💾 Archived View for zaibatsu.circumlunar.space › ~shufei › phlog › 20200915-Reply-X9-WalkawayNet.gmi captured on 2023-11-14 at 08:05:37. Gemini links have been rewritten to link to archived content
⬅️ Previous capture (2023-03-20)
-=-=-=-=-=-=-
Sorry to say I’m quite behind on my smolnet presence, especially responses deserving of consideration. I plead constitutional exhaustion from peasant grubbing at the money mines. Let it not be said my interest has flagged; far from it. But I’m plum knackered much of the time right now.
I recently on Fedi stressed the organic fluiditics of radio, and how the needful attention to the strange potentials in radio are neglected in radio network design these days.
In reply, X9 was kind enough to put together for me a snapshot vision, a manifesto of imperatives and pasteboard of references, for her vision of a WalkawayNet. I daresay for me some of this material is an alluring primer (for dilettantes like moi at least) on new alternative network developments. This is heartily encouraged reading!
the walkaway net: a high-level proposal
Much of this will be redolent of what we are here creating on what I fancy calling Smolnet (Gemini, gopherspace, &c.) Key amongst the themes are also humane priorities which will be familiar to anyone who has done Permaculture or sustainability in its broadest articulations:
long term thinking and deployment;
text primacy;
human legibility;
human comprehension;
privacy;
perhaps anonymity;
network resilience against attack, capture, or fault;
closeness to bare metal;
and attendance to the good social impact of all these facets...
Many people are working on projects which seem to be inching toward an infotech apotheosis of these concerns. FLOSS alone has not quite been enough, somehow, to face the challenges of our day. And yet how deliciously ironic that much of the response we seem to require has been a clarifying, simplifying imperative. I’m hopeful.
There’s a lot to X9’s outline. I’ve not done much reading on NDN. The augmentation of NDN with Tor anonymizing elements in ANDaNA looks delicious, I daresay. If NDN catches on, such enhancements would be a gimme, and hopefully popular.
The papers on gossip networks were more parseable to me, and exciting. Some of this is reminiscent of ideas which bubbled up as I began to play with APRS years ago, and later with phone traffic (manual routing) and HSMM-Mesh / BBHN / AREDN. Namely, that routing on radio needed to be much improved to leverage the viscissitudes of propagation on different bands. As it stands, ham, radio hacker, and professional alike tend to brute force networks on top of radio. And with push systems like AIS / APRS, there is precious little network repair even when there are elements of a gossip propagation model.
In short, Alice may use viral routing to talk to Bob over many APRS hops, but with many such modes, has no way to know if Bob reliably got her message through several cataracts. I had the amateur’s hunch that routing of traffic could be greatly improved with a dose of error correction and destination receipt reports added to acks, yet not sacrifice the “bare metal” concision and radio play which we do on packet radio.
The material X9 provides on Secure ScuttleButt (here referred to as SecSB rather than SSB, for reasons radio geeks will grok) is particularly interesting in this regard. The algorithm for fault tolerance provides a non-naive solution to such problems.
3.5. Fault Tolerance and Tree Repair When a failure occurs, at least one tree branch is affected. Therefore, eager push is not enough to ensure message delivered in face of failures. The lazy push messages exchanged through the remaining nodes of the gossip overlay are used both to recover missing messages but also to provide a quick mechanisms to heal the multicast tree.
Epidemic Broadcast Trees, pp 8-9
A walkaway net which is close to the “bare metal” of radio would almost certainly be a mesh which includes protocols attendant to different bands’ peculiar strengths. To some extent this is done on systems like APRS by doing “smart” stuff on the -IS side, but doesn’t yet include such attendance. That is, a packet should be happy to be carried by 300baud HF hops as well as a 5ghz mesh, but that doesn’t mean it need be agnostic. And if the principles of a Walkaway Net are to be serious, especially regarding resilience and long term human context, they cannot be.
HF, for instance, famously has global reach suitable for anonymizing traffic using broadcast obscurity. But whether Digital Radio Mondiale or Pactor, no larger network outside of .mil to my knowledge maximizes this. Walkaway Net could recycle frequency hopping from ALE and such to “sail” packets like boats on ionospheric clouds as they drift, with very little overhead. If gossip network tree repair were deployed, it would need much less handshaking than ALE and be more resilient as mesh than anything like Pactor, Winlink &c.
On the other hand, our theoretical packet sailboat could do its last mile along a regional VHF/UHF mesh trunk, and join its fellows that way. Selecting which packet goes to which band could easily be part of a gossip algorithm, I wager. Selection could be part of the routing as a variable for acceptable latency.
In the ultimate retrotech evolution of a Walkaway Network, ultimately humans need to be better integrated *as nodes* to such a network. And if it is “smol”, then this network will be optimally human legible to facilitate this, and facilitate resiliency which human networks provide.
To wit, oldschool radio traffic handling provides a good starting basis for human routing. As currently maintained by the National Traffic System and MARS across the USA and its dependencies, the routing is woefully stodgy and brittle. A test of the NTS a few years back showed a ~30% “packet” drop rate, if I recall correctly. This is being improved of late. But the routing is the primo issue, I believe, not the dedication of human nodes, as current events have enticed renewed interest. There is no FEC, no tree repair, usually no acks nor receipts beyond the single hop.
But there is no reason we can’t have a better and more elegant human routing system. And should. So this counter proposal for “cyborg traffic”: integrate human routing with an open standard, a smol standard which fits in a human mind easily. The standard routing should link in by machine easily, with minimal input. So it must be readable by human and machine. It must be text. The ASCII alphanumerics would suffice in both cases, as X9 reminds us is good and at hand to recycle. But the content need not be immediately human text, and might be anything from pictures to links. (There’s a method of encoding b/w raster avatar icons we used to use on Usenet, which would be fab.) A lot of this is already done in NTS using Red Cross standard compression codes, not to mention telegraph codes.
Here no doubt several incredulous eyebrows are well raised, so let’s give an example:
Alice in Erehwon needs to send Bob in Acirema a note about private funds. Her Walkaway front end encrypts this. It bumbles along normally as packets until it hits the cyberborders of Acirema. Uh oh, Acirema has decided to block all encrypted traffic today. And they are shutting down Tor gateways left and right. So it goes over HF radio, and perhaps a few VHF links. These are eventually jammed, but no matter because Alice’s message has robust FEC. Some bits of the message are sent via a human node named Charlie in their radioshack, who sends them by Morse code to Debbie listening in Acirema. Charlie only sees about 100 ASCII characters and sends them as Morse. Some of the characters are Walkaway Net standard compressions like “WN498”. Debbie acks, patches them into SecSB over the local mesh to Bob, whose own “app” sends a receipt back through Debbie or some other link.
I can imagine numerous such examples where human routing could be beneficial to network resiliency, so do beg those devs of perspicacity to consider such a layer as an enhancement to the core protocol standards. If longtermism / social sustainability is tenable, surely such traditions or skill toolkits are in order as part of a Walkaway Net nodal bottleneck.
How can human nodes contribute to such a network? Where is it appropriate? What sort of NTS-like corps or tradition is necessary? Which radio modes are potentially most effective for machine and human readability? (Morse is definitely hard for machines.). Could Debbie be a machine and Charlie a human or vice versa? How to best do that? Can Debbie and Charlie easily authenticate themselves to each other as via some human accessible and quick disposable key to help foil MitM or general grief?
X9 asks excellent tech questions as part of a challenge to create such a Walkaway Net. Certainly, quantum resistant encryption is directly in order. I’ll mull this, and more to the point, heartily encourage devs to mull this challenge. I hope my gabble encourages you all, at least.
We do need a full stack Walkaway Net in working order. A suped up junkyard hacker’s Fidonet for the new millenium. Yesterday. We all know the many whys. My intution strongly says we weirdos on these virtual islands of misfit toys are onto something vital. The blocks are there, or close enough. SecSB, smolnet / textnet, mesh, DIY power options, radio neohamming, FLOSS, FLOSH... A humane internet as a cyborganic community tradition. Time to pull it all together.
-EOF-
.