On the state of Indian education, labour regulations and corruption in India, Moxie Marlinspike at Signal and privacy, on the audio industry & future of podcasting and concerts, and AI hardware in India.

Interesting Links: November 1, 2020

“Moxie leads from the front, and he just leads by doing. One of his favorite quotes is ‘The only secret is to begin.’ If you want to get good at something or do something, you just do it, and you figure it out along the way.”

New Yorker, Taking Back Your Privacy (on Moxie Marlinspike)

1. ASER published their Annual State of Education Report 2020 Wave 1. Lots of media coverage, including Times of India, Indian Express (and also here), and the Hindu for example.

   In this pandemic year, the survey was calls to about 118k households, of which 58% responded, and of which 75% were surveyed. So in all about 52k households provided data (or about a 44% response rate). The survey covered 26 states and 4 union territories, with over 59k children ages 5-16 surveyed at about 9k schools. Enrolment in government schools increased over the past 2 years with a corresponding decline in private school enrolment.

2. Interesting paper on labour regulations and the cost of corruption in India. The authors research a suite of labour regulations in India that only apply to firms above a size threshold (in particular, with 10 or more workers). They estimate that “the regulations increase firms’ unit labour costs by 35%.” These regulations include workplace safety regulations, social security taxes, and business registration requirements. However, this 35% figure varies considerably across the states they studied. It ranges from a low of 14% in Kerala and 16.5% in Gujarat to a high of 69% in Bihar.

3. Nice New Yorker profile by Anna Wiener of Moxie Marlinspike, the founder of Signal. In a world where an increasing number of youth appear to be indifferent to privacy, this was a good read (well, ignoring the fact that uBlock Origin blocked 23 trackers when loading this New Yorker web page; so much for privacy….).

   It notes how he spent much of his youth immersed in anarchist literature and communities. Recently, we saw how Audrey Tang, Taiwan’s digital minister, was also heavily influenced by Japanese anarchism. Unsurprising, but interesting.

   “If I’m dissatisfied with this world—and I think that I might be—a problem is that you can only desire based on what you know,” Marlinspike said. “You have certain experiences in this world, they produce certain desires, those desires reproduce the world. Our reality today just keeps reproducing itself. If you can create different experiences that manifest different desires, then it’s possible that those will lead to the production of different worlds.”

4. Matthew Ball writes a great post on the audio industry.

   He shows how the fraction of Billboard 100 songs less than 2:30 in length went from roughly 1% until 2016 (so, 1 song, maybe 2), to about 12% by 2020, and how in 2018, only around 17% of Billboard charting tracks had one word titles to about 50% in 2019.

   He writes about the dramatic rise of video games, but crucially how it’s been additive as new technologies were launched. Consoles added revenue on top of arcades, and have continued doing so for over 35 years. PC gaming has been additive over console gaming rather than replacing it, growing considerably from the early 2000s to present. And similarly since the mid 2000s, mobile gaming has also been additive rather than taking revenue away from console or PC gaming.

   There is lots more there, including on the future for podcasting and live concerts. He has a fascinating chart showing that in the big increase in concert revenues over the past 20 years, the bulk of the increase in revenue has accrued to ‘non-top 100 tours’ rather than top 100 tours.

5. Finshots writes about computing hardware in India, esp. for AI. It discusses India’s AIRAWAT initiative where the government plans to build a series of supercomputers with the help of NVIDIA and Intel & where the supercomputers will be hosted at various academic institutes like the IITs, IISc, etc. The blog post is extremely light on any details, so check out the NITI Aayog working paper on AIRAWAT (pdf).

   Not surprising, but the report identifies some of the problem areas with respect to India’s supercomputers - India has very few supercomputers, its best supercomputers are considerably less powerful than the ‘best in the world’, and the existing infrastructure does not lend itself to be upgraded for AI workloads (e.g., are designed for specific purposes like weather modelling) and are already running at full capacity.