Black Friday Musings: Internet’s Scale + 5 Good Reads

Today is Black Friday, and while many associate it with shopping, for me it’s a day to wake up early and play with my inks, swatching them and deciding which one to use for the next few months. Of course, I don’t expect you to do the same.

Instead, I imagine you’ll be doing more typical activities, like browsing the internet. Of course, you can do your shopping online — and put the rest of your time to a different use. Perhaps you’ll read, listen to, or watch something enjoyable to kick off the weekend. (My 5 (including non-tech) reading recommendations at the end of the post.)

For example, you might be curious about why Apple’s USB-C cable is priced at $130, compared to $3 alternatives on Amazon. I also wanted to understand this, so I watched an entire video on the topic. It turns out, Apple’s pricing isn’t just about charging a premium for their brand—though that’s part of it. There’s actually much more to their cables! Watch

The reality of our modern world is heavily influenced by speeds and feeds. Today, it’s essential to have not only fast USB-C cables but also quicker internet and mobile connections. As someone who began using the internet with a 1200 baud dial-up modem, I find it astonishing to see my mobile phone connecting at speeds over 200 Mbps.

It has taken almost two decades, but we are finally witnessing substantial improvements in internet speeds across America. According to a recent report by HighSpeedInternet.com, the national average internet speed in 2023 is 171.30 Mbps, a 44% increase from 2022’s average of 119.03 Mbps.

Speaking of speed, Oak Ridge National Laboratory’s Frontier remains the fastest supercomputer on the TOP500 list. This exascale machine boasts an HPL score of 1.194 EFlops/second. Notably, Microsoft’s Azure Cloud’s Eagle, which also supports OpenAI, ranks third and is the highest-placed cloud system. This achievement is significant. It’s amazing and forms the basis of their major leap into the next generation of computing.

One of the main reasons Microsoft can push the envelope is money. Thanks to its huge software margins, similar to those of Apple, Google, and Amazon, the company has begun making its own custom silicon to meet its specific needs.

Just before ‘OpenAI: The Reality Show’ grabbed our attention, Microsoft announced two new custom chips for its data centers: Maia, an accelerator chip optimized for AI tasks and generative AI, intended for the Azure hardware stack, and Cobalt, an ARM-based processor designed for more general-purpose compute workloads in the Microsoft Cloud.

If I look at the sophistication with which the margin structure of the cloud business is going to change, it’s going to be around how smart are you with silicon.”

Satya Nadella

We are using the cloud for everything these days, though primarily for what could be described as “entertainment” or “edutainment” purposes. That should explain some of the bandwidth consumption trends. Openvault recently reported the following data: (link)

• • In 3Q23, the average monthly data usage increased to 550.2 GB, marking an 11% rise from 495.5 GB in 3Q22.
• • The proportion of users consuming 1 TB or more monthly climbed to 16.4%, a 19.7% year-over-year growth.
• Additionally, the segment of users consuming 2 TB or more grew by 35%, reaching 2.9%.
• • Subscribers on gigabit speed tiers reached a record 32.1% in 3Q23, more than double the 15.3% in 3Q22.
• • The share of subscribers on plans of 100 Mbps or slower fell to 10.1%, a 22% decrease since 3Q22.

These faster networks only accelerate how we use the bandwidth and what we do with it.

If you’ve been a regular reader for the past two decades, you’ve likely heard me say (and write) repeatedly: the network reinvents behaviors and usage patterns native to it. We’ve become so accustomed to on-demand video that it’s now the primary internet activity. Goodbye, linear television. Streaming music and subsequent programming through machine learning have relegated the concept of albums to the rearview mirror.

Radio is undergoing a similar transformation, evolving into podcasts. We’re spending more time online listening to spoken word content rather than traditional radio. One of the more interesting pieces of news I came across this past week was…

“In 2014, 20% of all audio time was spent listening to spoken word audio among people in the U.S. 13+. This has now increased to 31% of all audio time in 2023. Not only are Americans 13+ spending more time listening to spoken word audio, but this growth in time spent listening is being sparked by the 13-34 age segment.”

NPR + Edison Research’s fifth Spoken Word Audio Report

So essentially podcasting!!!

While the on-demand culture — encompassing cars, food, movies, shows, music, and media — suggests a desire for control over our time and destiny, our actions as humans often tell a different story. We habitually hand over decision-making to others. Despite widespread skepticism towards AI, we willingly relinquish our data, control, and intelligence to systems, whether these systems require it or not.

Recently, there was a trend on Meta’s Twitter clone, Threads, where users shared extensive personal information, including their likes and favorites, in the hope that the algorithm would provide customized content. Eugene Wei, one of my favorite technology essayists, astutely observed —

This trend was amusing but also revealing. People are beholden to the whims of blackbox algorithms to be heard online, to exist, and so they leave prayers and sacrifices to their new algorithmic gods like supplicants.

It won’t be the first time, nor will it be the last — we humans have always had a very complicated relationship with technology.

---

My 5 (long) Reading Recommendations:

• How an Indian Education Software Company Became a Lynchpin of a Cyberespionage Hacking Operation” – This Reuters investigation into Appin is a must-read.
• Against Scale” by Claire Evans is an insightful piece on growth and scalability, comparing natural processes with digital representation. Evans writes, “Here is the difference between growth and scalability: on one hand, an unruly seasonal explosion of flowers; on the other, the algorithmically-boosted presentation of those same flowers as a product, a highly-reproducible image driving traffic both online and onto California State Route 14. Link
• The article on meditation explores the unclear science behind how much the brain affects the body. As someone who practices meditation and traditional breathing exercises (Pranayama), I find this topic intriguing. Though I feel better from these practices, the scientific explanation remains elusive. Link
• What Happens When Salmon Vanish? This article focuses on the Yukon River community, directly affected by the disappearance of salmon. Link
• In “My North Star for the Future of AI,” scientist Fei Fei Li, a leading authority on AI, presents persuasive arguments about balancing technological progress with humanity’s needs. Link

Bonus read:

In a unique piece for Spin magazine, Richard Pryor’s ex-wife interviews jazz legend Miles Davis. The article is less about the content and more about the intriguing interplay between the interviewer and the subject. It raises the question: could generative AI transform this into a compelling visual conversation? Link

November 24, 2023. San Francisco