Wednesday, December 22, 2004

December newsletter II: 2005 predictions, A-Z

Early Indications is published twice monthly by John Jordan. He holds no direct financial positions in any company mentioned. Back issues are currently archived at

2005 Predictions

By now many observers have issued their 2005 predictions, so I decided to follow suit, utilizing an alphabetical format to address some of next year's most intriguing issues. In the spirit of full disclosure, many of these predictions are actually questions.

A is for Apple, which has to address some big questions. Having reinvented the mass-storage maket by equipping the iPod with a great interface and compelling legal content, the leadership must anticipate the eventual margin erosion in the hand-held segment and decide how long to ride the premium-price position in computational jewelry (i.e., what used to be called PCs).

B is for business intelligence, the fancy name for data warehousing. For information to enhance business outcomes, it has to fit more closely into real processes. The MBA analyzing data cubes has far less leverage than the people at the point of customer activity making better decisions in the moment. That objective means that data analysis tools will have to become more industry- and process-specific rather than generic, and they can no longer be so detached from operational systems.

C is for consumer devices, which are gradually having an impact on enterprise systems, which can no longer assume an inside-the-wall, locked-down configuration. Wireless data access is a prime example, but so are things like gaming interfaces, iPods as oversized USB drives, and instant messaging.

D is for distributed development. The issue isn't really India per se, because there will be new low-cost environments for certain kinds of work as India develops inflation, a middle class, and/or heightened political tensions with Pakistan. Managing distributed development is a more general issue than merely signing up resources in India or Spain or Estonia, and the tools for doing so are still genreally immature.

E is for energy, which remains a constraint for everything from mobility, in the form of battery life, to data centers, in the form of heat. Intel recently had to switch over to dual-core processors to maintain its stream of new microprocessor introductions because of heat, and the marketing strategy for Centrino (slower clock speed, better battery life) doesn't immediately translate into a parallel pitch for desktop and server chips that will no longer be positioned solely on speed.

F is for fiber optics, which remain a wild card in the the quest for widespread residential broadband access. Relatively speaking, Verizon is taking an aggressive position with fiber to the premise (rather than the node) in the Keller, Texas trial. Longer term, both capital and regulatory uncertainty loom. Meanwhile, wireless broadband deploys far faster and at lower cost, and it's completely possible that anyone who spends billions of dollars digging up yards in 2005 could be aced out by a wireless carrier within five years.

G is for gaming, in both the Vegas sense (a euphemism for gambling) and the Nintendo sense of the word. It's not clear how long state and federal governments will forego potential tax revenues from Internet casinos: for all the talk about the presidential election's being a contest of values, many states that voted for Bush have aggressively adopted lotteries, casinos, and other betting. Meanwhile, the battle for handheld console supremacy between Sony's PSP and Nintendo DS will have global implications for telcos, movie studios, hardware manufacturers, and other parties.

H is for health record, electronic. In the midst of federal budgeting, President Bush has to balance conflicting promises: to push for better use of digital record-keeping technologies in healthcare, and to reduce the deficit. It's too early to tell how the battle will play out, but to date few legislators have seen support for electronic medical records as a way to impress their constituencies.

I is for industry consolidation. Oracle's takeover of Peoplesoft will by no means end this trend, and 2005 may see such companies as Siebel, BEA, Cognos, Hyperion, and Lawson dealing with acquisition efforts. On the services side, it's less clear that a) bigness equals profitability and b) growth by acquisition works for human assets.

J is for jail. Sarbanes-Oxley section 409 is still being interpreted, but some provisions for timely disclosure took effect in August. The legislation uses the terms "real time" and "urgent" for these disclosures, which will add to the CIO's already substantial compliance burden - and provide tough penalties for failure. "Real time" for some purposes is four days, but retrieving a given e-mail or category of instant messages, for example, within that time is impossible for most organizations.

K is for killer application, or more properly the lack thereof. Intel has suffered as both consumers and business users find it difficult to justify new hardware purchases for such predominant tasks as e-mail, web browsing, and spreadsheets. On mobile platforms, meanwhile, cultural differences drive divergent adoption patterns of everything from mobile messaging to cameraphones to geolocation. Personal digital media management, in the form of iPods and TiVos, has sold well, but not all that well. In a global market, it's worth reflecting on total TiVo sales: 4.6 million for 2003, and probably less than 10 million total worldwide as of mid-2004.

L is for license revolt. Enterprise software buyers uniformly resist the move by enterprise software companies to counter declining new seat sales with increased support and maintenance fees. open source is one strand of the movement to limit the power of the vendors, but there are others: 3rd-party support firms, self-support, and refusal to buy upgrades. With increased vendor consolidation, expect to see the stakes increase, and customer resistance grow more vigorous.

M is for management software. Given that headcount remains a large and, thanks to health care costs, growing component of IT budgets, and given that the complexity of the IT shop is still growing despite efforts to rein it in, better tools for running the IT business are essential. Some are in early deployment. Consider that front office, back office, sales force, shopfloor, and field service all have been automated, but the IT organization typically runs on spreadsheets rather than audit-able, robust enterprise systems.

N is for network topology, inverting. Currently, most data network traffic flows outward, from servers to the edge. With the stunning growth in RFID and other sensors, voice over IP, music uploads, and networked games, that pattern is in flux. IDC predicts that edge-in traffic will outstrip inside-out flows in less than ten years, necessitating new approaches to security, reliability, and flexibility.

O is for open source. Linux and Apache are old news; 2005 could be the year that the MySQL database and JBoss application server move out of stealth mode and into more active market penetration.

P is for personalization. Amazon's consumer site is custom-generated on the fly for every registered visitor, representing an integration of cookies, behavioral profiling, databases, dynamic site-serving, and other technologies. That kind of performance and relevance will be expected as the norm in more settings. At the same time, the limits of the username+password model are being exceeded, and new identity solutions will get at least trial deployments in 2005.

Q is for QWERTY, or the keyboard paradox. For years, Bill Gates and others have predicted that voice-recognition and realistic speech synthesis are on the horizon. But a funny thing happened: young users of mobile voice devices have exhibited a preference for text input rather than speech. Demographics, Moore's law, and cultural norms will all shape the next wave of input preferences, but none will be exactly what the visionaries at Dragon and Lernout & Hauspie predicted back in the mid-1990s.

R is for RFID. Retailers already have the business case and many of the business practices in place to exploit the consumer-products and pharmaceutical tags; what will change dramatically are the behaviors and expectations in such places as hospitals, unionized warehouses, and courts. What are the rules for using tags (or automobile "black boxes") as evidence? What are the privacy rights of an employee suspected of theft or even of slacking? How will the black market adapt to the presence of tagged Oxycontin in both legitimate and shadow supply chains?

S is for search. Google's ambition and capability are both formidable: their agreement with leading university libraries to digitize some of their holdings parallels a less-visible effort at the Internet Archive and will be a landmark in information access. Yahoo, Amazon, and Microsoft, meanwhile, are devoting major investment and brainpower to various categories of search challenges. Given the magnitude of information volumes both at rest and in motion, traditional methods for storing, finding, and manipulating data will have to be reinvented - and more layers (in the form of geospatial, audio, and other aspects) are still in the queue.

T is for telephony. Voice over IP is changing the economics of the large national and Bell carriers, it's changing law enforcement, it's changing marketing, and it's changing enterprise computing. How it's changing these various things is not yet clear: regulators, lawyers, taxing authorities, and courts will have much to say about the eventual patterns of adoption. For example, the U.S. federal government took in over $5 billion in telephone tax receipts in 1999. Replacing that stream, along with a probably larger flow to states, will be controversial, more so in an age of deficits.

U is for unsolicited bulk e-mail, aka spam. The volumes keep increasing even in the face of potential penalties, but user dissatisfaction is growing even faster, to the point where e-mail is threatened with obsolescence as users simply give up and rely on voice or instant messenger (even with its obvious temporal limits). Vendors, meanwhile, continue to negotiate about control and related issues. The big issue is trust, a commodity in short supply these days.

V is for virtualization. Enterprise hardware utilization is low, and expensive. Making computing resources more fungible can increase performance while lowering costs. Rhetoric thus far outruns reality, and there remain many engineering and economic issues before this cluster of technologies (pun intended) lives up to its potential. One example: if Oracle charges per processor, how does it figure my bill if I run 11i on a blade server, devoting varying numbers of processors to different tasks as demand rises and falls?

W is for Windows, still the world's most profitable software franchise. Security remains a major question mark, as does the issue of platform extension: how can Microsoft most successfully maintain look, feel, and branding across PCs, cell phones, game consoles, TV set-top boxes, MP3 players, handhelds, and home entertainment centers? Where does extension inhibit rather than enhance entry into new markets?

X is for XML. Specialized variants, including BPEL (Business Process Execution Language) and XBRL(eXtensible Business Reporting Language) are entering the basic toolkit of various professions and task groups. RSS is reintroducing "push" media. We're well past the hype stage and into the gritty details of implementation.

Y is for Yahoo! The fate of this company will dictate and indicate a great deal about the industry dynamics of what used to be called the Internet sector. Structurally speaking, is Yahoo closer to Disney or to Google? Is Amazon closer to Wal-Mart or InterActive Corp? Is eBay a tech stock in any meaningful way? Ultimately, how much of the essence of the Internet will belong to Hollywood or New York and how much to Silicon Valley?

Z is for zero latency, otherwise known as real time enterprise. Driven by compliance requirements, customer requirements, and competitors, often from unfamiliar sectors, businesses often confront "impossible" performance requirements that can't be met simply by tweaking existing processes and procedures. As with so many other technologies, the really tough part of real time is behavioral and cultural rather than engineering.

Friday, December 10, 2004


A run of recent back issues of Early Indications is available here. (Never mind the photo of my esteemed former colleague - the newsletter is and has always been my project.) If you want something earlier, I have a personal archive that's almost complete back to 1997.

Thursday, December 02, 2004

December newsletter I - Trends 2004: Prediction Scorecard

Despite using or implying many automotive metaphors regarding dashboards, headlights, and rear-view mirrors, the technology prediction industry doesn't spend much time keeping itself honest about its hits and misses. Given that every other economic segment is being subjected to tighter scrutiny, we'll do likewise: here's a look at how we and other predictors (harvested from multiple sources) fared this year.

1) A lot of these developments were well underway a year ago, so something like an upswing in flat-panel display sales was no surprise. Similarly, megapixel cameraphones were linear extrapolations of previous results and R&D investment. Continuing the "correct, but no surprise" category I'd list the following:

-Big money in search technology. The Google IPO and Microsoft investments were in the works for some time, and Amazon's A9, while fascinating, didn't change the established order of things
-RFID: Wal-Mart announced its mandate in mid-2003, so nothing snuck under radar.
-Increased investment in offshore capabilities - but John Kerry's campaign rhetoric made companies shy about public announcements.
-Price pressure across the tech sector
-Increased speed of cellular connections

2) A few predictions may or may not have come true, but confirming data aren't available yet:

-A 2x increase in the number of wi-fi hotspots. Anecdotally, I don't see anywhere near twice the coverage in my habit trails.
-Bluetooth, by contrast, feels more prevalent than it was a year ago, particularly in Europe. There, shipments of mobile phones with Bluetooth have nearly tripled in the third quarter vs. last year.

3) A few tech predictions simply didn't pan out:

-Enterprises have been wary of or even downright hostile to the idea of using peer-to-peer technologies for storage, distribution, and even cycle harvesting. Security, defined fairly conventionally and often oversimplistically, still trumps most discussions.
-Micropayments remain in the same category as voice interfaces: a good idea that people will embrace any time now. Except they don't
-Really Simple Syndication (RSS) is still overwhelmingly used for weblogs rather than enterprise content. There are some early adopters, but outside of media companies pushing news feeds, it's hard to find many instances of RSS in the general corporate population. Cisco, for example, uses the technology to distribute press releases.
-The Apple iPod completely dominated the MP3 player market; forecasts of credible competition and price erosion at Apple simply didn't happen. iPod shipments this holiday season will be counted in millions, Windows-based players in thousands.
-PCs and TVs, in the mass market, are different animals. HDTV is getting hot, but not with a lot of PC infrastructure.

4) The hits among the misses: predictions that appear to have come to fruition.

-Voice over IP is in hypergrowth mode, along multiple axes: corporate (Cisco), wild-card (Skype), old Bell (AT&T), ISP (Earthlink/Vonage), cable TV (Cox), and RBOC (SBC).
-Weblogging continues to increase, but the rate of abandonment is still high: about 1 in 2 blogs gets left to die after 6 months. The presidential campaign showed how blogs, in the aggregate, can be front-page news.
-Personal digital media storage (TiVo and iPod) continued to grow faster than most technologies. With small high-capacity hard drives and capacitive scroll-wheel interfaces like the iPod's now coming to cell phones, the market reaction will be fun to watch.
-Computer gaming in its many forms - portable, online, console- and PC-based - continues to grow dramatically, to the point where the industry is bigger than first-run Hollywood movies and also a cultural vector for norms and fads. Last month, for example, Microsoft's launch of Halo 2 brought in $125 million in sales - for the first day of availability.
-AMD capitalized on both good strategic bets and Intel's poor execution to make the general-purpose microprocessor market a two-horse race for the first time in years.

Looking closer to home, this list of six mnemonic "M"s summarizes what Americas CTO John Parkinson and I predicted last November for the 2004-07 period:

-Miniaturization: "mobile networking, presence awareness (primarily via GPS), and machine-to-machine data and communications standards will co-evolve with advances in miniaturization, with the result that each realm delivers useful capabilities to the others."

-Mobility: "we can create many new types of mobile capabilities – and make both 'location' and 'locality' key properties of our business processes. Especially in North America, expect a lot of these developments to incorporate links through vehicles – which, as well as being robust mobile platforms that can shield their occupants from radiation, have the supreme advantage of acting as electricity generators. . . . more and more network traffic well be 'Device-to-Device' with software agents (rather than people) communicating with each other; when you’re talking on the scale of hundreds of millions of entities, human intervention just isn’t a possibility. The other requirement will be for giant directories, much like today’s Domain Name Service, to keep track of what’s where, where things are headed, and who’s allowed to see what information about a given thing."

-Mesh: "We will need new network topologies that are highly connected locally but only loosely connected regionally and globally. Networks designed like this have many powerful properties (self diagnosis and automatic repair, for example) and can be made to be extremely flexible and highly scalable [and] . . . eventually they will become the topology of choice for just about everything at the 'edge' of the Internet of Things."

-Management: "There are two facets to the management issue. First, keeping track and taking care of all the emerging varieties of mobile and/or connected computing devices will require new approaches to asset utilization, system design, and maintenance. The sheer volume of the world’s digital data production . . . means that metadata will have to become more serious business. . . . The other aspect of management will empower people to make better business decisions. . . . The complete emergence of the 'Real Time Enterprise' will probably take longer than our five year horizon, but most of the technologies we need already exist – and significant productivity gains are already accruing to businesses that embrace the feedback and control processes that make “real time” ideas work. . . . Adoption of real-time ideas is more likely to be constrained by the need to learn new working habits – and to become comfortable with constant monitoring and feedback – than by the availability of enabling technologies."

-Microsoft: "Over 95% of personal computers run a version of Windows, an operating system written for a hardware platform defined in the 1980s. Hardware has made such strides since then that the software now can’t take advantage of many of the available features – and attempts to do so have created obvious complexity and significant, expensive fragility as a result. It’s time to do something about this – and sometime in 2006, Microsoft will do so. The next release of Windows, currently code-named Longhorn, will significantly simplify how things work and at the same time expose many new capabilities from which to assemble a whole new experience when interacting with the technology. . . . This will of course have both intended and unintended consequences, but it’s worth recalling how important Windows 95’s embedded TCP/IP was for the growth of the Internet: Longhorn could well help unlock change of similar magnitude."

-Maturity: "Many industry observers ask if the technology sector is 'mature,' meaning, are triple-digit growth rates and investment returns still possible, or do hardware and software increasingly have the economics of automobiles or frozen carrots? Our take on the question is slightly different.

Let’s begin by looking at the core characteristic of information technology (and most other classes of technology) over its first fifty years: IT was scarce, and commanded the concomitant price premium. . . . The operating assumptions and design principles of a connected, mobile environment now revolve more around standards than differentiation, around services on a network than applications on a processor, around quality of decisions rather than quantity of throughput.

In short, rather than leaving some golden age of economic alchemy – of turning silicon into gold, as it were – the IT industry now addresses its most educated base of customers, who are competing for the highest possible stakes with ever more powerful and flexible technologies. The second fifty years of IT history promise to be far more dramatic than the first, and given possibilities in biology, education, politics, and transportation just for starters, it’s impossibly premature to write off the information technology market as a locus for growth and radical innovation."

So how did we do? Overall, reading a four-year prediction one year in should provide plenty of wiggle room. Nothing jumps out as truly bone-headed: software agents are probably more utilized than talked about, for example, and miniaturization is certainly driving solid growth in cell phones and iPods. Research into mesh network topologies is proceeding apace. If anything the organizational barriers to real-time technologies are even higher than we warned of. Margin erosion at most IT vendors keeps the maturity question front and center. IT management - of assets, people, and portfolios - remains a primary opportunity for both hardware and software vendors.

On the other hand, Microsoft's announcement that it was leaving WinFS out of Longhorn in order to make a 2006 ship date removes some of that OS's projected impact; some have referred to the modified release as "Shorthorn." Some of the biggest news of the year, such as new supercomputers, came from the Open-Source world. Google, meanwhile, is building much more than a search tool, and even the search facet got short shrift in our discussions of metadata and directories. Security, often the simplististic versions noted above, is still driving more behavior than we predicted, particularly in the adoption of web services and grid technologies. Mobility remains primarily a consumer rather than enterprise phenomenon. Overall, give us a B.

What do I think is coming for 2005? That's a topic for the next issue.

Wednesday, December 01, 2004

November newsletter: Ten Years On

Last month, Silicon Valley's Churchill Club held its annual Top Ten Technology Trends dinner, featuring VCs John Doerr from Kleiner Perkins, Esther Dyson, Roger McNamee from Silver Lake Partners & Integral Capital Partners, and Joe Schoendorf from Accel. Doerr started things off by positing that the internet was in fact underhyped. The browser is over ten years old and showing its age, but there's a burst of innovation in such areas as search, localization and mapping, and messaging. What he called the NextWeb (or Web 2.0 in others' nomenclature) has enormous potential. For example, Google recently bought Keyhole, the spectacular mapping site that does truly breathtaking things with satellite images; here's a free trial version: There's a lot to think about if you combine Google's search acumen with this dazzling visual capability.

History suggests Doerr is right. Carlota Perez, an economic historian at the University of Sussex, studied the persistent patterns underlying five techo-economic eras: the industrial revolution; steam and railways; steel, electricity, and heavy engineering; oil, cars, and mass production; and information and communications technologies. Each of these sometimes overlapping periods has followed a rough sequence of four phases: a new technology appears in the market, often disrupting existing arrangements. There's a period of rapid, often silly adoption from which a bubble emerges. After the speculative excesses burn off, the economic potential of the new technology is explored and exploited, until such time as the market matures and the potential is exhausted, often by a new technology paradigm. Speaking of "paradigm shifts," Perez's work might be seen as a close, and useful, parallel to Thomas Kuhn's classic The Structure of Scientific Revolutions.

Doerr is also probably right because we're still too close to the original event to have much perspective. Most of us can still remember seeing our first website or sending our first e-mail, and much of the first wave of any innovation typically automates existing practices: early cars were horseless horse carriages, television initially broadcast actors reading radio scripts, and oil from the Pennsylvania boom was first used for lighting more than locomotion. It's no surprise that the 1990s Internet accelerated existing processes of finding things: travel-related information, materials specifications, and undifferentiated goods like books or stocks. Now, technology promises to help people change many other activities. Compare how far we've come in a decade to the half-century it took for the automobile to reshape America's cities after the passage of the Federal-Aid Highway Act of 1952 that initiated the Interstates that both partitioned the inner cities of the North and created suburbs abound every major city.

We are now living through our own variety of unexpected consequences, including whole new processes invented in the past ten years that could not have occurred at any prior historical moment. Here's a short list:

*New economics of production and distribution
Linux and other forms of open-source software could not have been developed and tested with postage rather than electronic communication. Hedge funds and other hair-trigger financial arrangements exploited instant messaging and cheap computation; some have suggested the first chapter of this story is the invention of the spreadsheet about twenty years ago. Online movement of digital music (and soon video) is forcing Hollywood to reinvent its core business model and trading practices. The migration from proprietary to open networks and protocols is forcing some of the world's oldest and most visible companies - AT&T for starters - to completely reinvent their cost structure, go-to-market position, and customer set.

But as an excellent survey in the November 13 Economist illustrates, perhaps the most powerful economic upheaval caused by our technology is in the globalization of white-collar work. Detroit and Pittsburgh long ago lost their primacy as the world's factories; heavy industry has been migrating offshore for decades. What's shocking now is how fast such skilled positions as call-center troubleshooter, radiologist, and R&D engineer are moving to India, China, and the former Soviet Union. As with Linux, this trend would be severely inhibited if the Internet had not fueled education, communications, and aspirations in the developing world. The article points out that Toyota must allow between 25 and 37 days for an engine to move from Nagoya to Chicago, but communication of white-collar "products" faces very different and smaller uncertainties. For example, if a manager includes breaks for Thanksgiving, Christmas, and New Year's in a 12-week project plan, when are the holidays in India, Canada, or China during that same time span? If code is being written, how can the U.S. customer be assured that the offshore test setup is identical to the deployment environment?

There are some who think that this speed factor will accelerate the globalization of work, and indeed, a new report from a bipartisan Congressional commission, to be released in January, estimates the U.S. 2004 job loss at slightly over 400,000 - twice the estimate of the national Chamber of Commerce. While any estimates of job migration will be subject to wide variation in interpretation and credibility, 400,000 is nearly a quarter of the 1.8 million new jobs thought to be created in the same period, a number big enough to make a difference. Economists are busy fighting over whether migration of work to low-cost producers ultimately creates more and better new jobs higher up the stack in the country that lost the jobs, but nobody has conclusive results. A number of facts are beyond dispute: 1) workers are not immediately fungible (a call-center operator in Omaha typically can't answer a help-wanted ad in Sacramento for oncology nurses); 2) the process of job migration and invention takes time and ripples throughout an economy in unpredictable ways: education, health care, housing, and numerous other areas feel the impact; 3) low-cost producers face inflationary pressures (15-17% annually for wages at Wipro and Infosys, for example) and have limited barriers to erect against subsequent global competition. All in all, there's a lot of disruption still ahead of us.

Finally, as the Economist survey rightly insists, even though you can use one to do the other, there are substantial differences between sending work and/or jobs overseas (offshoring), and hiring someone else to do them (outsourcing). For example, Steven Bigari is a McDonald’s franchisee who used outsourcing of a business process - order-taking at the drive-through window - to increase the efficiency of his 12 McDonald's locations: drive-through order time dropped by 30 seconds to a little more than one-minute per transaction, which beats the chain's average of two-minutes thirty six seconds. Drivers are actually talking to a call center in Colorado Springs rather than to an employee at the restaurant, which are located in several states including Colorado. Doing so has helped increase the number of cars his drive-throughs handle by 15%, from 226 cars per hour to 260 cars. Other franchisees are buying the service from him as well.

The distributed, firewalled nature of radical Islamic cells makes them difficult to monitor and infiltrate. Ad-hoc modes of coordination, using principles similar to those underlying packet-based networks, prevent a simple quest to kill off the head so the body will die. What the defense analyst John Robb calls "global guerrillas" are also students of network theory in their target selection, particularly in regard to social networks (especially foreign infrastructure-support workers), the power of publicity, and oil pipeline topologies. It is no surprise that Gulf War II is behaving very differently from the first such conflict, if only because of the technologies involved.

From the U.S. perspective, re-orienting the armed services both to fight non-nation-state actors (albeit highly coordinated ones) and to fight so-called network-centric warfare (NCW) against more conventional enemies has been difficult. The very fact that NCW cannot be defined succinctly or consistently starts a long list of issues; other factors include the "revolving door" between the Pentagon and its suppliers that creates habits that are tough to break, and the hardened career paths that culminate in the command of very large and not terrifically mobile fighting assets. Military prestige continues to collect near the likes of aircraft carriers rather than squadrons of drones or software agents.

The iPod as a form factor is changing how people hear music, books, and other material; I'd probably rate it the most important tech development of the century so far. The mini-movies at BMW and Amazon, or the non-studio breakthroughs like 405 or Homestar Runner, give us a taste of what might come next, but it's hard to see the endgame in these early developments that have yet to truly unseat TV from its prime position. Many people predict photography will be the next "killer application" for the Internet, and Kodak's sad fortunes already testify to the technology displacement that's underway.

The daily newspaper is under heavy, multi-pronged attack: job seekers migrate to Monster, sports fans to ESPN and CNN/SI, stock-pickers to Yahoo! or any of a dozen other sites, car-researchers to Edmunds, and hard-news addicts to either serious outlets like the New York Times or BBC or to highly focused weblogs. Hobbyists that get only a nod from the dailies (gardeners, cooks, chess players, classical music enthusiasts) can get a firehose of information directed at any particular interest. Personal ads are similarly moving to consolidated Internet operators like, Yahoo, and e-Harmony. Where does that leave the mid-market daily paper? The fact that so many print media companies have been caught lying about circulation is no accident: many are caught in a race to see who will leave first, a critical mass of advertisers or readers.

Here's another reminder of how early we are in the Internet's lifecycle. In his book The Rise of the Novel, Ian Watt shows how the narrative form we now call the novel arose only after cheap printing presses and generalized literacy had made broad advances in a nascent middle class. Before Gutenberg, oral traditions gave the West epics and the Bible; it was impossible for authors to get an audience without the literal wrist-power of the clergy. Thinking in century-long cycles, what kinds of literacy and literature will we or our progeny invent to stand alongside the 30-second TV spot, the two-minute-thirty-second AM radio single, or even the sonnet or the epic?

*New forms of social interaction
Flash mobs, whether on line or connected by mobile phones, rely on the many-to-many capabilities that pen and paper and wireline telephony cannot mimic. Community, a word I've always distrusted in the Internet context, is in fact emerging as people meet and learn about each other online, to the point where a strong desire for physical meetings is not uncommon. Fark, a satirical site featuring news and user-driven commentary, has regular meet-ups in cities across the US and often Europe. Fine Homebuilding, a magazine aimed at high-end builders and architects, has a superb web operation in which tradespeople swap tips and approaches; earlier this summer 70 people, many of them physical strangers, met in person for a lobster cookout in Jamestown, Rhode Island. Webloggers frequently convene, and document the convocation, when physical schedules overlap.

Online interactions don't merely automate existing traffic. Think of how the cc: field is used and abused in corporate e-mail, then imagine trying to generate and distribute that many physical memos. Or as the ever-astute Clay Shirky argues, while it's common to think of flaming in online communities as a bug in the system, he says instead that such behavior is an innate part of the architecture, albeit an unexpected and not particularly pleasant one. We as an industry (broadly defined) are still learning how software behaves in conjunction with both particular users and groups. Massively multiplayer games, or worlds, are examples of environments that were inconceivable only a few years ago, and that do not automate any known pattern of interaction. The same can be said for eBay auctions, for that matter.

Clearly a lot of the utopians' hopes for the Internet to enrich their lives and their politics are silly, or worse. At the same time, webloggers both as news reporters and as fact-checkers (in a strong parallel to the open-source development model) had a marked impact on the recent election. Regarding news reporting, it was a dismal performance, one not approximated by the tired old media that waited for verification from real rather than phantom and/or wish-fulfilling sources. Dan Gillmor's We the Media (a book about bloggers as news sources) may be 60, 70, or 80% wrong - but it would be folly not to watch the trends he's describing to see what happens with regard to the parts he got right.

In a quarter-to-quarter business timescape, it's often difficult to assess these sorts of long-wave transitions. That is, unless there's a spike in some metric of usage, it's hard to realize when we're in the midst of change that will take decades to play out. The market for prediction remains more robust than that for assessment. Exercises in reflection are valuable, but the first one that comes to mind - James Gleick's What Just Happened - was disappointing. The alternative, for the moment, is to suspend consciousness long enough to remember what the world looked like before Amazon, before Google, before AOL - and try to make sense of where the road could lead from here.

*Carlota Perez, Technological Revolutions and Financial Capital: The Dynamics of Bubbles and Golden Ages (ISBN: 1843763311)
*Economist survey of outsourcing:
*2004 offshore job losses: Kimberly Blanton, "Offshoring accelerating," Boston Globe, November 15, 2004
*McDonalds process outsourcing: New York Times, July 18, 2004 and
*For more on outsourcing, see also Tom Malone's book, The Future of Work, which may be more accurately called "the future of organization structures") and
*John Robb:
*Clay Shirky: