NEW MEDIA OFTEN IMITATES OLD MEDIA before it finds its true footing. The early days of television recast radio programming, complete with voices, actors, scripts, and sounds, but adding the important visual element. Many of the original Internet sites were built straight from old-media analogues, from magazines on electronic news-stands to stores in now-obsolete Internet shopping malls. What is less appreciated is that sometimes, new transmission mediums imitate old mediums before they come into their own. Much-vaunted broadband applications have been slow in coming, with most broadband users trying to get to the same web sites or email servers, just faster.

Before a given transport medium can truly create applications unique and suited to itself, it must first achieve a certain ubiquity. The expense of creating content for a new medium often requires critical mass, and communications applications in particular require high adoption rates before they become practical. For example, Friendster and similar social networks were difficult to pull off even a few years ago, because the majority of any given network of friends was offline. Now, any online user is likely to belong to a peer group that is largely online.

Similarly, broadband applications like P2P and video-conferencing require a large number of broadband users before they are useful (a much greater percentage for video-conferencing than for P2P file-sharing, however).

One rapidly emerging new medium has, so far, failed to develop any new mass-market applications, and its users treat it very much like its predecessor. 802.11(Wi-Fi), with its free spectrum and near-free silicon provides a high-bandwidth mobile experience, but has been used as an easier-to-install replacement for its wired cousin, 802.3 (Ethernet). However, in a few places where ubiquitous Wi-Fi access is freely available, a number of interesting applications and behavior patterns have begun to emerge, and give us a glimpse into the future of networking.

In the late 1980s, Dartmouth College was the most wired campus on the planet, running 10Mb Ethernet into every dorm room. Today, Dartmouth is the most unwired campus on the planet, with 560 access points covering 200 acres. The recent Unleashed conference on wireless provided an opportunity to examine this new outpost of the future. The wireless network here is pervasive—Larry Levine, the head of computing services, challenged attendees to find a single spot on campus and surrounding areas that did not have 802.11 coverage. Even the boathouse, adjacent sections of the Connecticut river, the ski lodge, and sections of the ski slope are covered!

If you wanted to know where wired communications were headed in the late 1980s, all you had to do was go to the Dartmouth campus and look at their homegrown email application, Blitzmail. As any regular user of Blitzmail will tell you, it included a server-side address book and remote private and public folders before almost any other email application. Watching a regular user of Blitzmail, you could have predicted the rise of LDAP, IMAP, and most importantly Instant Messenger—Blitzmail was so fast and so ubiquitous that people used it for IM-style back-and-forth conversations long before IM became popular in the larger environment.

The new, wireless, Dartmouth is adopting voice-based instant messaging. Just as ubiquitous wired connectivity lead from email (sporadic and asynchronous) to IM (always-on and synchronous), so ubiquitous wireless connectivity takes us from cellphones to a push-to-talk model. A number of the staff and students there are trialing a very interesting device from Vocera, called a “communications badge.” It’s a small, two ounce device that’s basically just a microphone, speaker, battery, and 802.11 chip. People “push to talk” and use a voice-recognition enabled server to connect to other people. It’s simple, cheap, fast, and significantly lowers the time and effort cost of contacting someone. Just as with IM, the older folks are having trouble dealing with the interruptions. Eventually, you can see this device ending up in a hearing-aid form factor.

Portable devices dominate this campus:
90% of the students on campus have laptops, and 98% of the incoming class of 2007 purchased wi-fi equipped laptops. Wi-fi PDAs are everywhere. One person at the conference actually complained about the poor connectivity on his Blackberry and wished for wi-fi instead! One generation from now, you will hear the phrase, “Daddy, what’s a desktop computer?”

Voice is just another application:
Dartmouth is issuing Voice-over-IP handsets to students for $50 each, and voice is just a low-bandwidth and not very interesting application. Long distance calls are indistinguishable from short distance calls. The phone companies will suffer mightily.

Location based services are emerging:
Students there are already running calendar applications that alert them of their next appointment based on their current location and estimated travel time. People can walk up to a printer and hit “print,” with the computer automatically routing the job to the physically closest printer. At UCSD, students with PDAs can see each other walking around campus, projected on real time maps and send localized scheduling requests to nearby friends who might be available for a quick lunch.

Newsprint has zero value:
The USA Todays and New York Times’ lie uncollected on the doorsteps of hotel rooms on campus. With laptop in hand and ubiquitous wireless, one has Google News, Weblogs, EBay, CraigsList, and more, at their fingertips.

Better surfing requires better chemistry:
With iPaqs and laptops so critical, low battery life is a real problem. The tipping point will come when one of these devices can last an entire day or two of intense use, and recharge in just a few hours. This is the single biggest limiting factor to ubiquitous wireless devices.

This could also be good news for 802.11a—due to its higher data rate, for bursty applications, 802.11a chips can queue up more data, burst that data in a shorter period of time than 802.11b or 802.11g, and therefore spend more of their time “sleeping.” With 500 access points, termination to wirelines does not have to be very common. Most data can be sent for many hops (5-7 or more) before it has to terminate on a landline network. This means that future deployments, with some software assisted self-configuration, should be as simple as plugging a matchbox-sized AP into a wall socket in every room.

Individuals are early adopters:
Conventional wisdom is that most of the money for startups is in serving enterprises, not consumers. However, on campus, residential users use 4-5 times the bandwidth on the wireless network that the administrative staff does. Similarly, wi-fi adoption to date has been dominantly in the home, which bodes poorly for most of the recent crop of wi-fi startups, who are mostly aimed at the enterprise.

Prepare to be Googled:
While sitting at the conference listening to a lecture, I had the surreal experience of watching the person sitting in front of me pull up my bio on the August Capital web site. It used to be that we’d Google companies and people before we met them. With wireless PDAs, it won’t be long until people will Google-scan each other in realtime, as they meet them.

Mobile doesn’t mean distant:
In one study of the usage patterns of Dartmouth students, the typical student only accessed 9 out of 500 access points in 5 buildings each semester. This is similar to people owning cars but using them mostly to drive to neighborhood stores.

Bandwidth matters:
People often wonder what the point of a 70-100 Mbps wireless connection is if the WAN connection is limited to 1.5Mbps. At Dartmouth, people have found ways to fill up the available bandwidth—after normal HTTP traffic, most bandwidth is consumed by backup (people running Dantz Retrospect to backup their laptops over wireless LANs), business school students emailing around bloated Powerpoint slides, and of course, file sharing (which may increasingly head towards private LAN networks given the RIAA lawsuits). Even though every one of the APs at Dartmouth is terminated onto a landline, they are running out of bandwidth, and need to implement load balancing and increased AP density.

802.11a gets the A grade:
Any reasonable deployment has to support 802.11b for legacy purposes. Unfortunately, that eats into the same scarce 2.4GHz spectrum used by 802.11g. Therefore, 802.11a is the only solution for high bandwidth and low-latency services.

Saturation is a deployment tactic:
Microsoft operates one of the densest 802.11 networks in the world, with 4,223 access points accessed by 70,000 users. By positioning APs every 20 meters, they saturate coverage and use cheap hardware instead of expensive and complex planning and deployment software. Startups trying to make money from bandwidth management should remember that in the wired world, increased bandwidth is usually cheaper than complex QoS for most applications.The wireless revolution is possibly over-hyped, but don’t tell that to the people on wireless-enabled campuses today. They have gained wireless ubiquity, and are completely re-thinking how they use cellphones, PDAs, computers, newspapers, instant messenger, printers, power outlets, and most importantly, their time.