After cogitating on it for a couple of days and talking to other reporters, it's pretty clear that Fon has a great smokescreen with its free "Linus" model: The fact is, Fon is just a paid hotspot operator that wants to eschew all venue signing, installation, and management costs. Fon has three kinds of hotspot users, but attention has focused on the Linuses, those who opt to let others who are also Linuses use their hotspots for free. This will always be a subset of all users. Let's say Fon reaches 1m installed hotspots by 2010, one of the stated goals of the founder. If even 750,000 are "Linus" hotspots, only 750,000 people will be able to use those hotspots at no cost. Six billion people--okay, the tens of millions of travelers with Wi-Fi laptops or handhelds or phones--will pay.
Let's not lose sight of the fact that there are now hundreds of aggressive hotspot builders cherrypicking the best remaining locations. If Fon only gets the dregs of locations (suburbs, apartment buildings, coffeeshops without service right now), then they're a niche-filling for-fee network. If they offer roaming across other networks and resell to aggregators, then they become a backfill network for missing locations, but only if people in those locations find it worthwhile to collect small fees in exchange for sharing bandwidth.
Primarily, Fon is for-fee. Primarily, it faces the same location, location, location issues as other for-fee operators. Locations that want to offer free Wi-Fi will not install Fon.
Update: Esme Vos has a quasi-rant, too, on this subject in which she wonders precisely how Fon fits in the current ecosystem. She suggests that sharing a connection (or having an ISP sell you a device that shares your bandwidth) produces uncertain amounts of bandwidth for your own users. (A FAQ at Fon does mention that bandwidth throttling is a future planned option for the software.) And, she asks, isn't municipal scale better than house to house?