Saturday, October 29, 2011

Vision video

Here's an inspired video by Microsoft on how we might interact with our digital devices in the future:

There's a number of interesting ideas here to ponder.

First, notice how all the devices interact with each other. While she's in the taxi, her handheld device beams tomorrow's meeting details to the cab's display, which then highlights the building her meeting is in the next day. As she enters the hotel room, the hotel display lights up recommending running routes, and so on. This isn't something that will be very easy to build: Just imagine the security and privacy implications of having your handheld beam personal data to the cab opportunistically: The cab may not have driven by her next day's appointment, so the handheld would have needed to send more than was displayed to the cab. We'll also have to figure out how to keep these devices connected, as cell network connectivity is still very limited and short-range network technologies like Bluetooth wouldn't allow for the instant connectivity as demonstrated here.

Also notice all the gesture-based controls in the video. People are touching and swiping on surfaces, but also waving their hands to flip pages, whether at close range or sitting on the hotel bed paging through recipes on the hotel TV. I wonder how hard this will be to build. We've seen gesture-based controls in Kinect and Wii, but they only work with extensive calibration and the controls are imprecise at best for the types of interactions shown here, for example cropping and zooming the world map back at HQ. Then of course you'll have to teach humans how to use all these fancy gesture-based controls, which will require a lot of thought.

I love watching videos like this that show how the computing revolution is far from over. Lots of work left to do.

Living in the 21st Century

Here's a fascinating chart from the The Economist's Daily Chart blog, slightly edited for size [*]: 

The world's population and economic output are growing exponentially. No surprise there. But there's an interesting angle to look at population: Instead of counting the total number of people on the planet, you count the number of minutes lived by humans in a particular century. In the first 10 years of this century, we've out-experienced the 17th century.

We love comparing ourselves to others, and we see more people than ever having more experiences than ever. With Facebook, Google+, and Twitter permeating our lives, others' experiences are more accessible, so we're bombarded with things to compare ourselves to, giving us lots of opportunities to feel inadequate.

On the bright side,  humanity has certainly earned how to make products and services. With the 21st century already at about a third of of the output of the entire 20th century, people are living healthier, more productive, more comfortable lives than ever.

[*] I realize the y-axis in the chart is a percentage, and by cutting off 1st to 8th centuries so the chart would here horizontally, they no longer add to 100%. Yet I feel like the central point of the chart still remains.

Sunday, October 09, 2011

Less stuff, more happiness

A great TED talk about how owning stuff makes you happier, and how this century will be about editing rather than the relentless pursuit of more.

What's in the box?

Saturday, September 17, 2011

The Caller from Canada

My new car has Sirius satellite radio which comes with around 140 channels of niche channels, including one channel called POTUS which is mostly politics talk.

There's a show on POTUS called "Standup with Pete Dominick", where a former comedian talks politics with random callers from all over. When I was driving back to SF from Google on Thursday around 10 pm he couldn't fill his hour-long slot, so he offered any caller 30 seconds to talk about whatever they wanted to say. (Note that there is a 7-second delay to the show to bleep out coarse language.)

One of the callers was from Canada and his advice for Americans was to have more confidence in investing in the future of their nation. Dominick put him on the spot, and asked what specifically Americans should invest in. The caller fumbled a little bit - "there's really a bunch of stuff" - and before his time was up, the only example he could come up with was "running shoes".

Running shoes are clearly not the prime example of what America needs to invest in[*], but as the clock hit 10 pm, I found myself turning off the radio and thinking about what the right answer should be.

It's clear that the US needs better infrastructure - fix the roads, improve road safety, and build faster, better trains. We need more solar panels on roofs so we can burn less coal, and more wind farms that feed into the grid. We need better education and better educators so that kids no longer drop out of high school, but prepares them for the jobs of the century ahead. Better and cheaper health care, better medical records, the list of big-ticket items goes on and on ...

One problem is that the US is a country that's in love with the quick fix, and the things we need to do are long-term investments that need years and decades. And that requires confidence. The caller from Canada, while wrong on the running shoes, was right.


[*] To the caller's credit, he got cut off before he could launch into an argument. Maybe he was going to propose a scheme to give everyone running shoes to decrease obesity and thus long-term health care costs.

Tuesday, September 06, 2011

The only time you'll buy and read print magazines

In the future, the only time you'll buy print magazines on paper is at the airport. The only time you'll read them is during taxi, takeoff, and landing. That's when electronic devices have to be switched off.

Tuesday, August 02, 2011

VC Investment Returns: Europe vs US

Roughly paraphrased, the takeaway from this deck is that in Europe, startups compete for fewer investment dollars, which leads to better startups getting financed, and exits that are bigger. Thus venture investors get larger returns from investing in Europe than in the US.

You should flip through this, if only for the awesome graphics.

-- as seen in David Cowan's post on TechCrunch

Friday, July 15, 2011

Electric Cars

In the last few weeks, I've rented several Chevy Volts by the hour. I'm really optimistic about electric cars, and they're clearly the way of the future. I hope a decade from now, we'll see many or most miles driven in the developed world driven with electricity, ideally powered by electricity from solar or other renewable sources.

(Random thought of the day.)

Saturday, July 09, 2011

Solar Energy - Where's the Platform Play?

After my last blog post on investing in solar energy, a friend introduced me to Kerim Baran, cofounder of CivicSolar, an online distributor of solar equipment. Kerim gave me a little more data to chew on. Thus, here are some more thoughts on the space.

What does Solar on Your Roof Cost?
Installation costs right now are $5/Watt, which breaks down as follows: $2 for panels, $0.5 for inverters, $1.50 for balance of system (racks, wires, connectors, etc.), and $1 for installation. The average US residential solar installation costs around $20k.

Service companies like Solar City that plan and install solar panels on your roof typically make 30% margin on the installation of the job, and will provide financing costing an additional 10-15% which is backed by banks. So if you don't pay cash, and don't arrange the installation yourself, you're looking at $7 per installed Watt of solar capacity.

The price breakdown above also illustrates how the industry of getting photovoltaic (PV) cells on your roof is structured:

PV equipment manufacturers like Mayer and Burger, and GT Solar make equipment for a huge field of commodity PV manufacturers. Racking, cabling, and inverters are made by a different set of companies. Offline and online distributors like Kerim's CivicSolar get the wares to installers, which are often local contractors (although SolarCity also does its own installations). The local contractors often are electricians or construction companies that need third-party services like software to plan out the installation on your roof, and training on how to install the cells in the first place. These contractors are also hired by full-service firms like SunRun and Sungevity that present packaged deals of installation and financing to consumers who want solar on their roof.

Falling Costs
Don't mind the $7/Watt cost though - the US Department of Energy wants solar installations to go down to $1/Watt by 2020.

Right now, just installing the panels costs that much. Panel prices will fall, racking might become easier (think solar cells that you drape over your house like a carpet), and inverters could become cheaper (inverterless panels seem possible) or unnecessary (most devices in the average household require DC, not AC, so the extra conversion could be avoided).

The goal of $1/Watt is far away, but it illustrates just how far prices could drop.

Temporary Low
At the moment, PV manufacturers valuations are depressed because government solar subsidies are ending - Germany is one example of a country that's phasing them out. Some of the PV manufacturers' stocks are trading at 5 P/E, despite 10% profit margins and solid growth. Yet demand has declined and panel manufacturers are sitting on extra manufacturing capacity and stockpiled solar panels.

Despite this, betting against solar right now is like betting against Microchips in the late 70's: yes, demand temporarily fell, but given how much dormant demand there is, it would be stupid to bet that the solar industry will shrink long-term. Just think about all those big mansions in the Sun Belt of the US that want to reduce their $400 air conditioning bills.

Platform Plays
In the computer industry, platform plays always pay best: Build your own proprietary system that's an significantly better than the status quo and allow others to build on top of it. Think Android, Chrome, Windows, iOS, Macs, and so on - all of these are highly successful platform plays. I think it's reasonable to assume that the most powerful solar investment strategy would be in a platform play.

Where to Invest?
But what's the platform play in solar energy? All the hardware - PV cells, inverters, racking - is commodified as far as the eye can see. Photovoltaic technology is actually pretty simple, low tech stuff. What's improving is the PV manufacturing process. Yet betting on the PV equipment manufacturers feels wrong - you're betting on the second derivative here, because in order for these companies to grow, the PV manufacturers need to increase their rate of growth, not maintain it.

It's the installers like Solar City, SunRun and Sungevity that might end up becoming the platform. Homeowners can call them up and have them take care of installation, just like consumers can build their own computer out of individual parts, but prefer to buy one from Dell. If you will, these installers are building a service platform on commoditized hardware.

The problem is that it's hard to invest in these companies as an individual investor. None of them are public, and even when Google[*] put $200 million into Solar City, my understanding is that Google didn't get equity. The Sequoia Capitals of the world that invested in the early stages will reap the benefits.

[*] Yes, I work at Google but I don't do investments, nor am I speaking here on Google's behalf.

Tuesday, July 05, 2011

The Four Steps to the Epiphany PDF

I was just recommending "The Four Steps to the Epiphany" by Steve Blank to an entrepreneur. I had no idea it was available as a PDF. Here it is:

The Four Steps to the Epiphany PDF by Steve Blank

Download this to your tablet and read it on a long plane ride sometime.

Tuesday, June 28, 2011

Thoughts on Investing in Solar Power

I've been thinking about investing in photovoltaic solar energy. Here's why:

Energy consumption is growing world wide, and we need to build more capacity. As a country, your options are: Build nuclear plants, build coal plants, build wind parks, build solar parks, or encourage people to add solar to their roofs. Post-Fukushima, it's going to be hard to build nuclear plants. Coal and gas are heavy carbon emitters. Wind isn't suitable for everywhere. Solar is a great, emission-free option, especially for sunny climates.

Demand for photovoltaic will continue to grow. Solar stocks are depressed right now as some countries like Germany and Spain are ending their subsidies to install solar panels. This doesn't mean the end of solar installations though.

One crucial advantage of solar power is that the decision to install solar on your roof can be made and financed by individuals, and the electricity can be consumed where it's produced.

For example, in the US, demand will grow as people realize that they can knock down their air conditioning bill by installing solar panels on their roofs. In sunny states, air conditioning makes up 60 to 70 percent of the summer electricity bill. Solar power provides electricity precisely when air conditioning consumes it - in the middle of the day - no batteries or storage needed. From a recent visit in Arizona, it seems like few people have solar on their roofs, and there will be a lot of cells as consumers get more informed.

There have also been developments in financing solar installations. SolarCity, a company founded by Elon Musk and backed by Google, will finance your solar panels and lease them out to you. This allows people to realize savings without the upfront cost - more and more people are likely to use it.

Germany is a small, rainy country and had 7400 MW of photovoltaic (PV) solar power installed in 2010. The US, sunnier and bigger, meanwhile has 878 MW installed. There's plenty of room to go.

How to invest? It seems like there are three ways to bet on PV solar power.

It seems like the best bet would be to invest in financing companies like SolarCity. They will have reliable cashflows, high profits, and there's little competition in the space right now. Unfortunately, I couldn't find a single public company that does this. Google has made this bet by investing in SolarCity, but it's still a private company, so this isn't something an individual investor can do.

You could invest in the companies that make the equipment to manufacture PV cells. These are primarily German and Swiss companies - one example is Meyer and Burger. They have high profit margins, but by betting on the equipment manufacturers, you're betting on the second derivative: The growth in manufacturing capacity, not on growth in solar power.

Lastly, you could bet on the companies that manufacture the actual PV cells. This is essentially a commodity market, with many companies in the US (e.g. FirstSolar, GT Solar), China (e.g. Suntech, JA Solar), and Germany (e.g. Q-Cells, Solarworld). Some of these companies are trading very cheaply right now because of concerns of overcapacity. Just like there used to be 500 car companies in Detroit at the beginning of the 20th century, I expect a lot of this industry to eventually consolidate. Thus, it's crucial to pick the winners. I've ran the numbers on some of these companies. I'll share those in a second post.

Sunday, June 26, 2011

A 21-year old kid coding away on his MacBook somewhere

"In Silicon Valley, you get used to the thought that there's a 21-year old kid coding away on his MacBook somewhere, and that that kid will be the next Mark Zuckerberg. The underlying meritocracy of this thought is exciting."

Monday, June 20, 2011

Samsung Galaxy Tab 10.1 Unwrapping

The Samsung Galaxy Tab 10.1 is a beautiful device, and it's thinner and lighter than the iPad 2. I just got it in the mail from Amazon, and in honor of Apple product unwrap documentaries around the world, I decided to post some pictures of the unpacking process.

Note that this is the Samsung Galaxy Tab 10.1, not the Samsung Galaxy Tab 10.1v, which is its slightly fatter and older brother.

For some good reviews of this tablet, check out Anandtech's review which calls it "the sleekest Honeycomb tablet" and ArsTechnica's piece which says it's a "whole new world".

I've already ordered another one of these as a gift.

Click on the pics to get the full-size view.

Package arrives from Amazon:



Opened package

Cables - there are in-ear headphones,  power adapter, a special Samsung cable (looks like an iPod/iPhone cable) and a manual

Backside wrapped

Backside unwrapped

Turning it on for the first time

Samsung logo on startup - apologies for the glare, I work in a pretty bright office

It's soo thin

Home screen

Google Maps - correctly picks up my location as the Googleplex

Google Maps 3D View

Headphone jack on top

Speakers on each side

Samsung connector (iPhone/iPod-like) on the bottom

The tablet arrived with 4% remaining battery so I had to plug it in pretty soon after it arrived. In addition to my Galaxy, I also have a Motorola Xoom that Google gave me, and compared to the Galaxy it seems outdated - it's heavier, thicker, and less pretty.

Wednesday, May 25, 2011

Organizing your Email into Folders is a Waste of Time

Here's a brand new paper from a conference in Vancouver last month:

Whittaker S., Matthews T., Cerruti J., Badenes H., Tong J. .: "Am I wasting my time organizing email? A study of email refinding"

They instrumented the email client of about 345 users of Bluemail, an email client built by IBM research, and identified two subsets of users:

  1. Users that do a lot of foldering (organizing email into folders)
  2. Users that don't use foldering and rely mostly on search.

There are a bunch of great results in this paper, but let me zero in on the most surprising result.

The authors looked at operations where the user would try to find a particular email, either by starting a search, or browsing and scrolling through folders. They looked at whether the user was successful in finding the email he or she was looking for. "Success" was defined as opening an email and reading it for more than 29 seconds (the usual amount of time it takes to read two paragraphs of text), or if the user did something to the email after opening it. For example, opening an email and replying to it would count as a "success" as well.

The results are surprising: Both groups - folderers and non-folderers - found the email they were looking for 88% of the time. Non-folderers found their emails faster too: In 66 seconds instead of 73 seconds.

Organizing your Email into Folders doesn't make finding your emails easier or faster. People that put emails into folders spend more time organizing their inbox, more time searching their email, and don't find emails more often than people who just use search.

Thursday, March 24, 2011


If you're not using Hipmunk for flight search, you should be.

I'm going to Peru in May to see Machu Picchu. The flight I was about to book on Orbitz looked innocent enough, but it wasn't very clear that there was a 6-hour layover in there. Here's the same flight on Hipmunk:

As the proud holder of a corporate job, flight times matter a little more to me than money. Hipmunk is the ideal tool to help me inconvenience in tavel.

Disclaimer: I'm an investor in the site.

The Next Mark Zuckerberg

"There are a bunch of 19-year-old startup founders running around these days saying that they want to be the next Mark Zuckerberg. Some of them don't say it, but they're still thinking it. The reality is that if you want to get funding being 19 years old, you have to get traction and growth. That's what set Zuckerberg apart and that's still what it takes to get funded when you're that young. If you don't have a track record, it's too hard to tell whether you're going to give up and go back to school."

-- An associate who's fielding deals for a top tier VC fund, to me

Thursday, March 03, 2011

glob.h and glob.c for the Android NDK

I've been playing around with the Android NDK (native development kit) recently, which lets you compile C and C++ code directly to your Android phone's ARM chip rather than writing Java. This is very useful for performance-critical apps. Unfortunately, the Android NDK doesn't have a full set of libraries, and when you're compiling third party libraries to Android, you often run into these missing pieces. One of the biggest missing pieces is glob.h / glob.c, libraries which are used for file name matching aka globbing (think the "*" in "*.txt").

I've massaged a glob.h / glob.c from FreeBSD to compile on Android. They might come in handy in your Android NDK adventures.

Download: glob.h | glob.c.

You'll need to add glob.c to the LOCAL_SRC_FILES in your

Sunday, February 27, 2011

Only about 7% of retail is online

I got into an argument with a friend the other day about how much retail happens online vs. offline. Given the amount of money I spend on Amazon and the various branches of the empire that is Gap, I guessed that about 10% of retail happens online (at least in the US), while my friend insisted it was more like 2%.

My estimate was based on the convenience of home delivery and the quick, cheap delivery networks of UPS, FedEx, and OnTrac (a logistics company Amazon uses for a lot of its deliveries in California). In addition, residents of many populous states can avoid state sales taxes by ordering stuff from Amazon: For exmple, Amazon doesn't have operations in California and thus does not have to collect sales taxes. (However, in theory, California residents have to pay sales taxes on goods purchased out-of-state via their tax returns.)

After some research, I found this TechCrunch article from 2009 suggesting that in 2011, 7% of retail will happen online. Not quite 10%, but still impressive.