From The Street...
Friday, April 25, 2003
Wired 11.05: START The Shape of Things to Come
The bell curve, that beautiful form of regularity, is getting turned upside down.
By Daniel H. Pink
In the mid-19th century, a few of Europe's finest scientists and mathematicians noticed something peculiar about the way the world organized itself. When they measured large samples of various things - the length of people's middle fingers, say, or the price of certain goods over time - the results tended to cluster around an average. When plotted on a chart, the data took the shape of a bell. This bell curve, as it was dubbed, defined "normal distribution" - and it became a fundamental law of natural science, an elementary truth about the nature of reality. Charles Darwin's cousin Francis Galton - who pioneered the related idea that everything from stock prices to your descendants' IQs would eventually revert to the mean - called the bell curve "an unsuspected and most beautiful form of regularity."
EMEK But today Galton's beautiful form might not be so regular. Although bell curve distribution is still considered normal, a surprising number of economic and social phenomena now seem to follow a different arc. Instead of being high in the center and low on the sides, this new distribution is low in the center and high on the sides. Call it the well curve.
It's popping up everywhere. Take the size of organizations. Large companies - pairing off like beer-goggled students at a frat party - are becoming gargantuan enterprises. Think HP-Compaq, Citigroup, and AOL Time Warner. Meantime, small enterprises are also proliferating. The US Census Bureau reports record numbers of "nonemployer businesses" - teensy firms without any paid employees. Yet while the big grow bigger and the small multiply, midsize enterprises are waning. The pattern is similar in geopolitics. The past decade saw the rise of both huge multinational federations (Nafta, the European Union) and tiny secessionist movements and small independent states. But the political entities in the middle - countries such as Italy and Spain, for example - are on the unprecedented brink of losing population.
Consumer culture is going bimodal, too. Electronics manufacturers are racing to equip us with screens small enough for cell phones or large enough for home theaters - relegating standard screens to the scrap heap. High-end luxury hotels and low-end budget chains are doing well - but at the expense of midprice accommodations. In retail, Wal-Mart is soaring, boutiques are thriving, but middlebrow Sears is struggling. As The Wall Street Journal noted last year, "consumers are flocking to the most expensive products and the cheapest products, fleeing the middle ground in between."
Then there's the drooping middle class. The Federal Reserve Board's latest analysis of family finances showed that from 1998 to 2001, American incomes were up across the board. But when economists divided the population into five equal segments, a well curve emerged. "Incomes grew at different rates in different parts of the income distribution," the Fed reported, "with faster growth at the top and bottom ranges than in the middle."
We can even glimpse this inversion in the classroom, where most of us first encountered the bell curve. The National Assessment of Education Progress is a standardized test also known as the nation's report card. It measures how well American schoolkids can read and groups them into four categories: advanced, proficient, basic, and below basic. From 1992 to 2000, the percentage of students scoring in the top two categories increased. The percentage scoring at the bottom level remained distressingly high. But the portion scoring basic dropped nearly 10 percent. Even the nation's children, it seems, are graded on a well curve.
Of course, not everything we can measure conforms to this new shape. In national politics, the fastest-growing affiliation is Independent. Diversity and interracial marriage are rendering the old bimodal and trimodal racial categories irrelevant. Yet almost everywhere we look closely, we find ourselves staring down a distributional well. The implications are huge: insurers, marketers, and policy-makers may be basing decisions on faulty premises about what is normal. They're assuming a vibrant center - Middle America, middlebrow tastes - when the action has migrated to the edges. The 180 from bell curve to well curve has turned their logic on its head.
Galton and his contemporaries believed that conditions would deviate from the bell curve only during periods of transition. Every age, of course, supposes it is living through a unique era of profound change. But in our case, the conceit might actually prove true. The madness of our times might simply reflect our stumbling effort to revert to the mean. Either that or one of the world's eternal verities is less eternal than we supposed. This deviation may turn out to be anything but standard.
------------------------------------------------------------------------
Daniel H. Pink (dp@danpink.com), the author of Free Agent Nation, is writing a book about the rise of right-brain thinking in modern life.
Wednesday, April 16, 2003
John Dvorak on Foveon...
From his March 4th column in PC magazine...
=====================================
Whither Foveon? Dept.: So where is the heralded Foveon CMOS chip that is supposed to blow CCD technology out of the water? Here's what insiders tell me has happened. First, there were some technical gotchas that have been resolved. In the interim, both Fuji and Sony spent billions on new CCD manufacturing plants and are simply keeping the CCD prices artificially low?supposedly.
You may begin to see Foveon chips in cameras made in China. Talk about a marketing hurdle! Do you want a Nikon camera or a Good Luck Dragon of Serenity camera? Unless Chinese companies partner up with German companies, or any companies with reputations for making high-quality cameras, the whole idea is hopeless. Japanese companies look to be standing firm in rejecting the Foveon sensor. Canon is the only company seriously playing with CMOS.
================================
Open Innovation
Traditionally, new products are developed by in-house R&D; because of increasingly more complex "solutions", most all the work must be done in house to properly integrate the parts.
A Harvard Business School Professor is interviewed: here is part one and a link to the source at the bottom.
Finding smart people who don't work for you and getting them to participate; and making it all work by concentrating on the gaps, are all parts of a (cliche coming...) New Paradigm. A Hot Book on this is Open Innovation. The fella interviewed is the author.
===========================
IdeaFlow: creativity & innovation
Renee Hopkins has posted part one of her interview with Henry Chesbrough, the Harvard Business School professor who's just published "Open Innovation: The New Imperative For Creating And Profiting From Technology." His central argument: the traditional, internally focused model of innovation is becoming obsolete and that what's taking its place is a new paradigm of "open innovation" that has companies leveraging, licensing and buying existing technologies developed elsewhere.
An excerpt: "Closed Innovation is fundamentally about scarcity of useful knowledge. In order to do anything, you have to do everything. It is inwardly focused, and deeply vertically integrated. It takes little or no notice of external knowledge and resources. Open Innovation is fundamentally about operating in a world of abundant knowledge, where 'not all the smart people work for you', so you better go find them, connect to them, and build upon what they can do. It seeks ways to build upon and leverage external knowledge, and focuses internal activities upon filling in the gaps, and integrating internal and external knowledge into useful systems."
Idea Flow - Corante
Sunday, April 13, 2003
Gilder Storewidth Meeting Report - The Age of Cheap...
Brief overview of Storage in the Age Of Cheap...Welcome to Storewidth.com
Disruption in the Age of Cheap - The Convergence of Storage and Bandwidth Marches On
by Graeme Thickins, graeme@thickins.com
Another year, another eclectic gathering of technologists,
investors, and gurus from the storage and networking industries,
to assess the state of disruption and innovation in the converging
worlds of storage and bandwidth. But this one was different.
It was the third iteration of Storewidth, a conference that's
turned out (quite unintentionally, for certain) to parallel
"three awful years in the IT industry," as Forbes publisher Rich
Karlgaard noted in his introduction. He also set a tone for the
event by citing what he's now calling the "cheap revolution,"
led by such rising stars as Google, running on 12,000 cheap PCs
and the free Linux operating system. Another example: "Sun's
biggest competitor today is eBay, selling one-year old servers."
The warning is clear, said Karlgaard: "Never judge the health
of this industry by the leading companies."
Searching for 'Google-izers'
Storewidth has survived in a difficult conference environment
because its relevance won't go away. It's focused on a technology
truism that no economic turndown can stop, and a need so great --
data growth that's more than doubling every year -- that it simply
can't be ignored or put off. The enterprise, increasingly running
on networks, is finding storage the fastest growing, most time-
consuming application on those networks. And, with budgets
flat while storage doubles, it's no stretch to say trouble is brewing.
"It's now legions of people and bucketloads of Tylenol," said
Jonathan Martin of Veritas.
Though attendance has ebbed somewhat (to about 250 this year),
the conference continues to attract high-level attendees and
speakers, a strong showing of VCs and private investors, and
relevant, timely insight and debate.
Participants ranged from mostly senior execs of the computing and
storage networking leaders at one end, whose firms in large part
continue to hold their own (and even thrive) in the tech slowdown --
while, at the other end, there were promising young firms who
continue to get VC cash by claiming disrupting, innovative
technologies that can help enterprises solve their ever-expanding
data growth and storage management problems. (And, yes, more
than a few representatives of the former were there to shop for
the latter, as acquisitions continue at a strong pace in this
tech sector.)
Tech For Tech's Sake, Not
Disruption was indeed the underlying theme of Storewidth --
crowned by a keynote from Clayton Christensen, author of "The
Innovator's Dilemma," who offered hints of his next book -- along
with the more blatant theme of managing complexity, stated early
and often. "It's become crucial," said storage-industry guru Rich
Lary, "And the answer is NOT more complexity. Technologists
aren't that good at simplifying, because it's not as much fun."
Translation: simple is really hard, and it just may require new
players, unburdened by tradition, to come out of nowhere and
make it happen.
The big, whack-upside-the-head takeaway of the conference,
then, was this: more technology, in the basic sense, isn't what's
needed. Moore's Law is no longer the driver of this industry, in
the Age of Cheap. What's needed are better management solutions,
to address the big picture -- not whiz-bang technology for more
point solutions; enterprises have way too many of those already.
The successful disruptors in the "Storewidth Era," as conference
founder George Gilder likes to call it, will be those who make it
easy for enterprises to manage their growing data burden, and put
that data to good use.
"How can anyone make money when everything's so cheap?" Gilder
asked. "Peter Drucker says the profits go to the supplier of the
'missing element'." That element, the conference made clear, is
simple management of stored data. "Converting heterogeneous
petabytes to useful information," said Gilder. "That's
Storewidth."
Netting Out Two 13-Hour Days
No one complains of not getting enough content at a Gilder
conference. Here are some other major nuggets that filtered
out of a very packed agenda, and some further insights on
the above points:
- Labor Pains: Of the three major IT budget components, labor
is the only one increasing. Hardware and software costs continue
to drop. The growth rate of labor costs is accelerating in large
part because of enterprise IT's complexity problem -- of which
storage management is an increasing percentage. "Storage
provisioning is a 12-step program -- and we know how painful
those can be," said consultant Rich Lary. It can take days or even
weeks. "IT management is now saying, reduce my costs or pain,
or go away." The holy grail of storage provisioning, Lary says,
is to make it as simple as provisioning memory in servers. And
there are some encouraging signs that vendors, large and small,
are addressing this problem, other speakers said.
- The 'SMB' Love Affair: The small to mid-sized business
market is where the storage networking industry knows it must go
for growth, with big enterprise budgets flat. Few, however, have
made good inroads there yet. One company the industry may learn
something from in this regard is Quantum, said storage consultant
and author Jon Toigo (one of the moderators), with backup tape and
disk subsystems. Also, the adoption of iSCSI systems to facilitate
IP networking of storage is finally happening this year, first getting
traction at the department level and in SMBs, before it gets wide
adoption in enterprise networks -- where it will eventually displace
the more complex, costly fibre channel protocol. Most say the
company best positioned here is Cisco, but virtually every major
storage company has placed bets.
- Software Reigns: Much of the pressure in the storage business
is going to software -- or, as one conference session put it, "softening
hardware," to make it less fixed and unchangeable. And leading storage-
software company Veritas had a strong presence at the event (it was a
sponsor, too). It is one of several large firms that are actively acquiring --
in its case, to gird for what it sees as a new future to provision storage
as a utility. "Automated provisioning will remove the labor component,"
said Mark Bregman, EVP of product operations for Veritas.
- Utility Computing Is For Real: Several speakers referred to
a new future of "storage as a utility." This movement is part of a
major push by the big computing vendors to a new IT model called
"utility computing." IBM, HP, and Sun, among others, have large
initiatives and investments to deliver computing in a manner akin
to an electric utility. And storage is being elevated to a higher
level in the IT mix through these programs, because of the dire
need for better management, not to speak of the ability to charge
back departments for their IT and storage usage -- a long-awaited
vision that may start appearing in some solutions as soon as this
year, according to some speakers.
- Storage Services Has Legs: Near term at least, while we're
waiting for all this automation nirvana, low and behold, storage
services companies such as Arsenal Digital Solutions, CNT, and
others are doing well helping enterprises deal with the storage
they have, then helping to lead them to the promised land.
"Storage management is more art than science," said Randy
Whitehead, SVP of operations for Arsenal Digital Solutions.
"That's changing, but not soon." He noted his firm is growing
214% year over year.
- 'Value' - The Other Word for Cheap: Switching to other,
related industry segments, we're not only well into the era of the
"Value PC," but the "Value Transitor," too. "There's a sea change
coming in the semiconductor industry," said Nick Tredennick,
editor of the Gilder Technology Report, "More disruptive than the
IC itself." The emphasis moves now to embedded microprocessors,
a much bigger market -- but one that's happy with chips from two
or three generations back. "Fewer and fewer apps are helping pay
for the next generation of transistors," he said. "Intel may get to
its 60Ghz vision, but who cares?"
The Three Funniest Quotes
George Gilder: "Why not the Ritz this year?" [The conference
had been held at the better known resort across the coast highway
the past two years.] "In this economy, we didn't want a bunch of
tech execs drinking heavily next to a cliff."
Mark Bregman, Veritas: "Forrester predicted back in '97 we'd
need 8 million storage administrators by 2003. Heck, there are only
10 million Elvis impersonators out there -- and, unfortunately, it's
the same labor pool."
Steve Forbes (who delivered a keynote address the second morning,
covering many topics): "No matter what he does, Jacques Chirac
will not become the new Sun King."
"The Innovator's Solution"
In a keynote address held strangely *after* the closing conference
reception (yet very well attended), Clayton Christensen, Harvard
Business School professor and author, offered up some hints of
his next book, following on to the very successful "The Innovator's
Dilemma." He said it was the first time he'd spoken on the topic,
and that his new book is due in September. One somewhat surprising
point: he dissed MBAs big time -- saying such analytically trained
marketers are not right for growth companies, giving the example of
how they actually ended Sony's long string of disruptive innovations
back in the '80s. He also noted how his own institution's expensive
MBA program is now being disrupted by corporate universities.
=================================
Cheap Revolution
Several things I've seen recently suggest that the Latest Thing is Cheap. Technology is squeezing profit margins from most products. Forbes' Rich Karlgaard suggests how to survive - and win - in such an environment.
Forbes.com: The Big Cheap Chance
=============================
You can escape the Cheap Revolution's unrelenting margin pressure in only three ways:
* Improve your product offering faster than anybody else can. This tactic works, generally, only for market leaders with a good brand name, a greased distribution channel and financial might. (Think Intel.) And it works only as long as the market wants the added functionality and is willing to pay for it.
* Sell fast custom solutions that answer a customer's needs. Xilinx, with its programmable logic chips, does this. So does IBM, despite its size. IBM's trick has been to go modular and bring into its Big Blue tent an army of third-party solutions providers. The old, highly integrated IBM would never have been able to react quickly enough to customers' needs.
* Find an unserved market and serve it cheaply. This is the way of the disrupter, says Christensen. The product or service should be so cheap, in fact, that the industry's old guard thinks there's no money to be made and walks away. An example, says Christensen, is how Sony served teenagers with transistor radios in the 1950s. Their parents owned expensive tabletop vacuum-tube radios from RCA or Philco. For their part, RCA and Philco knew all about transistors. But they figured these solid-state curiosities would never replace vacuum tubes until they were capable of producing a superior sound. That left the teenagers unserved. Tinny-sounding as transistor radios were in 1955, they were--by the standard of no radio at all--good enough. By serving the unserved, Sony got a toehold in consumer electronics and never looked back.
Saturday, April 05, 2003
Dead Parrot -
Dead Parrot
This is the script of the classic Monty Python skit about a pet store owner selling a dead parrot to an incredulous customer. Better seen than read, of course, but it ranks with "Who's On First" as a comedy treasure...
Wednesday, April 02, 2003
FPGA's instead of Microprocessors....
The Next Big Thing in Semiconductors...
Gilder talked about this recently. Microprocessors are kludged into doing too many things, wasting "time" and power. With new portable devices acquiring computing and connectivity features, power is now real important. FPGA's can be reconfigured , in this instance, every forty milliseconds....
Here is a fellow who has taken the first steps towards doing a solution, it seems. The software is difficult, it says here. But the potential and implications are clear...
Forbes.com: Super-Cheap Supercomputing?
==========================
Super-Cheap Supercomputing?
Daniel Lyons, 03.25.03
NEW YORK - Star Bridge Systems employs 16 people and rents office space in a modest beige building near a Celtic memorabilia shop and a health food store on Main Street in tiny Midvale, Utah, which isn't exactly the middle of nowhere, but close enough. The company's founder and chief technologist, Kent Gilson, is a 37-year-old high-school dropout who for years has been derided by computer scientists as, well, a bit of a fringe character. The company's chief executive, Daniel Oswald, joined Star Bridge four months ago after running a foundation that dealt with ancient religious texts and Mormon studies.
In other words, IBM (nyse: IBM - news - people ) this ain't. Nevertheless, this band of outsiders claims to have created a reconfigurable "hypercomputer" that performs like a supercomputer but sits on a desktop, uses very little electricity, needs no special cooling systems and costs as little as $175,000.
The secret is in the chips. Instead of yoking together hundreds or even thousands of microprocessors--as traditional supercomputers do--Star Bridge uses a dozen or so relatively inexpensive field-programmable gate array (FPGA) chips. Each FPGA can handle thousands of tasks at the same time, in parallel. A microprocessor can only do one thing at a time. So even though a traditional supercomputer might have hundreds of microprocessors, a machine with only a handful of FPGAs can outperform it.
Also, FPGAs can be reconfigured using memory cells connected to the transistors on the chip. So unlike most chips, which can't be changed once they're made, an FPGA's circuitry can be redrawn over and over again. Engineers use FPGAs in satellites because this lets them upgrade computers without replacing chips; they just beam up new software and redraw the circuits.
Now imagine that instead of reconfiguring a chip, say, once a year, you could update it every 40 milliseconds, on the fly, to handle different tasks--like a factory floor that could constantly change shape to build different products. In theory you'd have an amazingly powerful machine, one that not only did parallel processing but also could constantly redraw its circuits, optimizing itself for each new task, applying resources as needed.
For years people have speculated about creating a general-purpose FPGA-based computer. The trouble is the software. It's incredibly difficult to write programs to make an FPGA-based computer do all this reconfiguring. So far the best thing engineers have done is create FPGA-based machines tuned to perform a single task. They might program a machine to, say, perform a compute-intensive genetic research algorithm over and over again at blazing speed. Those machines are cool but they are a bit like idiot savants--brilliant at one thing, dumb as a bag of rocks at everything else.
Now Gilson claims he has solved the problem. He says that after five years of work he has created a programming language called Viva that lets developers easily write applications for an FPGA-based machine and an operating system that can run those applications.
Star Bridge sells four FPGA-based "hypercomputer" models with prices ranging from $175,000 to $700,000. The "sweet spot" machine, called the HC-62, sells for $350,000 and contains 11 Xilinx (nasdaq: XLNX - news - people ) FPGA chips, which cost about $3,000 each. That model will perform 200 billion floating point operations a second. The $700,000 model contains 22 Xilinx chips and can perform 400 billion floating point operations a second, Gilson claims. In addition, customers must license Viva, paying $45,000 per year for a one-person license.
Gilson has not subjected his machines to industry benchmark tests. But if his claims prove to be true, the top-end Star Bridge machine would rank among the 200 fastest supercomputers in the world, comparable in power to multimillion-dollar machines made by Hewlett-Packard (nyse: HPQ - news - people ), IBM and Silicon Graphics (nyse: SGI - news - people ) that contain hundreds of microprocessors. The Star Bridge box would not compare to the world's most powerful supercomputer, which performs 35 trillion calculations a second. Then again, that machine, made by NEC (nasdaq: NIPNY - news - people ), has more than 5,000 microprocessors, takes up a four-story building in Japan and cost $350 million.
Gilson says he understands that people will be skeptical about the claims he is making. It doesn't help that two years ago one of Star Bridge's first customers sued the company, claiming he made payments totaling $200,000 to Star Bridge and never got a workable computer. A few months ago Star Bridge settled the dispute by giving the customer shares in Star Bridge in exchange for his $200,000.
Gilson insists his dream machine actually works. "I live in the future," he says. "Most people are pessimists who live in the present or the past."
Gearing up to start making sales, Star Bridge recently hired Edward McGarr, a former Novell (nasdaq: NOVL - news - people ) executive, and Rebecca Krull, former 3Com (nasdaq: COMS - news - people ) executive, to run sales and marketing, respectively. Star Bridge will focus its first sales efforts on defense and security companies, government agencies, biotech firms and oil-and-gas exploration companies, all of which need machines that can perform compute-intensive tasks. Gilson won't say what he expects to do in sales this year.
Over the past two years Star Bridge has sold about a dozen prototype machines based on an earlier design to the Air Force, the National Security Agency and the National Aeronautics and Space Administration, among others. It has also sold seven of the new models.
Silicon Graphics has also asked Star Bridge to send along a copy of its hardware and software. The $1.3 billion (fiscal-year 2002 sales) supercomputer maker wants to explore ways to make a Star Bridge system work with a Silicon Graphics machine. "For our customers that have specialized applications needs there is a place for reconfigurable computing, and we're interested in how Star Bridge software might play into that for us," says David Parry, Silicon Graphics' senior vice president and general manager for servers and platforms.
Olaf Storaasli, a senior research scientist at NASA's Langley Research Center in Hampton, Va., has been using Star Bridge machines for two years and says they are very fast but not yet ready to handle production work at NASA. "It's really a far-out research machine," he says. "It's more about what's coming in the future. I would not consider it a production machine."
One problem, Storaasli says, is that you can't take programs that run on NASA's Cray (nasdaq: CRAY - news - people ) supercomputers and make them run on a Star Bridge machine. Still, he says, "This is a real breakthrough."
Allan Snavely, a computer scientist at the University of California at San Diego Supercomputer Center, has been using a Star Bridge machine for about a year. He says he originally contacted Star Bridge because he suspected the company was pulling a hoax. "I thought I might expose some fraud," he says.
But after meeting with Gilson and seeing a machine run, he changed his mind. "They're not hoaxers," he says. "As I came to understand the technical side I thought it had a lot of potential. After talking to Kent Gilson I found he was very technically savvy."
Snavely and a colleague have been able to write programs that take advantage of the on-the-fly reconfigurability of FPGA chips. "You might do a two-part calculation where the hardware as it's set up for the first part is not optimally configured for the second part, so you reconfigure the hardware at that stage of the calculation," he says.
Snavely says it is not easy to write programs with Viva. "It's still a significant effort to describe a programming problem on their system," he says. Yet he believes the Star Bridge machine is ready to do some limited production work. And he believes FPGA computing is the way of the future. "Star Bridge has equipment that is raw. The full promise of it has yet to be determined. But the more I study this approach to computing, the more I think this is inevitable."
Gilson says that by the end of this year customers will be using his machines to do real work, though he concedes it may take many years for "hypercomputing" to catch on in a big way. He says he can wait. "This is a marathon, not a sprint," he says.
====================================
