Tuesday, February 28, 2006

February 2006 Early Indications II: Demo trip report

Earlier this month I spent several days at the 16th annual Demo conference, where 68 emerging companies and technologies launched and/or presented. Both bloggers and mass media outlets have fuller coverage (see below), but I wanted to point to a few important companies and play out some of their implications. If you want to see a company's pitch for yourself, Demo has videos of all the participants.

1) Distributed infrastructure
One of the trends that's having a significant impact on the global economy is decentralization. Apart from a relatively few activities - mining, manufacturing, and medicine - many aspects of productive infrastructure can migrate to one's laptop or thereabouts: newspaper printing presses, recording studios and CD-pressing factories, video production suites, travel agency ticket printers, call centers (as with Jet Blue), banks, phone companies, and photo labs are just a few examples. In about 10 years, each of these very expensive ventures has been replicated and in some cases undermined by a digital equivalent. Demo had a few more companies in this vein:

*Blurb allows customers to design books with photos and (unlike Ofoto and iPhoto) words. The custom-printed volumes are reasonably priced and handsome to look at. The company is finding a large market in the charity cookbook industry, 400,000 titles of which appeared in 2004; some Junior League "Taste of" cookbooks sell 25,000 copies, which would dent a best-seller list.

*The big media draw of the conference was Moobella, a make-to-order ice cream factory the size of a large soft-drink vending machine. In 4 minutes a customized cup of ice cream appears, and by all accounts tastes great. It's not clear how the economics will work (vs. a vendor cart for example), but the team solved some tough technology problems and prompted a lot of thought.

*iGuitar is a nice-looking electric guitar that's a USB peripheral. It obviates the need for lots of midi and outboard gear on the guitar-to-computer front, but also feeds a whole orchestra of synthesized music, to the extent that guitar-only pros are scoring TV shows formerly serviced by keyboard players.

*Locamoda connects cell phones, IVR systems, the Web, and physical storefronts in a powerful way. One application is in real estate: you're walking down the street and see a condo listed at an agency. You can interact with the 42-inch plasma screen via your cell ("press 9 to see an exterior shot"), send yourself a hyperlink, or leave voicemail for the real estate agent. The other play for this company is social computing - in space. Imagine going to a bar and texting a message to a physical screen where everyone can see the message. thetruth.com, the antismoking outfit, is testing these at skatebard (pun intended) parks to capture "wifiti."

*EQO (say "echo") connects mobile phones to Skype through a very clean interface. The economic implications of this kind of service are staggering if you're a voice-centric international carrier: VoIP is a big enough issue when the user is chained to a PC or home network, but turning the connectivity loose affects still more incumbents.

2) Search
As others have noted, living in a networked world where information is shared via standardized technologies changes how we find information. There's definitely a whiff of "me-too" in the industry right now: every startup dreams of a Googlesque IPO as their exit strategy. (Check out this videoconferencing company's homepage if you doubt my read on the zeitgeist.) At the same time, the presence of search as a pillar of both Microsoft and Apple operating systems, as a key to both on- and off-line shopping, and as a first-impulse information gathering strategy means that there is ample room for innovation, the human capital arms race in the search industry notwithstanding.

*Krugle is a search engine for code - it preserves sourcecode formatting rather than text strings. It also keeps tags of a whole session that you can keep as a trail of breadcrumbs to forward or post. Reuse has been a cherished ideal of software engineers for decades, but this approach makes reuse far more practical than any formal tool I've seen.

*Kosmix was co-founded by Anand Rajaraman and Venky Harinarayan, who came from Amazon after founding Junglee, which we wrote about in 1998. They're building vertical search in consumer areas (starting in health, travel, and politics). It looks like a company that will do well - their target markets are already generating deals.

*Transparensee does structured search: the demo was of real estate databases. Instead of having to requery after retrieving a set of findings, you use slider bars to re-weight categories to refine or expand results. If you want a $300,000 4-BR house in zip code X, let's imagine nothing comes up. The tool allows you to increase the price, decrease the BR count, or expand zip codes. The smooth interface and intuitive refining of the search were pretty appealing.

*Nexidia is building video search (which addresses a screaming-hot market: 20 billion streams were delivered in 2005, and 2006 looks to quadruple that, at least). The company uses phonetic indexing, which runs at 60x real time, and phonetic querying, so it's somewhat language-independent. The speed and accuracy of the search results got very high marks.

*Yet another collaborative filter for music comes from garageband.com, which focuses on "if you like this, try that" for independent bands and genres; it also pushes new music to your playlist as the user community discovers new favorites. Garageband also provides hosting and generally does a wider job of supporting the indie ecosystem than merely generating playlists. (Music Genone Project (pandora.com) is better and farther along on the recommendation front; musicplasma is quicker and dirtier but lighter weight).

*Riya is a pretty hot company, apparently, and focuses on image searching - very successfully, to the point where it can do facial recognition with scary-accurate results (it convinced me I want no photos of my kids on line, names or no names). It also matches text that's captured in a photograph, even sideways in the background, like on a book spine.

3) Data
Finding ways to manage, secure, and exploit information overload at both the individual and corporate levels is generating some useful technologies. Questions like "who owns the data," "how do I know this information is correct, up-to-date, and/or calculated appropriately," and "compared to what" are getting considerable attention.

*Cnetchannel.com uses C|Net's huge database of technology attributes to do smart up-sell positioning on web pages. The demo showed a laptop on a catalog webpage with upsells of a) memory that wouldn't fit, b) a wireless card for a machine with built-in 802.11, and c) maybe a bluetooth mouse for a PC without that capability (I forget exactly). The "after" matched the peripherals and accessories much more closely and, the company would assert, profitably for the merchant. The company also wants to use the underlying matching and control capabilities (there's a control console for MBA types to use) in other domains.

*Zimini is an attempt to be a next-generation couponing engine, but it's not quite clear how coupon-users will respond to yet another helpful online service asking them to "tell us a little bit about yourself" to drive the customization.

*Kaboodle is a "collaborative shopping" service, where you dump items you find into sharable pages; it's also a handy way to organize web shopping where you're comparing price/features across sites or pages. It's easier to like than to describe.

*Panoratio began as a project within Siemens to manage and move enormous data stores from the sensors at power plants. They have what are called PDIs: Portable Data Images, which are compressed but readily navigable. It isn't highly graphical (it still looks a lot like Excel) but the navigation is still better than what you get from your everyday spreadsheet.

*Bones in Motion collects GPS and time information to create online exercise diaries, which can also be published as blogs. It correlates effort across cities (Denver's altitude vs. Charleston's humidity) so users can compare training programs and share training courses: I could run or cycle for X effort in multiple cities using shared and equalized routes. It's a great example of capturing information as byproduct and organizing it. Look for it on Sprint mobile.

*Vivid Sky's story begins with a UPS-grade ruggedized handheld that you rent at the baseball stadium. From it you can watch video highlights throughout the game, order concessions, participate in online contests and surveys, order tickets, and check statistics, scores, etc. Pilots will be deployed this summer, and word is there will be football action later this year.

*Several companies played the tagging angle, in which users or communities contribute metadata to organize some body of content (Flickr is a great example). Tagworld is a bit like MySpace, but there's a marketplace piece as well. The demonstrator said it was trivial for a 14-year-old to buy things online, some from other members' classifieds, but that begs the question of what a 14-year-old is doing with a credit card online in the first place. Draper Fisher Jurvetson just funded these guys.

*Sproutit uses tags in an email-based tool for small teams of about 10 people. The objective is to make individuals less critical for customer interaction: if I as a co-worker know your context for a given interaction, I can cover for you when you're out of the office. In a similar vein, eeminder gives mobile workers very low-latency access to corporate data.

It's fitting for conference producer, and host, Chris Shipley to have the last word. In her opening remarks she made two salient points about this year's crop of demonstrators. First, the differentiation between enterprise and personal technology gets fuzzier every year. Second, the industry desperately needs to address the complexity issues that prevent more people from using more products and services. As she said in her keynote,

"Who needs more buttons and features and options – on just about any product? Can you seriously say that you've used all the capabilities of any of the software or devices that you already own? Do you really want more?

"Unless, as an industry, we commit ourselves to a better user experience, clearer choices, and greater value, I am afraid that a many people may just sit out the market. Needing no more new features, being unable to sift through any more search results, being overwhelmed by options, these individuals are going to stop, or at least slow down, the acquisition of new technology."

Food for thought, even tastier than the ice cream.

Other coverage:
Information Week

PC Magazine


Tuesday, February 21, 2006

February 2006 Early Indications I: The Price of Attention

Four data points from the past several weeks:

- Bucking the trend against investor wariness in the tech sector, Vonage filed for an IPO of $250 million. The SEC documentation shows that the company's acquisition costs are running roughly $214 per customer (up from $137 in 2004) even as average monthly revenue has dropped from $31 to $26.63. The churn rate has also increased from 1.7% in Q1 05 to 2.26% in Q3 05. Analysts are concerned that the service will struggle once big cable operators begin to compete seriously in consumer VoIP: the $250 million, minus commissions and fees, will barely cover one year of loss-making. The company has lost over $310 million, net, since its inception, about $190 million of that in the first three quarters of 2005.

- Satellite radio has evolved into a two-horse race between XM and Sirius, which as a company is younger but boasts stronger on-air talent (Howard Stern) and management (in the person of Viacom veteran Mel Karmazin). Both companies reported losses last week: Sirius pegged subscriber acquisition costs at $113 per subscriber for the fourth quarter and $139 for the year. XM had earlier reported an $89 cost per new customer, and a board member had quit, citing concerns over marketing and programming expenditures. According to a letter quoted on Bloomberg, Pierce Roberts said that "Given current course and speed there is, in my view, a significant chance of a crisis on the horizon." XM stock has fallen by roughly 40% since October.

- A new Gallup poll tracks blog-reading as it relates to other Internet activities, and the pollster's conclusion is that despite a continuing increase in the number of weblogs being produced, readership stayed flat in 2005 after a quick run-up to 20% readership in 1994 (9% of those surveyed read "frequently", 11% "occasionally"). By contrast, results of the same survey assert that 72% of Web users check news and weather on a regular basis, 52% shop, 40% pay bills, and 28% play games. The 9% of heavy users closely tracks the Pew Internet and American Life figure of 7% who reported reading a blog "yesterday" in September 2005.

What conclusions might we draw from the behavior underlying these findings and results? First, it's hard to imagine AT&T or NBC in their early development calculating customer acquisition costs. XM and Sirius are caught in between the expensive realities of a network build-out much like the aforementioned companies' and the standards-based Internet that lets voice, video, or other information services ride effectively free on existing infrastructure. The satellite broadcasters have to simultaneously conduct a 20th century battle on the capital investment and engineering front and a 21st-century conflict as they fight for attention with Net-based media.

Second, novelty will always draw a splinter population; the challenge, to use Geoffrey Moore's time-worn (but time-proven) phrase, is to "cross the chasm" between early adopters and the meaty part of a bell curve distribution. Nobody has figured out how to measure podcasts, for example, but one could easily hypothesize that the growth in blog-reading that didn't appear in poll data was siphoned off by the newest new thing.

Third, we tend to frame competitive landscapes in terms of obvious comparisons. XM and Sirius are outspending each other trying to win what they construe as a winner-take-all game, but several equity analysts have wondered whether podcasting is the real disrupter in the radio market. If that turns out to be the case, the hundreds of millions of dollars spent launching satellites and subsidizing receivers will have been undermined by an existing base of music players fed by cheap microphones and Internet connections.

Fourth, it's not clear that anyone understands how new media track between the big broadcast model (an oligopoly of well-capitalized networks that control infrastructure and talent) and the "audience of one" or narrow-cast model. Millions of people apparently want to watch American Idol or listen to Howard Stern, but a) they may not want to do it on the broadcaster's schedule or b) they may be doing so in a multi-networked mode: watching, chatting on line, and talking on a voice connection. There are signs that such crowd favorites as Seinfeld and more recently "24" are inspiring more communal, interactive modes of viewership that nobody knows how to measure or explain. In such a world it's easy to see the appeal of a Google or Yahoo ad model in contrast to Nielsen ratings, which feel inadequate to the task of measuring time-sliced viewing habits.

Finally and most important, supply and demand are at work. If something is available in glut, its price will drop relative to something scarce. With effectively unlimited information choices but only 24 hours in a day, people have control of the scarcity, which appears to be attention. Vonage is not only contending with Skype or Comcast; its success will measure, to a degree, how willing people are to spend money on talking as opposed to communicating by e-mail, playing basketball, or going out to dinner. To illustrate attention as scarcity, consider that at the same time that blog-reading stayed flat and podcasting exploded, instant messaging declined in popularity, to the extent that Gallup data are accurate.

While it's possible to buy more cars than one can drive, or more clothes than a person will ever wear, it's impossible to do more things than there is time do do them in. One of the key differences between a good and a service is that services are perishable. Lots of attention has been focused on this issue from the supply side: hotels and lawyers worry constantly about utilization, because every unbillable hour or empty room implies lost revenue that can never be recouped.

Perhaps it makes sense to look more carefully at the dynamics of consumption in a services economy: what are the constraints to information services, in particular, that the 24-hour day imposes? As highly as I may value mudbaths or psychotherapy, or podcasts or phone calls, there's a finite quantity that I can consume. What are the dynamics of partial attention? How can both providers and consumers of information services establish value and price parameters given bits' abundance? How will multitasking be measured? The answers, however far off they may be, will affect everything from network engineering to application design to the valuation of any attention-dependent company.