Wednesday, October 24, 2007

October 2007 Early Indications II: Ten big technology-related busts in the past ten years

Earlier this month we marked ten years of this newsletter's publication by noting ten developments that quickly permeated the market after being nonexistent or invisible in 1997. This time out, I'll list ten big failures that at one time or another looked like can't-miss propositions.

1) Online grocery

Grocery is a notoriously tough retail category, with thin margins, fickle and price-sensitive customers, and perishable inventory. At the same time, it's an enormous market -- absolutely everybody eats -- so in the late 1990s, the perceived invincibility of online grocery made for failure of dramatic proportions. Webvan combined aggressive expansion, a long leash from investors, and questionable management to create an $800 million sinkhole. The firm was operating in Chicago, Los Angeles and Orange County, Portland, San Diego, San Francisco, and Seattle at the time of its demise, and many customers were disappointed at the loss of a convenient, time-saving service, particularly after Webvan undid many of the successes of the HomeGrocer chain it acquired. The customer base remains tantalizing, particularly as commutes grow longer and free time shrinks, but the logistics of automating picking out a cart-load of groceries from among 200,000+ SKUs, some fresh, makes this a daunting entrepreneurial challenge.

2) AOL and Excite@Home

For a time, AOL ruled the world of dial-up Internet access. Its carpet-bombed floppy disks (later CDs) helped introduce millions of Americans to the Internet, or at least an isotope thereof. It combined access with content (in some measure, in the form of other people) to reach an astonishing price/earnings ratio of 700. But when broadband delivered by incumbent telcos and cable companies split AOL's access from its content, the supposed synergy broke down and the bubble burst.

Beginning slightly later than AOL, the Excite search engine (like Yahoo and Google, a Stanford creation) was bought by the @Home broadband startup in hopes of another content+pipes goldrush. The merger was a disaster: $7 billion of market capitalization vaporized. Cox, TCI/AT&T, Comcast and the other cable companies who owned physical plant and had operational responsibilities, were ill matched with the Silicon Valley engineering culture that emphasized features and glamour over reliability and customer service. That Kleiner Perkins owned stakes in both @Home and Excite compounded the enthusiasm for a rush to synergy, but the operational realities of rebuilding physical infrastructure, combined with the regulatory scrutiny drawn by @Home's proprietary relationships with one of several competing portals, meant that the cultural and leadership issues helped precipitate a train wreck of epic proportions in 2001.

(On AOL, see Kara Swisher, There Must Be a Pony in Here Somewhere (2004); on Excite/@Home, see Frank Rose, "The $7 Billion Delusion")

3) Iridium

Motorola was a major shareholder in and primary supplier to this satellite telephony venture. After its 1997 IPO, Iridium faced loan covenants that required it to sign up 213,000 customers soon after it began offering service in 1999. When only about 10% of that number materialized, Iridium filed for bankruptcy: $5 billion in assets was liquidated for $25 million, and only last month Motorola -- itself Iridium's largest creditor, to the tune of $2 billion -- appeared to have escaped further liability with a court ruling in New York. The service was never aimed at a mass market, with phones costing $3,000 and calls $7 per minute. Coverage was good in open oceans and deserts, but not in moving cars or cities -- and the handset, while technically sophisticated, was big, heavy, and sported an antenna "the size of a toothbrush," in the words of the Wall Street Journal. Satellites, meanwhile, have been similarly costly to rival radio providers XM and Sirius, which between them have accumulated historic losses of $8 billion and are now trying to merge.

4) Super Audio Compact Disc/DVD-Audio

Roughly 20 years after the launch of the compact disc audio format, which itself came about 35 years after the introduction of the LP record, the entertainment industry brought out competing high-resolution optical disc formats for audio. Sony and Philips introduced SACD in 2000, while the DVD Forum, led by Panasonic and Toshiba, brought out DVD-A at about the same time. Audio quality is much higher than CD from both formats, but market confusion has been a major limiting factor. Customers of a certain age who already had to buy music collections twice over were reluctant to commit to one of two competing formats, and while hybrid players now support multi-channel audio playback from either source, software is not widely available: artists and labels had to bet on one standard or the other, and the slow market penetration has resulted in relatively few, and expensive, titles being available. The format war coincided with the explosion of digital file sharing (hence strict and cumbersome copy protection schemes for both SACD and DVD-A), and customers have widely defected to portable, lower fidelity media such as MP3 files. The net result is that both high-resolution audio formats are essentially irrelevant, and the DVD standard itself is in the early stages of a similar format fight, with potentially similar results.

5) Quokka Sports

Rereading ten years of Early Indications and its predecessors, I was struck by how amazed I was by three or four software demos. One was Keyhole, the technology that became Google Earth after the company was acquired. Another was Quokka, which was devoted to delivering data-rich sports coverage over the web. From its origins in Australia, Quokka began with immersive feeds of long sailing races such as Sydney-Hobart: data relating to biometrics, meteorology, speed, absolute and relative position, and participant narratives made for engrossing viewing. Quokka bought the Internet rights to the Sydney Olympics in 2000 after moving to San Francisco, but the lack of a viable advertising model combined with common dot-com management failures to force a shutdown in April 2001.

Partnerships with NBC and Major League Baseball, along with further Olympic rights, cost money but failed to deliver returns. In retrospect, Quokka was probably better aligned with low-viewership sports like sailing and mountain-climbing that could find Webcast niches than with big-audience events with established television techniques and politics. Sports remains unevenly instrumented: NASCAR races are data-rich, but the single biggest predictor of a pass play's success on a football field -- how long the quarterback holds the ball -- is not recorded. Baseball, meanwhile, has generated hugely popular online fantasy leagues, with football following suit, in ad-supported models of which the Aussies could only dream.

6) OpenFund

If open-source works for software, why not try the model elsewhere? MetaMarkets, founded by two veterans from Barclays Global Investors, launched in August 1999 on the basis of full transparency as fund managers disclosed every trade, often with commentary. The fund started fast out of the gate: at year-end 1999, it was up 91% (by comparison, the NASDAQ was up nearly 50% in the same period). The fund fell 42% in 2000, and dropped another 26% between January and August 2001, when it shut down. In part, the fund was a victim of small scale: whereas most mutual funds need to run at least $100 million in assets for viability, OpenFund was at about $10 million when it was liquidated. Both management and critics compared OpenFund to a finance chatroom with real money: a Morningstar analyst noted after the fund's demise that "the entertainment, the gimmick, doesn't really have anything to do with investing." This sounds plausible: if my money is in free fall, I'm not sure chatting with the fund managers is going to help either my mood or the fund's performance.

7) General-purpose Speech Recognition

Ever since at least 1997, Bill Gates has been predicting that speech recognition will be an integral aspect of the PC experience. In his 5-to-10-year timeframe, it never happened, but not for lack of trying: Dragon Systems, headquartered in the U.S., was losing money selling speech recognition software before it was bought by Belgian competitor Lernout & Hauspie in the spring of 2000, just after L&H paid $1 billion for Dictaphone. The Dragon founders, however, had the misfortune of watching their company go into reorganization after accounting irregularities made the L&H stock worthless. Revelations of fictitious transactions in Korea and over-stated earnings elsewhere eventually sent the L&H founders as well as CEO Gaston Bastiaens (an industry veteran who helped launch the compact disc at Philips and later worked on the Apple Newton) into criminal proceedings that remain ongoing six years later: before Enron, Lernout & Hauspie was the archetype of corporate scandal. ScanSoft, which made optical character recognition products, bought the assets, but even now, neither Nuance (as ScanSoft renamed itself) nor Microsoft has made speech interfaces work for general-purpose computing. In vertical domains, however, speech interfaces -- particularly telephonic customer service and medical transcription -- are working well.

8) Digital Appliances

From high-profile efforts at Oracle (the NC) and Sun (JavaStation) to consumer efforts from the likes of Uniden, the late 1990s witnessed a variety of efforts to displace the personal computer with a network-intensive, easy-to-use, easy-to-manage device. The ideal of plugging a device into the Internet without need for hard-disk-resident applications or storage was motivated by a variety of factors, but ten years on, the vision has yet to catch on. For one thing, wireless devices allow much of the NC's functionality to be experienced on the go (cf. the Blackberry). Terminals and emulators never left the list of enterprise alternatives, as Citrix-based Windows systems illustrate: the PC remains a flexible platform that can be configured into diskless, mobile, or other alternatives. The relentless improvement in PC performance, particularly from 1990 until 2002 or so, made the PC's price-to-performance ratio continually appealing, until processing began to outstrip most of the application stack's needs. Finally, the lack of true broadband, until recently, made the devices slow in many environments.

9) Business-to-Business Exchanges

Talk about a shakeout: from 1520 exchanges in 2001, only about 10% were still active only two years later. VerticalNet, one of the first B2B exchanges, had 1700 employees and a $10 billion market capitalization at its peak; shortly afterward the CEO was faced with keeping 50 people on the payroll, using about $11 million that remained in the bank. Covisint, designed to make automobiles parts-buying more efficient, had a similar fate. That both survive today, albeit operating at minute fractions of their projected volumes, illustrates that while business-to-business commerce is huge, it is also difficult to reinvent.

Sellers stayed on the sideline as auction models presented the specter of purely price-based competition. Buyers, while wanting the price leverage, also realized that a) customer service and relationships matter and b) that bankrupt suppliers (as in the auto industry) are not in the buyers' long-term interest. Many exchange providers turned into merchants of purchasing efficiency inside the firewall, relying more on software and process expertise than on convening power. Running a market is also not necessarily attractive: as this newsletter noted in April 2000, in 1998 the New York Stock Exchange only made $101 million on 169 billion trades totaling $7.3 trillion.

10) Business Models Based on "Free"

At one time, at least two dozen Internet Service Providers offered free connections, usually over dialup. Free-PC was one of multiple attempts to get consumers to watch ads in return for hardware. Netscape famously gave away browsers to sell server software, a strategy that backfired for a number of reasons, one of which was Microsoft's anti-competitive behavior with Internet Explorer. Stocks in VA Linux, a company with real hardware sales but ample "free" hype, rose from $30 to $320 on December 9, 1999, the first day of trading, but later fell to 54 cents in July 2002.

More recently, eBay has encountered major difficulty making Skype pay off; Sunrocket and other VoIP providers are either shuttered or weathering tough times. There are also many businesses that have been collateral damage in free scenarios, some of them illegal or otherwise of dubious ethical standing. Music companies that have been slow to respond to file-sharing with appealing alternatives are the most visible of these. Even so, it has been repeatedly proven that you can in fact "compete with free" and in fact usually win.