Herbert West, Reanimator.
I can remember a long time ago, in the late 80s, about a system that was pre-Internet and based on dialup links and not related to FidoNET—it was UUCP, and it brought e-mail and news to any system that wished to participate and could find a link provider. Why is it important to hypernets? Because the reality of future hypernets isn’t going to be the slightly delayed speed-of-light handoff of packets between nodes in datacenters from anywhere on the planet. The reality will be data is couriered by people along their daily routines from place to place as they travel to and from work, school, and other social situations.
Never underestimate the bandwidth of a station wagon full of tapes.
— Dennis Ritchie (apocryphal)
Where some people are building Pirate Boxes and Wifi Extenders, the truth about bandwidth and radio is that it is a severely limited resource. When chunks of any kind of radio spectrum are deregulated by the State, devices multiply and saturate the available radio waves until they become unbearably slow. In a dense suburb, a cul-de-sac full of McMansions can create an immensely frustrating experience of impossible connections, slow web pages, and random connection resets where baby monitors, security cameras and wireless phones compete in the Battle Royale of Bandwidth.
UUCP was then the network of availability—modems and phone lines were available, so we used them. For a while, we adopted more synchronous protocols and migrated away from asynchronous batching of data transfers to packetized sessions over TCP/IP transported via SLIP or PPP. In short, we made ourselves addicted to novelty at every step of the evolution of the Internet. We liked letters and we made them faster with e-mail. We made “social networks” of e-mail when we made USENET news where the only downvote was your local killfile. We liked informative documents so we made Gopher and later made them the surface of communication throughout the planet with the World Wide Web and web browsers. We made so many documents and files to share that we needed to create directories like Archie to find them, which of course, was still again, Not Fast Enough, and we created full-text indexes of the Internet and search engines like Alta Vista, Inktomi, and Google. We even made meta-search-engines to aggregate the results of our favorite search engines. We made instant messaging and SMS and we made ourselves panic or upset when there was no responsiveness from the other party. We strapped on the yokes to collaborate with each other with Outlook and Exchange and SharePoint while we simultaneously distract ourselves with fire-hoses of novelty flooding us from websites like Reddit, Digg, StumbleUpon, Ffffound, Fark, Slashdot, Something Awful (I can probably find 50 more…), and of course Twitter, always at hand on your computer or your mobile phone. Our movies and entertainment no longer require a physical trip to the theater, they come to use streaming in Full HD at 3-5 MB/s, that is, if we’re not waiting for them to buffer. This is the Battle Royale of Attention.
The intersection of the Battle Royales of Bandwidth and Attention is where thoughtful and inspired content, communication will come. If you have ever been fascinated by websites like Letters of Note, you’ll know that in most cases, the letters are old (relatively of course). They were written before the Internet between people who were being awesome instead of being entertained.
So, what does this have to do with UUCP? UUCP was the support for a model of asynchronous-bandwidth that certain applications (USENET, e-mail) worked well on. Today, with modern computing, we have ways of representing audio and video that could barely be imagined in the early 80′s. We can bring it back, and make it work—again, for people who wish to be awesome instead of merely entertained. Where do we go with this? More than likely, signed-content-addressable networks (SCAN) with anonymous multicast publish-subscribe (AMPS). These of course are names I’ve mashed together to embody the qualities of the application that seem to be required in an asynchronous hypernet—that is, a hypernet that is built from batched input and outputs transported by hand, mail or car. In the end, the names of the technologies don’t matter, what matters is that we create a mechanism to connect hypernet nodes together for publishers and subscribers in such a way that content is authoritative (no imposters), and automatic (less frustration) without the convenience of a “live” connection to the network.
So, what should the asynchronous hypernet build on? We’re not going back to dial-up modems like UUCP relied-upon. The amount of data we want to share with each other is too much for modem-transfer. How about something cheap and small? SD memory cards you get for digital cameras have many excellent qualities: small, cheap, and ubiquitous. A 4GB SD memory card is less than $5 and small enough to mail with first-class postage in the United States for $0.65 from Portland, ME to Portland, OR—a 3,200 mile trek that would take you over 50-hours to drive. How much bandwidth for a full card would that be equivalent to? About ~23KB/s. Change to a 32GB SD card and you increase bandwidth to ~186KB/s. That is, of course, if you’re willing to wait and you’re not measuring how long it takes to read/write the card. Standardize on a 4GB card and every $5 adds another 4GB of potential-bandwidth to the hypernet. $100 of 4GB cards is 80GB of potential-bandwidth. A round-trip of data-transfer then costs $0.65 × 2 = $1.30 to move 8GB of information for each card. Rings can be created forwarding a card in a loop around the world returning to the original sender, a new kind of “token” ring—just add more postage or more hops if hand-delivered person-to-person. This becomes the physical network medium, for now we have to solve the protocol of what we should expect to see on the medium. This is the next step—but after we decide the first types of data and applications that it should support, for UUCP it was e-mail and news. Today, we’re probably looking at syndicated graphics (ebooks, emagazines, etc.), audio, and video as the hypernet’s key interests.
The stress-free future is lazy and asynchronous—if we let it be.
This remember me IP4 over Avian Carriers RFC
I’ve been having similar thoughts for a few years, now. I wouldn’t say that people handing around SD cards is going to be the future of electronic communication for everyone, but it’s an important capability to have. The phrase ‘signed-content-addressable networks’ identifies some techniques that would be helpful in making such a thing work. The Freenet developers occasionally mention that because of the way files are address in Freenet, USB storage drives could be used as a transport. I half-started a project that would use hash-based URNs and digital signatures as a blogging platform but that it hasn’t yet gone beyond the brainstorming phase. If you care to read my ideas, they’re half-assedly documented at http://www.dognds.net/
People don’t seem to realize the potential uses of low-capacity memory cards. If you can buy a 4GB card for $5, you should be able to buy 64MB cards for a dime a dozen. 64MB is still quite a bit of space, and the cards would be so cheap you could just stick them wherever, hand them out and not care if you got them back, hide them for people to find… all sorts of interesting possibilities when you can store a considerable amount of information on a tiny plastic card with a cost of about a penny.
Unfortunately, nobody sells bags of 64MB cards for a dime a dozen, because nobody makes them that small anymore. 1GB and under are hard to find now. It’s a shame the potential isn’t realized. (Or maybe I’m overestimating the scale of manufacturing costs, and a 64MB card would still cost as much to make as a 4GB?)
That’s a kind of beauty of the network, if we make a standard “packet” whatever $5 buys, the network just gets more powerful. This year, 4GB is what is available for $5, and when they stop making them by next year, 8GB will become the next “default size.” Sure, if you want to spend more money now for a larger packet, you can do that, but you don’t have to. 64MB cards would be useful for smaller amounts of data—like slowe-mail, but at this point, you probably have to go mining for them in land fills to get them.