We traveled to Shenzhen for the Scalable HCI Symposium alongside MIT Media Lab fellows, Harvard and Stanford graduates and researchers from around the world, and long-time hacker friends to get closer to the current state of hardware.

One thing that stood out is the speed of the feedback cycle.

When you can iterate close to manufacturing, the system optimizes for implementation. Things get copied, remixed, and shipped quickly because the goal is momentum and throughput. In that environment, a lot of details that feel sacred elsewhere become negotiable.

You see the tradeoff.
Convenience can be extremely high.
Differentiation can be extremely low.

When multiple products come from the same factories, the logo starts to feel like a sticker. The signal is not in the mark. It is in whether it ships, whether it works, and whether it can be repeated.

That does not mean there is no taste. It means taste is not always the bottleneck.

A large part of the westernized hardware industry is storytelling: brand, promise, and projected futures. But proximity changes the signal. When you get closer to the source, it becomes easier to tell the difference between what is imagined and what is actually being made.

Some people know how to sell. Others know how to build. Rarely do they do both.

In places like Shenzhen, design for manufacturing is not a buzzword. It is a reflex. There is an overskilled labor force capable of creating almost anything you can imagine, and doing it repeatedly at scale.

Meanwhile, in San Francisco, New York, Boston, and similar cities, companies pay ultra premium salaries into a limited talent pool. Often, that talent is several layers removed from fabrication, assembly, suppliers, and production constraints.

This trip was about closing the distance between ideas and where they’re made.

We were surrounded by what felt like an XXL freshman class of hardware founders, the Class of 2026.

The biggest gap in hardware is not talent. It is distance.

Distance from the factory.
Distance from the supplier.
Distance from the feedback loop that determines what is real.

From far away, it is easy to confuse a pitch deck for a product. Up close, the noise collapses.

Being closer to the source is not just faster. It is more economical.

Hardware becomes social because scale is a chain of commitments you cannot make alone. The moment matter is involved, truth stops being singular. It lives in tooling, yield, vendors, lead times, customs, and the quiet heuristics of people who have shipped before. You can rewrite software at midnight. You cannot rewrite a mold, a purchase order, or a container shipment already on the way. That delay forces synchronization, and it turns trust into throughput. Learning accelerates when these partial realities meet around the same object in the same room, because the object creates a shared reference point and makes disagreements testable. Hardware scales when a network aligns on what will actually ship, on time, at quality, and in a way people can rely on.

The willingness to be transparent, and to actively augment the maps we move through, echoed how we think about schoolscapes, as explored in a recent Record of Thought, Augmenting Education.

It underscored a quieter shift in emphasis: away from the abstract question of “What do you want to learn?” and toward the more grounding, consequential one, “Where do you want to learn?”

Honestly, a personal highlight moment for me was not just the keynote or a panel.

It happened later, when Cedric and I DJ’d.

Vinyl in one hand. A Transport from USB Club plugged in from the other.

Analog and digital met from opposite ends and converged into the same working system. The format of old school and new school met.

Both sides sharing an appreciation for the curation of not just music but moments, people and the places we share.

Massive respect to the MIT Media Lab and SUSTech Design School for building environments where ambitious ideas are expected to survive contact with the real world, while taking on the deeper macro challenge of scaling hardware.

Particular thanks to Cedric for stewarding the program and for consistently insisting that the future earns its place by becoming physical, legible, and human scale.

Grateful to the people who helped shape the experience:

First to the organizers Cedric Honnet (MIT), Seungwoo JE (SUSTech), Zinn / Ziyin Liu (SciArt Co-lab, AIRS); to the many practitioners and educators that brought food for thought including Marcelo Coelho (MIT Design Intelligence Lab), Jiabao Li (Northeastern/Stanford), Jun Rekimoto (UTokyo, Sony CSL), Ken Nakagaki (uChicago); and to the many people who made the room what it was: Cyril Engman, Liz Dorman, Alex Ollman, Seeed Studio, Yuxiang Cheng, Joey Castillo, Spencer Chang, Anna Brewer, Char Stiles, Andy Kong, Kelin Carolyn Zhang, Sebastian Bidegain, Harry Isaac, Weiwei Hsu, Evan Kahn, Yufeng Zhao, Omar Rizwan, Jesslyn Tannady, Miyu Horiuchi, CJ Pais, Shih Wei Chieh, Sidney San Martín, Jenny Zhang, K. Michelle, Zain Shah, Jonny Cohen, Miranda Lin Li, Matthew Siu, Taylor Tabb, and so many other contemporaries.

Not every contribution fits in a byline. The real value was in the fast, practical exchanges: a supplier name, a known failure mode, a good place to eat, a workaround that saves a week.

There were generations before us, and there will be after us. For more info check out the Scalable HCI program.