Bob Cringely writes a follow-up column about his WhyFi idea, this time spelling out the impractical details more impractically: Cringely comes clean with the details of his WhyFi idea to spread free Wi-Fi hotspots nationwide. I ripped apart his previous column because it was long on bad ideas, short on execution strategies.

He expects that every participant in the project who offers free Wi-Fi will eat the bandwidth bill in exchange for free equipment, which will be loaned not given to them. Only those providing hotspots get free access to the network. (Original business models of Joltage [dead], SOHOWireless [apparently dead], and Sputnik [now an enterprise software developer].)

The free hotspots will apparently be part of a nationwide authentication network that will only allow members of this club to get in for free. Otherwise, users are charged for use. Cringely estimates the cost of a million hotspots at $150 million. He suggests someone underwrite this project to make a pile of money.

So now I can tell you exactly why this idea doesn't work, especially now that he's dropped the whole part from his first column about requiring special firmware or MAC filtering.

Hotspots cost more than $150 each. As I noted in my response to his first column, Cringely has magically eliminated the overhead costs for running a national network with a database of legitimate users. There's no dollars in here for running the backend, shipping out products, helping with installation (even by phone), dealing with customer/technical support ("my account doesn't work," "the hotspot is dead"). I would estimate given his plan that the cost per location for a million locations is about $300 per location for a single access point (which many won't be; see below), and about $20 to $50 per month for all of the associated support. More likely, the support costs are about $10 per month per free user on the network. It could cost more to support the paid users, and Cringely doesn't postulate a payment.

Hotspots aren't a single access point and you can't put them just anywhere. If you exclude homes and coffeeshops and a few small retail establishments, locations that have value and lots of traffic control their spectrum and require expensive or at least complicated, multi-AP installations. A mall or an airport can prevent tenants or airlines from installing APs. This is an ongoing battle right now in airports.

Arbitrary density doesn't promote the right kind of density. If every small business in the country was a hotspot, there would still be fast dead areas. If I have 20 businesses in a row across two blocks with WhyFi nodes, but a mall a few blocks away is dead as are all the hotels in the vicinity, that doesn't fulfill the WhyFi dream.

Going from zero to a million. The value of this network is only apparent when it's well underway. Unlike the Internet, in which adding a resource makes it useful to the entire global network at once, squaring the value of the network with each additional node (Metcalfe's Law), WhyFi is additive: each node only barely improves the overall utility of the network's free access.

Venue owners need to be cultivated. You can't just throw 1,000,000 APs in the mail and have a network spring up. The cost of cultivating each location or chain involves money and time.

Free doesn't feed the business bulldog. A hotel doesn't need free Wi-Fi when it roams. It's a company. So they're not going to install free Wi-Fi in a WhyFi network manner and allow random free usage. Rather, they might offer free Wi-Fi on their own terms to their own guests. Wyndham Hotels and Resorts, for instance, offers free broadband and Wi-Fi to members of its free affinity club. The Marriott chain is building out free Wi-Fi in 1,700 budget-value properties (i.e., places families stay or budget-minded business travelers).

That old bandwidth problem again. Bob believes that a million people and businesses can share their broadband connections, the majority of which have usage agreements prohibiting sharing. So no company is going to pony up $150M or $1B or whatever the actual cost would be to encourage people to violate their usage contracts.

What's in it for them? Bob suggests that the benefit for the million WhyFi nodes is that they get free Wi-Fi. How many people actually have the simultaneous interest to offer free Wi-Fi by subsidizing it off their broadband connection and have PDAs or laptops where they think they'll get a benefit from free roaming on other WhyFi locations.

We don't need Wi-Fi everywhere. We just need it in the right place. Whether it's free, municipal, subsidized, or commercial, Wi-Fi hotspots have to be in the right places where people congregate, not in arbitrary locations self-selected by nodes.

Free commercially funded Wi-Fi is an idea that's spreading, but it's going to spread in the vernacular: not with a centralized database and huge funding. Each business or group of businesses will make their decisions and roll it out, and eventually it's going to be trivial for someone to find free, commercially supported access (not to mention free community and free municipal) in any business district.

There will always be limitations to free service: speed, time of day, quality of service, uptime, and bandwidth. These factors coupled with what will probably wind up being an unlimited flat rate of $20/month for nationwide roaming for Wi-Fi hotspots will make the alternatives easy and clear.