital into companies, inflated valuations, and demanded user numbers that rise to the level of their investments. When the growth proves to be unsustainable--whether causing fraud or simply poor user economics and overextended operations--we get news stories like this one.
Meanwhile, Kopi Kenangan, a large Indonesian coffee player, has announced $109M in new funding from investors including Sequoia (and a totally random assortment of others including NBA player Caris Levert). I am personally quite partial to Kopi Kenangan, despite the industrial quantities of condensed milk they use and the paranoia their espresso tends to inspire in me when combined with the climate in Jakarta.
Jio’s Scrooge McDuck bathtub of money
Another week, another set of investors targeting Jio, which has now raised >$8bn in the last several weeks. This time, it’s General Atlantic ($850-950M target) and the Saudis, via their $320bn sovereign wealth fund. Much of this activity and attention seems to have been catalyzed by the Facebook/Jio partnership, and the knock-on effects of this relationship on mobile payments, digital commerce, and perception of internet connectivity (and smartphones) as crucial utilities across a broader segment of the Indian population. I would not be surprised if there are a number of imminent investments in the services layer, from entertainment to food delivery and beyond. This type of investment has been somewhat fraught in the past (e.g., Oyo), as valuations aren’t necessarily aligned with true market opportunities and unit economics often fall short of the mark. Now, with Covid’s impact on businesses like Zomato, a darling of the India startup scene a mere few months back, I suspect it will continue to be a complicated landscape for investors to navigate.
Digital payments continues to be an area of promise and increasing consumer engagement. Whatsapp is preparing to expand its payments offering through 3 banks, utilizing India’s UPI (unified payments interface) rails. In other emerging tech markets, like Africa, there is a similar move to mobile and digital payments. Africa has long been seen as an early adopter of mobile payments like M-Pesa, as many individuals simply skipped bank accounts & credit/debit cards. Now, the threat of virus transmission seems to be accelerating mobile (and, in particular, contactless) payments in these markets and in the US/EU markets.
Souncloud “Verzuz” IG
Check out this interesting Music x Tech x Future piece on Soundcloud, Instagram, and their respective visions of how to build for musicians and listeners. It traces the “tech products shaping the music industry” trend back to MySpace, which briefly happened upon music as a viable vertical even after Facebook emerged. I loved Soundcloud in the early days but have been disappointed by how unwieldy the app has gotten for listeners. It’s filled with interstitials and sign-up flows, while the discovery & playlisting tools are less intuitive than in early generations of the product. I still find some of the most interesting new artists on Soundcloud, but have to overcome a lot of friction to do so. Now, as IG and other social platforms enable artists to leverage their existing followings for live events and other promotional activities, it feels like Soundcloud needs to rediscover its differentiation.
No comment
Disclaimer is that I haven’t tried Quibi, the heavily-funded short-form mobile video product from Jeffrey Katzenberg and Meg Whitman. “I attribute everything that has gone wrong to coronavirus,” Katzenberg tells the New York Times. “Everything. But we own it.”
Visualizing AI
Take a few minutes and check out the Google blog showcasing interactive visualizations of AI applications. Fairness in AI has become an increasingly popular topic in recent years, and rightfully so. These Explorables provide a more intelligible way to think about some of the underlying issues.
The ambient album you never knew you needed
Luke Schneider just released an album called Altar of Harmony. Schneider is a virtuosic pedal steel guitar player; pedal steel is a type of guitar that uses a variety of levers (and pedals, shocker) to modulate the sound and tuning of the instrument. You play it with a metal slide, which lends a fluid and dreamy sound that you’ll recognize from modernized Hawaiian music and a lot of American country. Turns out it also makes for a really good ambient creation tool. This album is more interesting than background music, and is worth a listen. My dog hates it with an admirable passion, and stares at the speakers with a quietly intense fury every time I play it.
See you all next week.
#