At the beginning of the year, I came across an article written for New York Magazine’s The Cut titled “In 2026, We Are Friction-Maxxing”. Despite being published as a parenting article, the key premise of “Friction-Maxxing” seemed to catch on with people.
Why it resonates.
The author suggests that much of life has been targeted by tech companies intending to distract, provide convenience, and otherwise give us what they’ve learned we want – and in doing so, it’s had a consequence of making life without their products feel inconvenient: “Reading is boring; talking is awkward; moving is tiring; leaving the house is daunting. Thinking is hard.”
This concept resonated with me. I’ve been a bit of an AI luddite in the past, primarily because I share some of the same concerns as the article’s author. Why would I want this technology to do things I’m otherwise capable of doing or learning? Is there a danger in offloading that effort to a product being sold to me?
However, I know in practice it’s not so easy to avoid the situation entirely. Most of my tools as a developer are now interspersed with AI features, if not based entirely around them. Most technology everybody uses is the same way, too. Every question you Google leads with an AI’s best guess at an answer, your new phone asks you to enable an AI assistant when you set it up, and your great-aunt keeps reacting with heart emojis to AI-generated videos of dogs saving babies from disaster.
Is there “value added”?
As a developer, I’ve come to understand the value added to how I do my job because of this new ecosystem of tooling. Autocompletion of boilerplate code saves me the effort of repetitive tasks, and back-and-forths with certain agents in Cursor have helped me plan and iterate on features in my project when I had no idea how to approach them. Those are two of the simplest examples of what’s available to me now, and I can’t imagine those suddenly going away. Past me would hate to hear it, but they’ve genuinely helped me do a better job. And frankly, if I don’t use them, I know there’s a big chance of falling behind as a professional.
Despite my so-far positive experience with AI in my work, the article and concept of friction-maxxing stuck with me because it mirrors many of my own personal concerns with what I might accidentally leave behind as I move into this new and robotic future. What mental muscles am I allowing to atrophy in favor of prompts to an LLM? Does learning how to rely on a magically generated solution mean I’m forgetting how to learn without one?
What about friction-maxxing in a technical role?
As I mentioned, it’s hard to imagine these tools going away, so my own friction-maxxing can’t manifest as cutting AI out of my work entirely. But if I can do things myself, I certainly try to.
Just today, I was about to prompt Cursor with a question about setting values from one Uint8Array into another in JavaScript, and the consequence of the source array having more values than can fit in the destination array. Midway through my prompt, I had a realization. It would take me the same amount of time to just do a proper experiment in my browser console as it would to write out my question. So I did just that and learned precisely what I needed.
It was a small task, but I still felt a small bit of pride in rejecting the ease of using AI in favor of doing what I suppose I am paid to do: a little bit of programming. Likewise, despite the pain of doing so, I like to write all of my blog posts without any AI assistance, whether for ideation, drafting, or actual writing. It’s up for debate whether this has a positive or negative impact on the quality, but I like knowing that my writing has my own voice. Also, the effort I expend writing will make me just a little bit better the next time.
So while maybe I don’t need to maximize friction in how I work, I do believe that there is still some friction worth having. The power of these tools comes from how we use them, and I intend to be careful about it.