There's nothing wrong with this Robert X. Cringely column that a little expertise and knowledge couldn't solve: A little background on why I'm spending so many words to tear this essay apart. It's not that I'm opposed to the premise or some thought experiments. It's that it's so horribly uninformed.
Cringely is a well-known fellow in the industry who has written and produced videos about the business side of the Internet. Cringely is a nom de plume, but it's how the fellow appears to want to be known. He writes a regularly column on PBS's site and whenever one appears, it's quoted and linked to all over the Net. So it's my duty, unfortunately, to explain how bad this one is.
(This is the same Cringely who stated two years ago that he was using a passive repeater to obtain Wi-Fi service to his hill-top home in a manner that community wireless folks said was impossible. He promised to provide details and never did, although a personal tragedy was part of that.)
In the first part of the article, he completely botches explaining the hotspot industry in comprehensible terms. He calls several different categories of business aggregators, and completely ignores a whole other set of trends. It's not important that one adopt the cant of an industry, but when well-defined and well-used terms exist, you could use them or invent equally good distinctions if you must. He does neither.
A hotspot operator or wireless ISP installs hotspot infrastructure and manages the network, offering a revenue split and other incentives to real-estate venues to allow the hotspot to be set up; or a real-estate venue may contract with a WISP to have hotspot service installed. Wayport is the categorical WISP at this point, a practically pure play in this space. They contract with venues, install hardware, handle billing, and work with end users. T-Mobile HotSpot is also a WISP in its current model.
Infrastructure builders contract with real-estate venues to install hotspot service that they resell to hotspot aggregators. Cometa and Concourse Communications are infrastructure builders. They typically create vendor-neutral installations in which many operators and aggregators can pay for access for their downstream customers.
Aggregators resell access to other networks, typically handling the billing and authentication negotiation. Boingo Wireless and iPass are aggregators. Most of the cell companies are also aggregators, reselling access to a number of WISP networks, but sometimes also to their own network. SBC and Sprint PCS are building their own locations, but also include Wayport in their network plans.
Cringely writes the big money lately seems to be going into WiFi hotspots and hotspot aggregation. Not so. The Cometa investment, valued at $150 million (not hundreds of millions as Cringely writes), is partly services-in-trade from the three partners, Intel, AT&T, and IBM for installation and bandwidth. That amount was set over a year ago; it's not new investment at this point. Wayport has raised a lot of money in the past, but is only trying (still?) for $10-$15M more for expansion. MobileStar went down with over $80M in venture. The cell companies are self-funding their own network build-out. Boingo has a reasonable amount of investment raised since 2001. iPass raised tens of millions in an IPO a few months ago, but they stated very clearly that 99 percent of their revenue comes from dial-up, and the remaining one percent is broadband wired and wireless -- and most of their broadband is hotel in-room wired. I don't see the cash. Wireless LAN switches are where the real venture has flowed lately.
The first problem that all these aggregators have is the existence of the others. Actually, it's not. They're competing for venues, but all of the operators and infrastructure builders I've spoken with -- all of them, just about -- are looking for exclusive infrastructure and then to resell those locations to as many aggregators as possible. There are enough venues that they're not bumping elbows as I wrote in an article in The New York Times a few months ago. In some ways, more aggregators means it's more likely that more venues would install service because they would be more convinced that demand was there from customers of these services or their upstream aggregators.
For WiFi to really succeed it would be nice if there was just a single aggregator, but that's not the way things work in a market economy. There's no reason for this. Many aggregators can offer different combinations of plans for many sets of networks. iPass sells a metered rate en masse to corporations which essentially average their usage across all of their roaming users; Boingo sells an individual unlimited subscription that requires that each user maximize the utility of that subscription each month. A single "aggregator" in his terminology would also mean that every venue was presented with a single set of terms: like it or lump it.
IPass, another aggregatgor that works slightly differently from Boingo, (iPass is effectively an aggregator of aggregators) claims 2800 hotspots. Ignoring the typo, iPass is just another aggregator; Bob has misdefined the term and this trips him up. iPass actually has 10,000 under contract, even though they only claim 3,000 (not 2,800) today. Given his point about ubiquity, it would be worth mentioning the higher figure, which will be available near term, and include significantly the first reselling of T-Mobile's nearly 4,000 locations in the US.
Cringely now gradually segues into his big idea: let's have a million hotspots. How? By having everyone install hardware provided for free and then hook it up into an aggregated whole that requires Wi-Fi-like -- key word there, like -- hardware.
He relates the story of his friend who uses product placement on TV shows to sell jewelry. He says that his friend bypassed all the lawyers involved in contracting with each show by providing jewelry at no cost to the show's costume designers. It's a cool model because it means there's no money in this space for TV producers, or they'd be sucking it dry. His friend is milking a good run out of an otherwise unexploited niche.
But applying this to the hotspot world is tricky: his million hotspots are in a million venues, owned by probably 250,000 separate companies. Let's say 500,000 venues are owned by 1000 companies and 500,000 by 240,000 companies. Cringely writes, First we need to encourage what are essentially noncommercial hotspots and we do that not by revenue sharing but by providing free equipment. Anyone who wants to start a hotspot gets a free WiFi access point and a free WiFi client card for a notebook or other computer.
Who is this we? Who gives this hardware away? He never says. Who accepts it? There's no equivalent of a costume designer in a random venue. Every one of the 15,000 hotels in the US has a general manager and most have higher-ups. None of those people could accept free hardware -- and it wouldn't be a single piece. A hotel can cost tens to hundreds of thousands to fully wire or unwire. Even a coffee shop might need two access points. Every individual coffee shop in the US has already been approached by many, many hotspot operators -- at least when I talk to the coffee shop owners -- and some have already offered them free hardware, and they said no.
It gets worse. If you want to be part of the WhyFi network, you have to accept WhyFi equipment. Cringely doesn't define quite how WhyFi and Wi-Fi differ except by using a few buzzwords. He wants authentication through a database -- but who runs that infrastructure? Each location? A central authority? Who handles customer service, account management, backups, emergencies, tech support? That's a massive infrastructure right there if you scale above a few dozen people -- even a few hundred requires tons of resources.
The features of the access point and gateway in his vision include bandwidth throttling, firewalling, and other aspects of control that are the heart and soul of the Sputnik firmware, AP, and Central Control. Bob, you don't have to invent it. It already exists and it uses Wi-Fi standards and other networking standards. And it has all kinds of open-source and extensible components to it.
your WhyFi card gives you free unlimited access to the entire network through MAC address filtering. Still not sure how you get the millions of Wi-Fi cards replaced with WhyFi cards, and then how MAC address filtering -- a completely ineffective method of controlling access to a network as the MAC addresses must be sent in the clear and thus can be kiped and cloned -- really addresses anything. Then there's the issue of real-time processing of MAC addresses on a local WLAN authenticated through a centralized database: what happens when you lose or replace your WhyFi card? How do you register in the first place?
Right now many readers are thinking that most ISPs frown on hotspots and connection sharing. That's true but they also function in a competitive environment such that I don't think any major ISP could make stick such a prohibition if there was widespread cheating. Some strategic lawsuits a la the RIAA prosecution of teenagers stealing music would be part of it. Cable companies believe they can sue under theft of cable service laws. Some ISPs have been willing to cancel accounts when they find sharing going on. If it becomes widespread, there may just be more cancellations and prosecutions. Or Speakeasy Networks will get a lot more business for their legitimate and encouraged network sharing. (He mentioned Speakeasy.)
Because there is no revenue sharing the software to manage the WhyFi network can be much simpler. There's also no money to fund the new hardware or software or firmware or infrastructure necessary to run this system. And Cringely talks about charging just a moment later. The software to manage millions of MAC addresses and accounts even for free requires huge investment.
Because the hardware and service are free to hotspot owners there is likely to be great demand leading to those one million hotspots. I'm in the Twilight Zone. WHO is providing this for free?
And even the free subscribers don't present a burden on the system because each of their WhyFi cards extends the hotspot they are connected to by building a little ad hoc WhyFi access point of its own....Note that what I propose is simple technology. No mesh networks here. Not that I am opposed to mesh networks but I doubt that even WhyFi would reach the kind of density needed to make that concept work.
So is this entire essay a speculation on what could happen if every current Wi-Fi user were given (or purchased) new free equipment, a new standard were developed, and WhyFi was a bridge and access point and client at the same time? It's possible, of course, but when you start bridging and handling local traffic and have lots of individual access points in a small space, there are huge coordination issues under Wi-Fi, as well as a necessity of using the same channel, massively reducing available bandwidth. It's possible to approach this better with mesh, but it means even more so that entirely new hardware would be needed...which I guess he's proposing in the first place.
When it is finished the WhyFi network would have one million hotspots and hotspot extender cards, 30 million paying subscribers. Back up: who pays? I thought you needed a WhyFi card to access the network and then you used it for free. Without a WhyFi card, um, how do you access the network?
And how do you manage a network with 30 million subscribers on an ad hoc basis with no infrastructure and no centralization?
Here's my final flaw in Cringely's essay: the elephant in the room he's not mentioning. Free wireless. It's all over the place. Community groups. Municipalities. Businesses. Groups of businesses. Free wireless is a huge inchoate "movement" in which thousands of locations offer it without any coordination among most of them. Pyramid Research estimates that in a few short years most hotels that offer broadband (Wi-Fi and wired) won't charge separately for it -- it'll be an amenity like a bed or a desk.
The million points of WhyFi lead me to ask Why WhyFi? Why not Wi-Fi? The current set of overlapping commercial motivations to offer for-fee service and the huge wave of free and open commercial, public, civil, and community access mean that the nooks and crannies will be filled in.
Perhaps I'm spending too much time deconstructing Cringely, but his reach extends far. I'm not sure why we need a new standard nor who will pay for any part of it. This might have been just speculative fiction on his part, but there are too many errors and too many missing pieces to take any of it seriously.