Joi Ito's conversation with the living web.

Most recently in the Hardware Category


Last year, a group of Media Lab students visited Shenzhen with, bunnie, an old friend and my hardware guru. He's probably best known for hacking the Xbox, the chumby, an open source networked hardware appliance, and for helping so many people with their hardware, firmware and software designs. bunnie is "our man in Shenzhen" and understands the ecosystem of suppliers and factories in China better than anyone I know.

With his help, my students saw and experienced a ecosystem that we all benefit from, but mostly don't see or even realize exists. I have been living vicariously through the stories and reports of my students until last week, when I finally got my own tour of Shenzhen with bunnie.

bunnie insisted that we keep the group size very small because we would be going to places that couldn't fit many people and we wanted to be nimble. As chance would have it, Reid Hoffman, my old friend and founder of LinkedIn and the provost of MIT, Marty Schmidt, both were interested and available so this formed our odd little tour group.

The first stop on the tour was of a small factory run by AQS -- a manufacturer with operations in Fremont, California as well as Shenzhen. They mostly focus on putting chips on circuit boards. The factory was full of Surface-Mount Technology (SMT) machines which use computer programmed pneumatics to pick and place chips and other components onto circuit boards. In addition to the rows of SMT machines, there were lots of factory workers setting up the lines, programming the equipment, testing the results using x-rays, computers and eye balls and doing parts of the process that made more sense economically or technically to do by hand. AQS is the factory that is manufacturing the circuit stickers designed by Media Lab student Jie Qi and Media Lab grad, Ayah Bdeir's, littleBits. What's great about AQS is that, with the help of bunnie, they have started working closely with startups and other projects that previously would have had a very hard time finding a partner in China because of the small volume, high risk and usually unconventional requests that go hand-in-hand with working with entrepreneurs and our creative students.

What was more impressive to me even than the technology were the people that bunnie introduced us to, such as the factory boss, John, and the project managers and engineers. They were clearly hard-working, very experienced, trustworthy and excited about working with bunnie and our friends. They were willing and able to design and try all kinds of new processes to produce things that have never been manufactured before. Their work ethic and their energy reminded me very much of what I imagined many of the founding entrepreneurs and engineers in Japan must have been like who built the Japanese manufacturing industry after the war.

In all of the small factories that we visited, including AQS, the factory workers lived in dorms surrounding the factory and ate together and lived together. All of their living expenses were supported by the factory and their salaries went entirely to savings or disposable income. Also, all of the managers and even the boss lived together with the workers. I'm sure we were picking good factories to visit, but everyone seemed happy, open and very close.

After AQS, we visited King Credie, which made the actual printed circuit boards (PCBs). The PCB manufacturing process is a sophisticated process involving adding layers while also etching and printing all kind of materials such as solder, gold, and various chemicals involving many steps and complex controls. They were working on some very sophisticated hybrid PCBs that included ceramic layers and flexible layers --  processes that are very difficult and considered exotic anywhere else in the world, but directly accessible to us thanks to a close working relationship with the factory.

We also visited an injection molding plant. bunnie has been helping me with a project that requires some relatively complicated injection molding. Most of the plastic parts for everything from cellphones to baby car seats are made using an injection molding process. The process involves creating "tools" which are the huge steel molds that the plastic is injected into. The process is difficult because if you want a mirror finish, the mold has to have a mirror finish. If you need 1/1000th of an inch tolerance in production, you have to cut the steel molds at that precision. Also, you have to understand how the plastic is going to flow into the mold through multiple holes in the mold and make sure that it enters evenly and cools properly without warping or breaking.

The factory we visited had a precision machine shop and the engineering expertise to design and machine our injection molding tools, but our initial production volume was too low for them to be interested in the business. They wanted orders of millions of units and we only needed thousands.

In an interesting twist, the factory boss suggested that we could build the precision molding tools in China and then send these tools to a US shop for running production. Due to our requirement for clean-room processing, he thought it would be cheaper to run production in the US -- but the US shops didn't have the expertise or capability that his shop in China had to produce the tools; and even if they did, they couldn't touch his cost for such value-added services.

This role reversal is an indicator of how the technology, trade, and know-how for injection molding has shifted to Shenzhen. Even if US has the manufacturing capacity, key parts of the knowledge ecosystem currently exist only in Shenzhen.

bunnie then took us to the market. We spent half of a day there and only saw a very small part of the huge network of buildings, stalls and marketplaces. The market was several large city blocks full of 5-10 story buildings with stalls packed into each floor. Each building had a theme or themes ranging from LEDs to cellphone hacking and repair. I realize it's cliché to say this, but it REALLY felt like blade runner in a way that even Akihabara never did. I think it had a lot to do with the fact that many of vendors were selling to factories so were focused on wholesale and not retail and the volumes were huge and the interfaces were rough.

We started in the section of the market where people were taking broken or trashed cellphones and stripping them down for all of the parts. Any phone part that conceivably retained functionality was stripped off and packaged for sale in big plastic bags. Another source of components seemed to be rejected parts from the factory lines that were then repaired, or sheets of PCBs in which only one of the components had failed a test. iPhone home buttons, wifi chipsets, Samsung screens, Nokia motherboards, everything. bunnie pointed to a bag of chips that he said would have a street value of $50,000 in the US selling for about $500. These chips were sold, not individually, but by the pound. Who buys chips by the pound? Small factories that make all of the cellphones that we all buy "new" will often be short on parts and they will run to the market to buy bags of that part so that they can keep the line running. It's very likely that the "new" phone that you just bought from ATT has "recycled" Shenzhen parts somewhere inside.

The other consumer of these parts are the people who repair phones. Phone repair starts with simple stuff like replacing the screen to full-on rebuilds. You can even buy whole phones built from scrap parts -- "I lost my phone, can you repair it for me?"

After this market where phones were "recycled" we saw equivalent markets for laptops, TVs, everything.

A "SVMSMUG" phone

Next we went to another kind of market. When we walked in, bunnie whispered to me, "EVERYTHING here is fake." There were "SVMSMUG" phones and things that looked like all kinds of phones we know. However, the more interesting phones were the phones that weren't like anything that existed anywhere else. Keychains, boom boxes, little cars, shiny ones, blinky ones -- it was an explosion of every possible iteration on phones that you could imagine. Many were designed by the so-called Shanzhai pirates who started by mostly making knockoffs of existing phones, but had become agile innovation shops for all kind of new ideas because of the proximity to the manufacturing ecosystem. They had access to the factories, but more importantly, they had access to the trade skills (and secrets) of all of the big brand phone manufacturers whose schematics could be found for sale in shops. These schematics and the engineers in the factories knew the state of the art and could apply this know-how to their own scrappy designs that could be more experimental and crazy. In fact many new technologies had been invented by these "pirates" such as the dual sim card phone.

The other amazing thing was the cost. There is a very low cost chipset that bunnie talks about that seems to be driving these phones which is not available outside of China, but they appear to do quad-band GSM, bluetooth, SMS, etc. on a chip that costs about $2. The retail price of the cheapest full featured phone is about $9. Yes. $9. This could not be designed in the US -- this could only be designed by engineers with tooling grease under their fingernails who knew the manufacturing equipment inside and out, as well as the state of the art of high-end mobile phones.

While intellectual property seems to be mostly ignored, tradecraft and trade secrets seem to be shared selectively in a complex network of family, friends and trusted colleagues. This feels a lot like open source, but it's not. The pivot from piracy to staking out intellectual property rights isn't a new thing. The United States blatantly stole book copyright until it developed it's own publishing very early in US history. The Japanese copied US auto companies until it found itself in a leadership position. It feels like Shenzhen is also at this critical point where a country/ecosystem goes from follower to leader.

When we visited DJI which makes the Phantom Aerial UAV Drone Quadcopter we saw a company that was ahead. They are a startup that is growing at 5X / year. They have one of the most popular drones ever designed for the consumer market. They are one of the top 10 patent holders in China. They were clearly benefiting from the tradecraft of the factories but also very aware of the importance of being clean (and aggressive) from an IP perspective. DJI had the feel of a Silicon Valley startup mashed together with the work ethic and tradecraft of the factories we had been visiting.

We also visited a very high-end, top-tier mobile phone factory that made millions of phones. All of the parts were delivered by robots from a warehouse that was completely automated. The processes and the equipment were the top of the line and probably as sophisticated any factory in the world.

We also visited a tiny shop that could assemble very sophisticated boards in single-unit volumes for a price comparable to a typical monthly cable TV bill, because they would make them by hand. They place barely visible chips onto boards by hand and had a soldering technique that Americans will tell you can only be done by a $50,000 machine. What amazed me was that they used no assisted vision. No microscopes, magnifying lenses, etc. - workers in the US can do some of what they do, but they need assisted vision. bunnie posits that they do it mostly by feel and muscle memory. It was amazing and beautiful to watch.

We visited PCH International where we saw supplies coming in just in time to be assembled, boxed, tagged and shipped. What used to take companies three months from factory to store, now only took three days -- to anywhere in the world.

We visited the HAXLR8R, a hardware incubator in the middle of the market district run by a pair of French entrepreneurs.

What we experienced was an entire ecosystem. From the bespoke little shop making 50 blinking computer controlled burning man badges to the guy rebuilding a phone while eating a Big Mac to the cleanroom with robots scurrying around delivering parts to rows and rows of SMTs -- the low cost of labor was the driving force to pull most of the world sophisticated manufacturing here, but it was the ecosystem that developed the network of factories and the tradecraft that allows this ecosystem to produce just about anything at any scale.

Just like it is impossible to make another Silicon Valley somewhere else, although everyone tries -- after spending four days in Shenzhen, I'm convinced that it's impossible to reproduce this ecosystem anywhere else. What Marty, Reid, bunnie and I talked a lot about was what could we learn from Shenzhen to help the Boston and Silicon Valley (and more broadly the US) ecosystems and how can we connect more deeply with Shenzhen.

Both Shenzhen and Silicon Valley have a "critical mass" that attracts more and more people, resources and knowledge, but also they are both living ecosystems full of diversity and a work ethic and experience base that any region will have difficulty bootstrapping.

I do believe that other regions have regional advantages - Boston might be able to compete with Silicon Valley on hardware and bioengineering. Latin America and regions of Africa may be able compete with Shenzhen on access to certain resources and markets. However, I believe that Shenzhen, like Silicon Valley, has become such a "complete" ecosystem that we're more likely to be successful building networks to connect with Shenzhen than to compete with it head on.


I recently did a TED Talk where I provide a higher level context for my trip to and observations about Shenzhen.

Andy Rubin

Andy Rubin

Photo by Joi Ito

In designing user interfaces, we aim to empower the "user" to understand and control the system at hand. Output via screens and speakers, with input from a keyboard, a touch screen or gestures. Between them, the "user" is understood to be our conscious "mind" - the logical bit of our brain that thinks it's in charge.

This "mind" is actually not nearly as "in charge" as it thinks it is. In fact, our larger and often much more wise mind - the emotional, sub-conscious, parallel-processing, pattern recognizing part of our nervous system even manipulates and deceives our conscious mind. Articulated long ago as Dual Process Theory, Kahneman formalizes them as System 1 (this vast, quick and automatic aspect of thinking) and System 2 (the small "conscious" mind that logically considers and judges).

There is a basic fitness function to having our conscious mind feel confident, whether fighting, mating, or even making the small decisions that people make to get through a day. But the confidence we are building is with the small and logical part of our minds, deceiving ourselves that things are ok when another part of ourselves might know otherwise.

This is articulated in an experiment described by Trivers in which subjects are asked to listen to a series of voices, some of which are their own. Depending on the confidence of the subjects, some tended to attribute their voice to others ... or conversely, mistake other voices as their own. The interesting thing was that the galvanic skin response that connects to our parasympathetic nervous system always reacted consistently to our own voices, even when our conscious minds were deceived. (Trivers 1985)

Whether it's the decisions we make or the assessments of how we feel, we are consistently persuading ourselves that the world is organized and coherent, and that we understand what's going on, most of the time. In fact, the world is complex and chaotic. Most of what goes on in the world -- and even in our own bodies -- is beyond the comprehension and (luckily) the control of our little minds.

Thus, good design communicates with the broader, faster, more emotional system. What we call the "flow state" or "in the zone" is just our little minds getting out of the way so that our bigger and more intuitive mind can run the show. Whether throwing a basketball or driving a car, if our logical minds were coordinating each step, it would be impossibly difficult to coordinate all of the steps. However, our little minds are "smart" enough to get out of the way when we have mastery and allow the rest of the system dominate.

Why is it then that we seem to insist on building and assessing our systems based on what our little mind thinks? Think about the testing in schools that only measures local knowledge and logical skills, or designing user interfaces around what the user is focused on like pull-down menus and the mouse pointer.

I believe that we must focus much more on creating interfaces that send information to -- and receive controls signals from -- the rest of our system. This could apply to sensors for health, assistive robots, the Internet of things, thermostats, or future vehicles.

The problem is, individually and collectively, our little minds don't like to give up control. We have to trick our minds to get out of the way sometimes. That's where deception emerges as a design pattern.

In the late 1800s, James Naismith, a pastor and a physical education teacher in Springfield, Massachusetts realized that he needed a way to deal with young kids who would become restless and unruly during the harsh New England winters. He knew they needed the exercise, collaboration and competition they got the other nine months of the year.

So Naismith invented basketball, allowing kids to exercise indoors, to compete and collaborate, all through playing this fun new game. It worked swimmingly, and quickly spread through YMCAs and became the sport it is today. My bet is that if he had called it "social ball" or "don't-beat-each-other-up ball" it probably wouldn't have been nearly the hit that it was.

Was this subtle deception immoral? Was it effective? Which part of the mind was Naismith looking to address, and which part did he find ways to speak to?

Today, we spend so much time telling our conscious and self-deceived minds what we want it to do. What if we spent more time trying to induce our minds to get out of the way, through meditation, play, prayer ... or even deception. We need to think less like industrial designers (designing for the intentions of the conscious user) and more like game designers (designing for the desires and quick, "irrational" behavior of our mind.) We need to design our medical devices, computers, vehicles and communication tools to be influenced by what we really do and think. Not just what we tell ourselves we are doing or thinking.

Trivers, R. (1985). Social evolution. Menlo Park, Calif., Benjamin/Cummings Pub. Co.

Originally posted on LinkedIn

Yesterday, we started planning our veggie garden and started a compost bin. I'm trying to figure out what percentage of my total food intake I can grow at home. We have a relatively large yard by Japanese standards so most of this will be a matter of personal energy. I'm going to start small this year but try to increase my nutritional independence from commercial networks every year.

My goal is to be able to cover nearly all of our fertilizer needs through the composting of all of our biodegradable garbage this year.

Thinking through the various scenarios, I realized that I could significantly reduce inputs and outputs from our house by going this route. When I imagine walking over to the garden every morning, picking my veggies, then chucking the waste into the compost bin, I get a happy feeling inside. I realize this is pretty simple and not so significant, but "just add water and sunlight" is very appealing.

I think that I can also make a significant impact on my energy inputs through photovoltaics and maybe some day get off of the power grid. This requires a larger financial investment but is an area that I've already done a bit of work in this area from my time at ECD.

In my lab/office/Tokyo pad we just finished setting up (thanks to the folks at WIDE) a dark fiber connection to the WIDE box at the Japanese Internet exchange. It is currently a 1G connection. WIDE is a research project and I'm only paying for the dark fiber. WIDE is routing for me. I am not going through a single licensed telecom provider for my Internet connectivity. Consequently, going from 1G to 10G is just a matter of buying more hardware and has no impact on the running cost. More bandwidth is just about more hardware. The way it SHOULD be.

It's exciting to think about making my footprint smaller and smaller in nutrition and energy and thinking about nutrition, energy and bandwidth more and more as assets that I operate rather than services from big companies.

I was going to Twitter this as I was sitting here drinking my morning tea, but it turned into a blog post. Thanks Twitter. ;-)

I am going to be flying on Lufthanza to go to 23C3. If I'm lucky I'll have Flynet (Boeing Connexions) wifi on both legs. It's suppose to terminate on the day I return from Berlin. However... I don't have enough Watts. The Lufthanza seats (as with most airlines) only do 70 Watts. My MacBook Pro takes 85 Watts and My Dell XPS M1710 130 Watts. Arggh! This is so frustrating... If you haven't experienced trying to draw more wattage out of an airline plug, it's a pain. It looks like it's working for a minute, but it will just shut down and the LED turns red when you try to draw too much power.

It may be because of all of the complaining from people like Larry, but Apple released an airline power cable for the MacBook Pro which allows you to plug it into DC Power connectors which most US airlines use. It doesn't recharge the battery and appears to solve the problem. However, this solution doesn't help you on the airlines that just have an AC power outlet. (Most of the rest of the world.)

I wonder if it's possible for the airlines to increase the maximum power on the seats. I sure hope so.

In a last ditch effort to get my computer operational I reformatted what appeared to be a corrupt disk and borrowed an external disk to boot from. My OTHER MacBook Pro is in the shop and I had wiped it clean before sending it in. This SECOND MacBook Pro was the backup so the only backup I have is a backup backup which is about 6 months old.

Erasing the disk that possibly had the only copy of 6 months worth of data on it was an interesting thing. I knew that if I sent it to some service or used some tool that I might be able to recover some or all of the data. However, I imagined the time, stress and grief that it would cause me to engage in such an activity. I tried to take inventory of what I had done in the last six months and what items were unique and what I could recover from other people or from the Net. When I clicked "erase" on the Disk Utility, it was actually extremely liberating. Like decided to "let go" after dwelling on a loss in the family or something...

I realize this may sound a bit high drama, but I'm sure I'm not the only one whose brain shuts down to almost all outside input during a broken computer incident. Now I'm running on a fresh install with very little baggage and it actually feels quite nice. This also means no World of Warcraft and possibly more blogging. ;-)

I've started having a weird problem on my MacBook Pro. It hangs in a mode where the mouse moves and all of the applications are basically frozen. It process stuff in bursts every once in awhile. It doesn't show weird CPU activity on the activity monitor and I have about 10G of disk space free. It happens when I only have a few applications open. I've done an fsck and a permissions fix with Disk Utility. Does anyone know what the problem might be? When I reboot, it gets better, but the starts to degrade again after running for awhile...

On another MacBook Pro note... My first one is in the shop because my fan started to rattle and my CD bay and the delete key broke. This is a "backup" that I bought and it's already broken while the first one is in the shop.

OK, one more gripe. The MacBook Pro draws more power than the airline seats are designed for so the breaker on the seat will pop as soon as the MacBook Pro starts charging while on or even with the battery out, you run something that taxes the CPU. A laptop that doesn't work with airline power really cramps my style.

So although the speed is addictively fast, I'm not sure it's worth all of the problems that I've been having with it.

Anyway, I'd be very grateful if someone could help me with my most recent problem. ;-)

Have started working with Justin... or rather Justin has started working on editing video. I'm about to learn to use Final Cut Pro and turn this blog into a video blog. ;-)
Justin Hall @
Open Source Physical Objects: Limor Fried and her x0xb0x Synthesizer

Open Source Physical Objects: Limor Fried and her x0xb0x Synthesizer - a conversation between hacker/artist Limor Fried ("Lady Ada") and Joi Ito with Phil Torrone of Make Magazine. Fried talks about her popular x0xb0x synthesizer kits, and the increasing elaborate revisioning of the product that's coming from her users. With Ito and Torrone, she proposes that this is a promising model for "open source physical objects" - extending the permitted hackability of software to hardware. This is an interview from South by Southwest: Interactive, in March 2006; the camera was held by Merci Hammon, the editor was Justin Hall, and Joi Ito was the executive producer - this is part of a series of videos released online from that event under a Creative Commons Attribution 2.5 license.

Kevin Marks has created a Quicktime movie of the Steve Jobs keynote at WWDC 2005 with chapters which makes it easier to view. This is the first time I've seen chapters. Pretty neat.

UPDATE: Here's how you do it.

Technorati Tags:

Click image to see flickr image
including notes on the objects.
whatsinyourbag Originally uploaded by Joi.

I saw this fun whatsinyourbag flickr tag on Minami's blog so I decided to dump the contents of my shoulder bag on the floor (at 5AM) and take a picture.

As usual, there were a lot of PowerBooks at this conference. Interestingly, Esther Dyson, Lawrence Lessig, Bruce Sterling and I were the only people I noticed who had stickers on our PowerBooks. Other people who I know who have stickers on their PowerBooks are Mena Trott and Cory Doctorow. I wonder what this means? What do those of us who are willing to vandalize our pristine PowerBooks with stickers have in common?

Anyway, just an observation...

I saw this picture on Boing Boing. It's a ancient (about 30 years old) hard disk that probably fit about 256K according to a Boing Boing reader.

This iDuck can hold 1000X as much as that disk drive.

And these little 0.2mm RFID chips hold 128K each.

I wonder when they will start selling memory at the drug store in Petabytes per gram...

At least Kerry has good taste in computers.

via Markoff

During my session at Bloggercon, I got Stealth Disco'ed by Halley Suitt.


Who took the video?

Had dinner tonight with Ken Sakamura, the father of TRON, the realtime embedded OS which is a dominant and essential part of most embedded systems in Japan today. He is also the Director of the Ubiquitous Networking Laboratory. He brought a bunch of amazing gadgets to dinner. The most impressive were the 0.2mm 128K RFID chips in a little vial.

(c) Ted Kaehler 2003
Here is a site with a graph of the SARS epidemic. Incidentally, it is powered by Squeak.
Epidemics usually follow S-shaped curves. The predictions here are based on pure exponential growth. When the middle of the S-shaped curve is reached, the rate of infection will slow, and exponential growth predictions will no longer be useful. The reported data shows that the epidemic is still in an exponential growth phase.
Via Dave Smith but blogged first by Frank.

Monthly Archives