New business models, the explosion of data, and new uses for technology topped the agenda at last week’s IBM Almaden Institute, which brought together a group of very smart researchers and industry experts from around the globe. They talked about how to innovate by taking advantage of the rising tide of data; topics included “Life on an Instrumented Planet.” Here are some highlights from a handful of the talks at IBM's research center in the hills above San Jose, Calif. I think the speakers did a good job of reaching high elevations in their talks. This is the first of two posts on the conference.
Hal Varian, chief economist at Google, opened the conference with a talk about how components, complements and standards have become critical to innovation. Components are things like screws or microprocessors that are the essential ingredients for a more complex system. This was pretty elementary stuff, but Varian had some insights that you only get if you look at things from a high elevation.
Complements are things like DVD disks and DVD players. They both have to arrive in the market at the same time in order for the DVD market to take off. Varian says that coordination isn’t easy because such complements often involve companies in different industries with different business models. One of the fundamental drivers of coordination has been Moore’s Law, which holds that chip capacity doubles every couple of years or so, putting demands on improved performance across the entire technology spectrum.
Often, the value of one product is enhanced by the presence of another, like hamburger and ketchup or microprocessors and hard drives. Sometimes one complement tries to absorb another, like when Sony (consumer electronics maker) bought Columbia (movie studio) but that frequently results in poor results (AOL Time Warner).
Standards are great for the creation of product ecosystems, but it’s dangerous when one or two companies (Wintel) monopolize the supply of a standard component. If one partner in the ecosystem cuts prices, the entire group benefits. Revenue sharing turns out to be one way to get ecosystems to accelerate. For instance, DVD rentals didn’t take off right away because initially, Blockbuster had to buy movies from the DVD makers. They only became a hit when DVD makers gave Blockbuster the disks for free and then, in turn, Blockbuster gave back a share of the revenue every time a disk was rented. Hence, in today’s increasingly complicated world full of complex product ecosystems, innovations in business models may be key to making a product take off.
Kris Pister, founder of Dust Networks, talked about the progress made in the last decade by mesh networks based on a “smart dust” wireless sensors. He admitted that a lot of the early excitement around such networks was overhyped. As a professor at UC Berkeley, he pioneered the research from around 1997 to 2002. The goal was to create mesh networks with thousands or millions of nodes, each around a millimeter cubed in dimensions. Such nodes could be deployed in places such as grape vineyards where they could measure data, such as dryness or temperature, and pass it along to the next node via a short-range wireless signal.
Pister founded Dust Networks in 2002, and the company has been laboring to build the market ever since. The company provides the entire system, not just the nodes, or motes. By 2003, Pister's company could get 2-millimeter-cubed wireless chips; battery life is now five years to 10 years and power consumption is as little as 4 milliamps. But the market, projected to be 700 million units a year in 2007, didn’t happen. Part of the problem was continuous changes to the Zigbee radio standard.
Dust Networks, meanwhile, tried to focus on getting to 99.9 percent reliability for mesh network data transfers. It got to the point where nine of 10 packets could make it from one node to another; with redundant paths, the reliability is good enough. Now the networks are being deployed. Coal plants are using them to monitor whether motors are running or on the verge of breakdown. Costs are about $1,000 per node, but that’s cheaper than the cost of manually monitoring every motor in a factory, particularly in remote places such as a North Sea oil drilling platform.
Honeywell offers a service where it installs a wireless sensor network in grocery stores, monitoring chillers and other energy systems. It can cut energy costs by 10 percent to 25 percent. Streetline Networks uses the sensor networks to monitor open parking spaces in the city of San Francisco. It can also identify parking meters that have expired.
One of the next steps will be locating objects, Pister says. The networks can be used to locate valuable equipment such as defibrillators in hospitals or where you left your car keys. RFID was supposed to do this, but a lot of suppliers found that it really only told them “the last time an object passed by a scanner in a warehouse.” Pister didn’t talk about the privacy implications of this. That’s a huge obstacle. I think that Pister’s company can do well in the market for monitoring the performance of million-dollar machines. But it’s going to be a long time before we use mesh networks to find our car keys.
Andreas Weigend, former chief scientist at Amazon.com and a tech consultant for a variety of companies including MySpace and Nokia, predicted that Web 2.0 would lead to a data revolution that in turn would spur big gains in “relationship economics.” He pointed to Google’s ability to mine data from the searches we do. By collecting data on everything from searches to Google Maps, Google has far more data than a typical point-of-sale retailer, he said.
This data that we voluntarily give to Google will be so valuable at some point that Weigend expects companies will start paying us for sharing our data with them. That’s already happening a little. The same goes for data such as our whereabouts, which companies such as Dash Navigation can discover through our use of GPS navigation devices.
Jigaw, he noted, has made a business providing contacts for business people. Members must upload contact information from their business card collections in order to receive access to the contacts they want. There are now eight million contacts available on Jigsaw. This is, as with Pister’s sensor networks, another example of a data-mining business with some big privacy issues. You can lose control of your private information just by giving someone your business card.
Weigend observed that consumer data collection has turned upside down. At first, companies paid their own employees to collect data. Now, users are sharing a rich amount of data about themselves and their relationships. And in return, they expect to get value from releasing this information.
Horst Simon, associate director of the Lawrence Berkeley National Laboratory and keeper of the top 500 supercomputers list, says that nine petabytes worth of data are being generated each year from research into climate change. That’s understandable, given the need to collect all sorts of weather data from around the world. But it’s also one of the reasons why we’re headed for a huge amount of data collection, storage and computing in the future. After all, climate change is just one of the many big topics that require supercomputing. The National Energy Research Scientific Computing Center is seeing its data archive grow about 1.7 times per year. IBM’s BlueGene/L supercomputer is No. 1 on the top 500 list, by the way.
Mark Dean, director of the IBM Almaden Research Center, spoke about how to bring technology to Africa and lift that continent out of poverty. IBM has had staffers there for 55 years and is expanding its operation there dramatically. It donated a 14-teraflop BlueGene supercomputer to in Capetown, South Africa. Part of Africa’s problem: it doesn’t produce enough end-consumption goods with its natural resources. Ghana produces 70 percent of the world’s cocoa beans but only has one or two chocolate brands. It costs less to ship goods from the U.S. to Botswana than it does to ship from Botswana to neighboring Tanzania. The economy is benefiting from barter, including the trading of cell phone minutes; Dean says that farmers who use cell phones to sell their goods can get 30 percent better profits than those who don’t. It doesn't take much to make a difference with technology, but I can't say I heard a lot of reasons to be optimistic.