Archives for the month of: February, 2011

I just got back from a week long trip to the west coast. The main purpose of the trip was to present our paper at the International Solid-State Circuits Conference (ISSCC), which is one of the top conferences in semiconductor where Intel, AMD and IBM present their state-of-the-art processors. I had been to this conference before without a paper, but this was the first time to present my work. It was a totally different feeling and I was very excited during the entire conference. I was also fortunate to be invited to give talks at SoC companies including Qualcomm, Apple, Nvidia and Marvell during my trip. I received invaluable feedback from great audiences, and had so many interesting conversations with brilliant people. I enjoyed the trip so much and came back energized enough to work hard on another chip =)

I might have another post on things I learned from the trip, but this post is going to be about the dreadful flight experiences I went through during the trip. I flied three times — Boston -> SFO (United), SFO -> San Diego (Virgin), San Diego -> Boston (Jetblue) — and all three flights were delayed by over an hour. The BOS -> SFO flight departed on time, but took 1.5 hours longer to get to SF than scheduled due to strong headwinds. The SFO -> San Diego flight was a short 1.5 hour flight, but was delayed by 1.5 hours! Lastly, the flight back to Boston from San Diego first got delayed, but when I showed up at the gate on time, no one was there. It turned out that there was no plane leaving from San Diego, so everyone had to hop on a shuttle and go to Long Beach Airport. I was so tired after my week-long trip that I was more than ready to fall asleep on the red-eye flight, but instead I had to endure a 2-hour bumpy shuttle ride to Long Beach, go through security again, and walk through a rainstorm before finally getting on-board to sleep.

I was just so amazed how every single flight could be so messed up like this. Maybe it was just bad luck to blame. Maybe I was carrying bad weather all the way from Boston to cause the delays. Maybe three is too small of a sample size to make any definite judgement. Still, I just couldn’t understand how these airlines could stay in business with these poor services. Maybe every single airline in the US have terrible service. Maybe that’s why they can all stay in business even when they torture their customers, providing them with experiences they never want to go through again. I don’t know much about how the airline industry works, but I just feel there should be a fairly simple solution using high-tech to solve these problems.

This is a slightly different story that my wife and I went through when we were coming back from vacation in Chile. We were at Santiago International Airport to catch an American Airline flight back to the US. We arrived 3 hours in advance, but there was already a very long ticketing line. It turned out that there were 2 other AA flights with hundreds of passengers leaving at pretty much the same time. Hundreds of people stayed in line waiting for hours. Although we got there 3 hours before departure, we barely made it to our flight. Okay, I can stay in line if there’s a lot of people. What was really annoying? There were a bunch of self-ticketing machines lying around with no one using them. For some reason, the machines required a “ticket number” that was not written in our priceline itineraries. You couldn’t get your ticket from the machine with your passport or itinerary. I wonder how many people actually have their “ticket number” that’s not even printed on your priceline itinerary. It seemed not many people had that number in hand given that almost no one was using the machines.

I really couldn’t understand why they placed the machine there in the first place if no one could use them. I couldn’t understand why they required these weird numbers for the machines, while they required only passports to be shown when they handed out tickets in person. If the long lines were inevitable even with the machines, they could have used a simple machine spitting out numbers for customers. If your number shows up on a LED screen, you could show up to receive your ticket. This kind of simple solution could prevent everyone from standing in line for hours dragging around their luggage, and instead allow them to relax at a cafe waiting for their number to show up on the big screen. It just seemed like airlines had no motivation to come up with solutions to comfort their customers.

I had a similar experience at Hertz renting a car during my trip to SF (OK this is the last example of a bad experience). There were a bunch of people on the line, maybe because it was Friday of a long weekend. People were waiting over an hour in average to get their cars, but Hertz employees kept on leaving for lunch, making the waiting time even longer and longer. The more annoying part was, again, there were a bunch of self-sevice machines that didn’t work. Hertz was running short on mid-size cars and had to upgrade those who reserved mid-size to higher-end vehicles. The machines weren’t smart enough to do that and kept on refusing to give cars to any customer who reserved a mid-sized car. Again, a very simple high-tech (maybe not even high-tech) solution could have made the machine smarter, and could have slashed the waiting time for customers by a tremendous amount.

How can we force these companies to change? Is there a way to make them care more about their own customers? I think nothing will change unless a brand new company shows up with much better customer service, better enough to generate an exodus of customers from existing companies with crappy services. I thought Virgin, Jetblue and Southwest could be those type of companies, but it seems that’s not the case or they aren’t successful enough yet. Maybe Zipcar is that company in the car rental industry.

New companies come up with fresh ideas to compete with existing giants. This competition is key to innovation in the industry as a whole, and this is what benefits customers. Hope there’s enough competition, enough to motivate traditional industries like airlines and rental-cars to change their quality of service. Glad that I don’t need to travel too often.


Technologies change so fast. People tend to forget very quickly what life was like just several years ago without all the fancy technologies we nowadays take for granted. I remember 10 years ago in 2000, I was in high school and very few of my friends had just started getting cellphones. I remember thinking cellphones are luxury items for teenagers, and a beeper would be more than enough for me. While I was debating whether I should buy a beeper, cellphone prices began tumbling down. By 2001, more than half my classmates started using cellphones and I ended up getting one too.

Fast-forward to 2011. We have handsets with more compute power than desktop computers in 2000, with battery life that last an entire day, and small enough to put in our pockets. Even better, they are affordable. I plotted the price decrease of the Amazon Kindle since its release in late 2007. I felt the Kindle had been around forever, but realized it hadn’t even existed when I came to Harvard in 2006. The price dropped from $400 to $140 in just 3 years.

How about the iPhone? The latest models are priced at $200 (with 2yr contract), but I wonder how many people remember the first iPhone with a whopping $599 price tag (with 2yr contract). Amazingly, this was just 4 years ago.

More powerful, cheaper gadgets flood the market every year, and the driving force behind this is Moore’s Law. Any techie might be fed up with Moore’s Law by now (I see it in the first slide of almost every academic talk in our field), but one cannot emphasize enough its impact in the entire IT industry. Thanks to device physics researchers, transistors have been getting smaller and smaller, allowing chip designers to pack in more transistors in a single chip while shrinking the chip area. Smaller chip real-estate translates to cheaper chips, similar to smaller apartments costing less than larger ones. More transistors enable more functionality in chips. As a result, you get more powerful, but cheaper chips that power your smartphones.

Smaller transistors decrease manufacturing cost per chip, but non-recurring cost is becoming significant as well. This is a one-time cost of designing the chip before sending it to the fab that manufactures the chips. Chip design became increasingly complex due to increasing number of transistors in a single chip. Processors in commodity phones were much easier to design than those in state-of-the-art smartphones with support for audio/movie playback and web-surfing. Hiring more engineers to tackle complex chip design leads to higher cost. Fortunately, companies came up with design automation softwares that automated large parts of the chip design, allowing them to suppress the explosion of cost in engineering manpower. This also helped decrease chip design time, reducing time-to-market and allowing companies to churn out new processor designs every year in time for new smartphone releases.

However many times I think about Moore’s Law, it’s just unbelievable that it has been sustained for the past 50 years, and still on-going. I can’t think of any other industry like the semiconductor industry. Price falls every year, but that drives explosive increase in market penetration and creates new market segments starting from laptops to smartphones, ebooks and tablets. Increase in volume is more than enough to justify the decrease in price. Better technology at cheaper price -> larger market, revenue-> invest in better technology is the cycle that drives explosive growth.

So where are we going looking forward? Can price just keep falling forever? Can we come up with better technologies forever? Will customers say they don’t need anything better anymore? What does this imply to handset manufacturers and processor designers?

There seems to be no sign of slowdown in the price fall. Huawei, a Chinese company that both designs processors and manufactures handsets, teamed up with T-Mobile and announced the Comet smartphone priced at $10 with 2-yr contract and $150 without contract. Compare that to $200 with contract and $600 without contract of high-end smartphones! With the Comet, users can surf the web, watch movies and listen to music, albeit at a sluggish speed. Of course performance of the Comet is not as good as the iPhone, at least for now. But what would happen in 2 years? Huawei will be able to offer phones with performance similar to current iPhones at a much cheaper price. Customers might decide they don’t need more performance and settle with the cheaper Huawei phones. Would there be enough functionalities for the iPhone or other high-end smartphones to keep adding to differentiate themselves and justify more expensive price-tags?

I purchased a Samsung Galaxy S a month ago. I did plenty of research in engadget and played with a lot of smartphones, but I didn’t notice any serious difference between the iPhone and the other Android contenders from HTC, Samsung and Motorola. There are more apps in Android than I can possibly imagine downloading. Scrolling and pinch-and-zoom were smoother in the iPhone, but I didn’t think that was a major difference. Right now the high-end Androids have the same price-tag as the iPhone, but what if emerging companies like Huawei come up with similar Android devices at a much cheaper price? Would customers be willing to pay $200 more for smoother scrolling and pinch-and-zoom?

Compared to Android, iPhone could be at a better place to lock in customers thanks to iTunes. Just like Amazon can lock in ebook users to Kindle thanks to their enormous amount of book contents, Apple could prevent users from ditching the iPhone for cheaper phones by making iTunes an essential part of their lives (I seldom use iTunes in my Mac, but I suspect a lot of iPhone users use it for music). Maybe locking in users with rich contents is the right way to avoid losing them to commodity handsets. Actually, Amazon Kindle’s success makes me wonder what would have happened if News Corp had come up with something like the iPad before Apple did and package it with The Daily using their power over contents. After all, Amazon didn’t have any experience in manufacturing before the Kindle. I guess making a multi-purpose tablet would have been much harder than designing a device mainly for book reading like the Kindle.

There could be a limit in squeezing in more functionalities to justify pricy high-end smartphones. Maybe they are eventually going to turn into commodity phones, maybe free phones that anyone can afford with contract. The key to avoiding this could be in tightly integrating the hardware with high-quality contents that users are willing to pay $$$ for. All these scenarios assume hardware having nothing to improve in the future, hardware being a commodity that anyone can manufacture, and contents being the main differentiator.

Although I agree with the power of contents, I still believe there is a lot of room for innovation in hardware. Hardware still has the power to bring wonders that nobody ever imagined. The past 50 years show hardware can exceed all expectations (like this wrong prediction by McKinsey on cellphone penetration). In the future, there could be watch-sized smartphones thanks to 3D-stacked chips. There could be huge flexible displays that you can fold to fit in your pocket. There could even be wearable computers connected to your brain to assist your thought-process. These are things customers will be willing to pay for. These technologies will keep driving hardware innovation and continue the wonders of the semiconductor industry. These technologies will bring better gadgets to customers at a cheaper price. These technologies will prevent hardware from becoming mere commodity.

I still believe in the future of hardware innovation, and I’m happy to be part of it =)

%d bloggers like this: