Jeffrey K. Belk, senior vice president of marketing at Qualcomm, makes many excellent points in his open memo (republished on Alan Reiter's blog) on using wireless data on the road. I want to separate out and rebut some of his contentions, however, as I believe that while his fundamentals are almost always correct, he has set up some strawmen, and has misstated the state of the Wi-Fi hot spot industry.
Throughout his comments, he fails to separate the financial model that might underlie Wi-Fi hot spot deployment and its actual technical limitations. This is partly due to the ubiquitous, slow-everywhere mindset that he's coming from.
The crux of Belk's contentions begin with this statement:
Because we all know that although we have 56kpbs modems, we don't get 56kbps, we get 30-45kbps. So, if I were to travel on a one week trip in the US, say going from San Diego to San Francisco. Then from San Francisco to Chicago to New York and home, I could access use my Wireless Wide Area PC card and be productive---anywhere I have wireless coverage, anywhere I can make a phone call. As much as I need to...at a fixed rate of $80 per month.
This $80 per month is a rate for 1x CDMA2000 through SprintPCS and Verizon Wireless through Qualcomm's corporate rate. This statement seems more or less correct and fundamental: anywhere, he can get something approximating a modem call with no hidden charges. That's his starting point.
London is Falling
I checked into the One Aldwych hotel in London... Unfortunately, he appears to have chosen one of the very few locations in the world that requires its own software and a weird setup to activate. Only locations that have chosen to only work with Boingo Wireless's system have any comparable bar to entry, and Boingo's software is superior to Windows in managing Wi-Fi configurations -- and it doesn't have to be uninstalled later, as Belk notes is critical with the software One Aldwych is using. The problems with One Aldwych have been documented extensively elsewhere, including in columns by travel writers Joe Brantcatelli (JoeSentMe.com) and Joe Sharkey (New York Times's Business section).
The service is also ridiculously expensive, running eight pounds for two hours in the trial and 16 pounds for 24 hours after the trial was over. By comparison, most US-based Wi-Fi hotel service is $7 to $10 per day with a few exceptions; even in most countries, nearly $30 for 24 hours access is egregious.
Rebuttal: strawman. This is an unintentional strawman because Berk chose this hotel specifically because of its vaunted Wi-Fi service. The service is both difficult to configure and expensive, and not part of any network. Most US hotels use a simple gateway page, work with Boingo's client, or use a room-based authentication system (for wired). Most are also not expensive, and most are part of networks. This will also happen worldwide as international hotel chains with significant US presence roll out more consistent pricing or loyalty programs worldwide.
Stockholm Success, Helsinki Pricing
Belk has no real complaints about Telia HomeRun, although its pricing is, again, a little high: roughly $15.26 for 24 hours access. Regular Swedish travelers can get a monthly rate, of course, and Telia is associated with other Scandinavian networks for roaming.
Again, not a bad service, the user interface was well implemented, except for the location directory... Unfortunately for Belk, most Swedes are carrying phones that allow them to get hot spot locations through a pretty simple interface, which might explain why the Web interface isn't as good as those for countries that have to rely on it to direct users.
In Helsinki, he was going to be charged $58 (50 euros at the time) for 24 hours access on Sonera's network, which I believe everyone everywhere in the world would agree is insanely high.
Rebuttal: fair. Pricing is inconsistent, and probably too high. Part of this reflects an emergent market without the competition that's starting to appear in the US.
Los Angeles for Less
Beck returns to Los Angeles now and checks into a hotel. On the desk, a sign for "High Speed Internet Access". Now, this was wired Ethernet. Cost $9.95 for a 24 hour period. ...I did not use this, as being back in California I was able to fire up the trusty Thinkpad with my CDMA2000 card and get my mail...
Rebuttal: great example of tradeoff, but ignores monthly subscription. This is a perfect case of using the all-inclusive (but slower) cell data system instead of an expensive Wi-Fi solution for that kind of spot traffic. I use my 9600 bps GSM connection all the time to avoid $10 per day charges when I just need a little connectivity. However, Belk isn't at a point where the Wi-Fi networks are broad enough that he would have had a flat monthly subscription (at probably $50 per month) that would have allowed him a "free" connection to the hotel at probably T-1 (1.5 Mbps) speeds.
Some Speciousness and Real Strawmen
Belk now tries to analyze apples and oranges, and here's where I really kick in. He totals his potential cost in Europe at $127 for five hot spots, even though he sent back the $58 card and used a cheaper service.
He then says that a similar trip in the US would have cost only $13.30 as a prorated part of his 3G service.
This is specious in two directions: first, a similar trip wholly in the US would have cost him probably no more than $50 and possibly as little as $25 for Wi-Fi service in the same locations. Second, he can't prorate his $80/month service: he's paying for the full month. If he only travels five days, the service costs him $80 per month. If he uses it every day, then, of course, the utility is divided by that.
Belk now sets up what he expects the "Wi-Fi" rebuttal to be, although I'm not sure who he means by Wi-Fi: Hot spot operators (wISPs)? Aggregators? Community networkers? Manufacturers? It's hard to say "Wi-Fi" and mean only one of those, but he almost certainly is referring to individual networks, like Wayport.
Backhaul. Belk notes on a Hotspot service, ALL Wi-Fi connections speeds are limited by the backhaul (i.e. the way the Access Point is connected to the Internet). Likewise, however, backhaul can be easily increased. If a location has 512 Kbps fractional T-1, they can, for a fee, generally easily upgrade to full 1.5 Mbps T-1. Cell data is highly limited by spectrum availability, cell locations, number of simultaneous users, and other factors. He can get tens of Kbps right now, and generally will be able to, but cell data will be highly susceptible to non-point-to-point backhaul/carrying capacity factors.
Because he's vaunting the 50-80 Kbps speed of his 3G subscription, anything that's wireless-to-backhaul has to offer him a significant improvement, in the hundreds of Kbps to over a Mbps to anchor him to a specific location. That's perfectly reasonable.
But he starts to fall down when he says that hundreds of Kbps is what he gets from hot spots not 11 Mbps. I agree that hot spots may advertise 11 Mbps networks, and that's overstating the case. But 512 Kbps is not 50-80 Kbps. It's 10 times faster. For many people, working on 512 Kbps is like working in an office, while 50-80 Kbps is sucking at a straw. It changes your behavior, which is why broadband users don't act like dial-up users.
Locations. Here's where Belk really starts to confuse ubiquity with utility. My axiom on this is: Ubiquity's advantage fades in direct proportion to its speed relative to nearby faster service at a comparable price. If I can get 80 Kbps for $80 per month everywhere, but find myself within short walks (or on top of) 1 Mbps for $50 per month in certain places, the right answer might be a subscription to both services. They're not exclusive. Most modern business travelers need bandwidth. Another axiom: The need for high-speed access is directly in proportion with the speed of a company's business relationships. Thus speed can equal money, and a lot more speed can be worth much more than its marginal cost.
Belk sets up and knocks down all the locations he isn't interested in using data services, which form the majority of the hot spots in the country: Starbucks (because he gets it delivered to his house), Borders, McDonalds. Let's tell their tens of millions of customers each month that nobody goes there; a sort of reverse Yogi Berra statement.
He then knocks down the 300 foot diameter claims for Wi-Fi. Of course, that's line of sight service with built-in antennas. He's really knocking down the claims of certain providers, not claims of the industry as a whole or any reasonable person in an IT department deciding on how to deploy it.
Belk takes the potential area served by all the hot spots expected to appear and comes up with a laughable number. But it's laughable only if you expect ubiquity. Ubiquity, speed, and cost are three dependent variables: turn one up or down without adjusting the other, and your model thrives or fails. Keep speed low, cost medium, and ubiquity at 11 (yes, Spinal Tap joke), and you have 3G service today. Turn ubiquity down and speed to moderate to high and you have Wi-Fi; the cost number might be at all kinds of settings today, but moderate to low tomorrow.
Here's the final strawman in this section: However, folks go to a pub to socialize and drink pints, they go to a bookstore to browse and buy books, and they go to McDonald's to eat. Belk is conflating current behavior with future behavior. I still doubt that businesspeople will use McDonald's for Wi-Fi, but then I'm told that millions of business travelers, like salespeople, use McDonald's as their temporary offices every day. Belk is saying that because he can't see the use today, it won't happen tomorrow. But then again, where does he expect 3G users are using their data service? It's in McDonald's. And everywhere.
Hotel ubiquity. Belk hasn't kept up with announcements. By June, many US hotels will have added Wi-Fi where none existed. Hotel-based services like Lodgenet will add Wi-Fi access points to their in-room entertainment centers. Conference rooms, public areas, and other locations in hotels will get Wi-Fi.
It's not necessarily inevitable, nor is it cheap. But as a hotel consultant told me, hotels spend a few million dollars on carpeting every few years. A few million on a long-term Wi-Fi investment will happen if the competition among hotels requires that Wi-Fi be installed to keep visitors staying loyal.
He also notes, the Hotspot environment, in the U.S. and Europe will be fragmented, with lots of different providers. This means lots of different log-on methods, lots of different usage experience, lot and lots of different pricing policies, and lots and lots of Big Bills! In fact, the reverse is rapidly becoming true. Roaming convergence with single cell/Wi-Fi bill (T-Mobile) or single login/aggregated service (Boingo, iPass, GRIC) is becoming a matter of course, not the exception.
3G and Unlimited Bills
Beck now actually makes a number of arguments in favor of offloading data to Wi-Fi. He points out that GPRS networks are already congested for voice, and that adding data will cause even more problems. (This is marketing as well as fact: their technology is competitive with GPRS.)
So Wi-Fi and GPRS makes sense as a package, because cell carriers want to offload their expensive data traffic onto Wi-Fi networks which are relatively cheaper and allow them to continue to carry huge amounts of voice traffic on the expensive cell spectrum. (See my InfoWorld story which quotes Sky Dayton of Boingo saying the same thing.)
Then he says that UMTS providers will experiment quite a bit and offer all kind of plans and options because they'll have the spectrum available for it. We'll see. I think he set up one strawman here -- scarce voice spectrum -- that can't be so easily knocked down.
The finances just don't make sense. If it costs a small fraction of your cost of licensed spectrum and cell towers to build or buy into a national network that offers substantially higher speeds at partner locations, why not preserve that licensed spectrum for higher-cost services?
Where the Wi-Fi Roams
And WHERE THE TRAFFIC IS GENERATED will not be people traveling to a Pub, McDonalds or Starbucks to connect.
I'm afraid he gets it wrong here. It's a combination of people who already use the spaces that will have Wi-Fi and people who will migrate to spaces that offer it. It's not a mass migration.
Schlotzsky's Deli put in free Wi-Fi in a small number of Texas stores. They said they now have six percent of their customers to those stores coming because of the free Wi-Fi. And then buying drinks and sandwiches which they estimate will add as much as $100,000 in revenue per store per year.
Likewise, I don't know any early adopter who doesn't check out the locations of local Wi-Fi from Starbucks to community networks to sticking their laptop out the window before traveling. High-speed connectivity at broadband speeds not fast modem speeds are now a necessity, not an option, for many travelers. Current 3G can't fulfill that; Wi-Fi is the only option that does.
Belk contends that Wi-Fi will continue to be fragmented, available in limited places, and expensive. All trends point towards unified roaming at a fee that will probably be no more than $50 to $75 per month for unlimited usage, and maybe less with some exceptions.
Given a unified T-Mobile network that offers $30/month (one-year commitment) unlimited service, and Boingo (incorporating Wayport, several airports, etc.) for $50 per month, you could make a case that we already have an $80/month plan that covers virtually all major population centers in the country. And it's only getting denser.
The real closer in this gap is airports. When we hit the top 35 airports in the country having significant Wi-Fi presence, which I still predict as 25 of them out of 35 by the end of this year, then you'll see business travelers make a massive uptake.
Would I choose a few Mbps in an airport for $8 (or $50 per month) versus 10s of Kbps in an airport as part of my prepaid flatrate fee? You can bet Mbps wins -- when it's something I expect and it's available in significant locations.
Belk isn't looking at the time people spend in one location: hotel, airport, other venues. He's thinking about the time in transit. 3G spans that transit time as a modem replacement; Wi-Fi fills in those long static times for broadband.
(Side note: And does anyone remember MobileStar? MobileStar was the Wi-Fi Hotspot service provider that went BANKRUPT and was purchased by T-Mobile as the basis for their Starbuck's Wi-Fi network. MobileStar's business model had all kinds of problems, and T-Mobile bought MobileStar's assets, not the business. They're a strawman of all strawmen because they spent what I estimate as $90 million without having an adequate tracking model for costs or revenue.)
Belk closes with a good analysis of the upcoming potential for 3G data and with a couple of practical experiments for you to think about the differences in 3G and Wi-Fi.
I don't believe a compelling model for profit has yet emerged from the hot spot world. Revenue remains low. Customer price sensitivity is still being tested. The crossover from massive home and business deployment of Wi-Fi networks hasn't resulted in a nation of roaming mavens.
The success of free networks, both commercial (like Schlotzsky's) and community (library, groups, etc.) has shown that people will use Wi-Fi when it's available without a hassle or a cost attached.
The real question in Wi-Fi's future is not whether it's a complementary technology to 3G even when deployed in as few as 10,000 hot spots. The real question remains whether the hot spot operators will find enough users fast enough in the next year, and overcome industry nonsense to develop effective flat-rate roaming agreements while rolling out additional airport service.
For-fee hot spot service should have a future, but exactly what that will look like is unknown. But point-based high-speed service represents something that people want and use. Broadband has rapidly replaced dial-up service in people's home because of the convenience of having that big pipe and always on connection. Likewise, Wi-Fi niche in a cellular world will continue to be a narrow but deep one that becomes an increasingly important hand-in-hand partner to developing cell data service.