These two firms have independently prototyped methods of increasing downstream speeds for home users by banding them together: The idea is that consumer broadband is dramatically underused on average even as it is dramatically oversold in general. The ratios mentioned in the article, for DSL range from 14 to 200 oversubscription, meaning that as much as 200 Mbps of DSL service has been installed for every one Mbps of actual Internet access. (It's another reason why I've been surprised that despite companies like Akamai and products like Real Networks' reflectors--still extant?--that inside-the-edge servers at ISPs haven't been a much bigger industry. Perhaps it is and it's very quiet.)
Both Mushroom Networks and WiBoost's potential products would increase downstream bandwidth by borrowing from equally equipped neighbors' connections. Let's say A, B, and C had 1.5 Mbps/384 Kbps DSL. When A is at work and B is reading email, C could be browsing high-bandwidth Web sites and having them load, potentially, three times as fast less some overhead in managing connections among the special devices that would sit just to the customer's side of the DSL modem, much like where a VoIP adapter should be located topologically.
One rub is that without the agreement of an Internet service provider to install special software that would handle dynamic bonding of these separate DSL connections, these aggregated gateways could only split traffic quite grossly. For instance, they might be able to send image requests for a Web page to several neighbors' DSL lines, but not divvy up an FTP transfer.
The other rub is that ISPs often prohibit behavior of this type under the rubric of sharing a connection.
Now, some ISPs might see this as an opportunity to promote a "we all hang together" scenario, as New York Times reporter John Markoff cleverly quips in the article's opening paragraph, and offer neighborhood's superior rates for all signing up for longer commitments. Everyone participating would receive, on average, much higher speeds, and the ISP could offer a reduced group rate paid partly through a reduction in marketing expense and churn.
WiBoost gave me an early demonstration a little while ago, and the technology is de facto, not proposed. The next step for both companies is to find manufacturers and/or ISPs ready to take the plunge into a new form of network bonding.
I thought this was one of the bigger potential benefits of Mesh Wifi networking. Imagine a street with 5 Broadband connections in 5 houses that each ran a wifi AP. If we could aggregate the broadband, they'd have 20Mbps between the houses and then burst bandwidth of 5 times a single broadband line. And/Or redundancy between 5 different ISPs. But it's tricky and may not be possible done purely on the subscriber end.
I'm afraid that this is not quite as simple as it sounds... Apart from the technology issues concerned, security and privacy would be Major issues. It would also create a situation where everyones broadband connection would become inherently unstable, depending on the use somebody else... Also purely technicaly speaking you cannot just "bundle" bandwith by connecting several broadband connections into on virtual network. It sounds nice, but it just isn't feasible I am afraid.
[Editor's Note: I'm not sure I agree it's infeasible, but I do think that the points you raise require an ISP's involvement. If the devices secured packets from the ISPs side to the individual customer's gateway regardless of which path they took to get to the gateway (using a VPN tunnel), then there's no opportunity for neighbors to see traffic. Stability can be addressed by dynamic throttling, so that a given user always has priority for the maximum bandwidth of their individual connection. The bundling is quite simple if you have control of both ends of the connection. ISPs and others have bonded together multiple connections for years--it's done with T-1s all the time--but it can't be done unilaterally on the users's side.--gf]