Welcome to People vs Algorithms #81.
I look for patterns in media, business and culture. My POV is informed by 30 years of leadership in media and advertising businesses.
Sometimes it’s nice to read in the browser.
Algorithmic Sameness with Kyle Chayka. Podcast out now.
Media as automation
The media world is being automated. This is an uncomfortable idea because media is a complex nuanced human process of reasoning, taste and serendipity by which we collectively proces the world around us. Automation is cold and clinical.
In all things, automation pushes the humanness of something to a new plane. Framing media transformation as such seemed like an interesting idea to explore for a minute.
Automation is a steady evolutionary drip. A process of replacing or augmenting humans with machines, automation sticks when humans benefit from it. Naturally, we agonize over its broader consequences and beneficiaries. Often nostalgia powers a recursive and messy semi-linear process of past to future. But, we learn to absorb the change. The march of automation is inevitable. We are restless toolmakers.
Along the way, automation changes how an economy prices human enterprise. It always shifts where value is created.
Windmills harnessing nature to grind grain. Powered looms transforming textile production. Process automation revolutionizing car manufacturing. Chatbots automating customer service. We used to kill animals and cook them on fire outside of caves. Relentless automation means we now order Shake Shack burgers on phones. Though it’s still much better to eat them at Shake Shack.
Media is subject to the same automation imperative as any industrial process, though we may not see it this way. Privileged groups of people used to make media on paper and distribute it on donkeys. Now everybody makes content and algorithms deliver it in mysterious personalized segments inside of small pocket computers.
At the heart of modern media automation are “recommender systems,” machines that sort through trillions of signals each second to determine what we see. This is what we refer to when we say “the algorithm.”
The algorithm is the beating heart of media automation. It may have an agenda, but this is probably less true than we think it is. The algorithm's agenda is to optimize the matching of sellers and buyers inside of an attention ecosystem. People will always try to hack it because they always do these things. Algorithms will always have unintended consequences.
I used to think of algorithms as useful little tools to tell you what to read or watch next. Increasingly I think of them as a sort of global consciousness, the brain that stitches together an increasingly virtualized existence. The connection between you, and the next frame of your reality. These recommendation engines also power all of the other things that take your attention and money — the routes you travel, who you connect with, what products you buy, etc.
Your job is to understand what they are doing and live more harmoniously with them.
The automators
I started to think about the scale of media automation transformation reading about Meta’s 2022 decision to invest in AI capacity ahead of demand. Today that decision seems visionary.
The world freaked out the moment OpenAI released ChatGPT. It became clear that generative AI tech would change things and a mad scramble for the processors that power these experiences ensued. These new processors are called GPUs, like CPUs but with a “G”. You need them to do AI. Powerful GPUs, mostly from Nvidia, became the new gold. Meta had, fortuitously, reserved a whole bunch.
As it turns out, Reels, Meta’s TikTok rip off video feeding tube, now the default way we consume social content, is insanely compute intensive. The recommender engine at the core of these systems require tons of GPU power to give you that buttery media experience. The decision to invest in GPU cloud capacity to meet demand for Reels (including a sprawling ad system that pays for it), was a good one.
Nvidia’s CEO, Jensen Huang, highlighted the importance of recommender systems in an interview with Ben Thompson a couple of weeks ago. According to Huang, much of world would need new GPUs to power not just AI chatbots, but everything:
For example, we know that there are a trillion things on the Internet and the number things on the Internet is large and expanding incredibly fast, and yet we have this little tiny personal computer called a phone, how do we possibly figure out of the trillion things in the internet what we want to see on our little tiny phone? Well, there needs to be a filter in between, what people call the personalized internet, but basically an AI, a recommender system. A recommender that figures out based on the nature of the content, the characteristics of the content, the features of the content, based on your implicit and your explicit and implicit preferences, find a way through all of that to predict what you would like to see.
I mean, that’s a miracle! That’s really quite a miracle to be able to do that at scale for everything from movies and books and music and news and videos and you name it, products and things like that. To be able to predict what Ben would want to see, predict what you would want to click on, predict what is useful to you. I’m talking about things that are consumer oriented stuff, but in the future it’ll be predict what is the best financial strategy for you, predict what is the best medical therapy for you, predict what is the best health regimen for you, and what's the best vacation plan for you. All of these things are going to be possible with AI.
Media automation is essentially a matching function in a two sided information market. Today the supply of content is vast and fragmented. We are all sources of content. The demand side, you and me and our phones, is ever more hungry in fidelity (video, interactivity) and our evolutionary expectations as segments of one.
Naturally, the economic power in the new ecosystem is in interface ownership, and control of the algos and silicon that power an individualized journey.
To the platform owners, your wellbeing is not the product. The efficient automation of an attention marketplace is.
The automated
Should we be concerned about a media reality powered by an algorithmic feeding tube we don’t quite understand? New Yorker writer and internet smart guy, Kyle Chayka thinks we should be. We had Kyle on the People vs Algorithms podcast this week to discuss the conundrum.
In his new book, “Filterworld”, Kyle explains how our algorithmic-driven social media addiction has led to the “flattening of culture.” He describes a world in which algorithms increasingly dictate our experiences and shape our cultural tastes — the music we hear, the news we consume, the people we interact with, the places we go. The idea is that these algorithms act as filters, narrowing our choices, force feeding us lowest-common denominator crap. He laments a world in which creativity is suffocated by the need to feed an algorithm that determines whether or not someone will see your stuff. Many creative people are annoyed by this for good reason. Kyle offers an important critical perspective.
There’s another side, though. One that delights me everyday when I dive deep into YouTube to understand how, say, GPUs work, or the joy I get watching a heartwarming interspecies TikTok video of a monkey riding a dog. The satisfaction of reading the daily output of my new friend Emily Sundberg’s Substack, FeedMe. The modern internet might be God’s greatest gift to the restless mind.
Though a deeply romantic type might find it rewarding, it would be silly to grind wheat with stones to make a pizza crust. The world moved on long ago. Rewards are found in new parts of the culinary stack.
Media is no different. We can pine for the much less chaotic world of old, where a small number of gatekeepers channeled their observations about the world into their “professionally curated” packages. Or, we can choose to subject ourselves to the new endless algorithmic flow of the feed. One channels the thoughts of the few. The other, the engagement data of the masses. Both are easy to get lost in and, incomplete.
Everybody hurts, sometimes
You can choose to rise above.
Rise to question the invisible segmentation to which they are unwillingly subjected inside of the algo. Make the effort to step out of its stream, or to dig deeper in it. Seek out human sherpas with whom they can explore the endless frontiers the internet lays bare. Maybe engage ChatGPT in a voice conversation and see where things end up. Stroll over to Substack. Queue up one of a thousand beautiful podcasts.
These are journies I relish. I grew up with a crappy local paper, top 40 radio and the CBC. To me, modern information space is a wonderland.
Like everything else, media’s being automated. Nothing’s gonna change that. Our job now is to sit above the automation layer in new ways. Sadly, rising above the drone of the feed is not easy.
Navigating the modern panalopy is the part I struggle with most. Bouncing between cherished newsletters inside of my cluttered email box, a sprawl of open Chrome tabs, keeping up with old media while uncovering the new, surgical strikes on YouTube, nagging text threads, a list of unheard podcasts, on and on… my daily journey is chaotic and disorganized. Faced with information sprawl, I feel wildly inefficient.
I have not found a satisfying solution. Automation, and my response to it, has made considered consumption and the cultivation of my personal taste frustratingly difficult if not anxiety creating.
I asked a massively prolific Millennial newsletter writer to share her secrets of personal information processing. Her unhelpful response was to get a Yale intern, open more text conversations with smart industry insiders and down more caffeine.
People vs Algorithms. There’s always tension. Both can win…/ Troy
Bonus content: Death of the Follower & the Future of Creativity on the Web with Jack Conte | SXSW 2024 Keynote
This is a good companion video wherein Patreon founder and Pamplemousse guy, Jack Conte, offers a personal account of what happened to creators as the algos took over the web.