.png)
7 hours ago
2
There’s a familiar story about a pottery class. One group is told they’ll be graded on how many pots they make. The other, on a single perfect one. At the end of the term, the best work comes from the group that made the most.
The takeaway isn’t that more is better. It’s that quality emerges through repetition—by putting in the reps, making, reflecting, and making again. Possibly there’s also a statistical edge in more attempts, but the real value is in what you learn from them. You refine your judgement. You build taste by doing, not theorizing.
Lately, slowness has been positioned as an antidote to slop. It’s become a form of branding: a way to signal integrity and resistance to the accelerating pace of tech. In this framing, craft is synonymous with restraint. Especially in the AI era, speed often reads as suspicious, careless.
But speed isn’t the opposite of craft, and slowness isn’t inherently better. That tension played out in a recent exchange between
and Nan Yu. Marc proposed becoming a “speed propaganda account.,” encouraging people to do something cool with AI in 30 minutes, then do it 30 more times. Not to reject craft, but to remind us that speed can be a site of learning, too.
Marc is one half of Danger Testing, a duo dropping apps like songs each week. Their work is spontaneous, cultural, sometimes absurd—meant to resonate, not necessarily endure. Nan, Head of Product at Linear, works on a product known for its polish and precision. One works in rapid bursts; the other, in fine-grained systems. Their perspectives reflect a broader tension in design today: how to balance speed with care, novelty with depth.
Nan offered a distinction: there are two kinds of speed. The first is how quickly someone gets good, the pace at which they build fluency through repetition and feedback. And then there’s the speed they earn once that fluency is established—the efficiency and ease that comes after mastery. Think of it as the difference between learning to cook, and being able to whip up a great meal. With AI, those two timelines have been decoupled. You can jump straight to the appearance of fluency, even if you haven’t developed the underlying skill.
When you’ve put in the time, fast doesn’t mean careless. It means practiced. You can move quickly and still land somewhere good. Like Paula Scher, you’ve spent a lifetime designing logos so that you can dash off one on a napkin in under five minutes.
What’s changed is that the second kind of speed is now available to everyone. You don’t need to study game design or typography or animation to produce something that looks finished. You can skip the training and still get something that appears finished without understanding how or why it works. But looking finished isn’t the same as being good.
And that’s the trap. Tools produce polish, but not perspective. You’ll see people post their one-shotted output on Twitter—raw, unedited—and ask sincerely, “Chat, did I cook?” The answer is often: no. Sure, the aesthetics are there. The structure’s plausible. It resembles good work without necessarily being good work.
That’s the illusion of fluency: a resemblance of quality, without the process that produces it. Without mistakes or thoughtful revision, it’s hard to know what’s working—or if anything is. You see it, you like it, you ship it. But you haven’t pushed it. You haven’t learned from your own choices because you haven’t made any yet.
Photo editor Emily Keegin describes this as the gap between thought and click. “It looks great really quickly, but that’s the danger,” she told me. “You don’t iterate the way you used to. You don’t push past the obvious.” What’s lost isn’t just labor, it’s the friction that helps sharpen instincts. Tools that collapse the process can collapse the learning, too.
Marc framed it this way: if speed is free, the leverage shifts from craft to curation. When anyone can produce something that looks good, the question becomes not “Can you make this?” but “Can you tell if it’s any good?” Or: “Can you make it better?”
This is where many of us are getting stuck. Our ability to generate has outpaced our ability to evaluate. You can create dozens of variations in a day, but your capacity to interpret them—what worked, what didn’t, and why—hasn’t changed. That still takes time. And it’s in that slower layer of work that taste is developed.
Taste is not a fixed attribute. It’s trained. It comes from time spent in the medium, from decisions made and revised, from paying attention to what resonates and what doesn’t.
Fast can surface instinct, but only if that instinct was built through practice. Speed alone won’t do that. You still have to notice what’s working. You still have to edit. You still have to care.
AI removes friction. That’s part of what makes it powerful. But creativity needs constraints. Without them, it’s easy to confuse movement with progress. Time is one constraint we can reintroduce. One hour to make something. One day to sit with it. Not to delay, but to pace. To reintroduce reflection without letting it turn into paralysis.
It’s not a new idea. There’s a long history of using time as a constraint to force clarity. 100 days of making. Thirty days of vibecoding. Two-week sprints. In 2016, three of us at HAWRAF did A–Z: 26 projects in 26 hours, one for each letter of the alphabet. Every hour, we pulled a random letter and made something. It was public. It was messy. Most of it was bad. That wasn’t the point. We weren’t aiming for polish, we were learning how we worked together, what we reached for under pressure, what stuck, and what didn’t.
That kind of structure can feel counterintuitive now, but it’s often the only way to get past the obvious. Once the easy ideas burn off, better ones have a chance to surface.
This is what’s often missing in AI-first workflows. When the first draft looks good, there’s no clear reason to keep going. But without moving through variation and failure, you don’t build the kind of discernment that leads to originality. You miss the chance to make the thing only you would make.
Craft is a kind of attention. It’s not just the result; it’s everything that led to it. The choices, the edits, the judgment behind the scenes. Even if you can’t see the process, you can often feel its presence.
This is why we still respond to effort. Why watching someone land a skateboard trick feels satisfying. Why hearing a live performance resonates. Why we care about provenance, even when the outcome appears similar. There’s a difference between a fast idea and a practiced one. We know it when we see it.
Slow built your instincts. Fast lets you use them. But you still have to know what you’re reaching for. You still have to care.
And you still have to throw the pots.
Thanks to Marc and Nan, for reading drafts of this piece and for having this discussion on the timeline. <3
