The shortest distance _is_ still a straight line, it's just not a cartesian line.
Hamming distance (which is what the entire article is about and somehow doesn't mention it, or edit distance, or anything along those lines) is still a metric space, and still follows all the standard triangle inequality properties.
If you convert your standard tree into a BK tree, suddenly all the assumptions about tree structures come back.
(I also briefly met Bram Cohen of BitTorrent in SF once, but didn't really keep in touch).
Anyway, Ian Clarke uses a different routing system which he refers to as a "small world topology". BitTorrent uses "mainline DHT" based on Kademlia, by Petar Maymounkov (who I also met 6 years ago, when I started Intercoin.org ... he had been a lecturer at NYU so it was easy).
MaidSafe (now called Autonomi) is slightly different, it uses its own Kademlia DHT, but with at least one major improvement: it removes the IP addresses after one hop. So you can't DDOS the network, for example, or discover all the nodes easily.
I'm big into decentralized systems. I've met pretty much everyone in the space, from Tim Berners-Lee in 2014 when he was doing SoLiD (who left and started Inrupt) to Stefan Thomas and David Schwartz from Ripple. I even emailed back and forth with Leslie Lamport for a while, about the "Buridan's Principle", which helped me a lot to understand how to properly do scalable consensus across large distributed systems.
The original point of banks was a place to handle the security aspects of money. Crypto dumps the burden of security on its users.
That wouldn’t be a big deal if most crypto transactions where tiny and people where keeping trivial sums in their wallets, but that’s not what’s going on. Thus defeating the point of low transaction costs as there’s a small number of transactions going on and a great deal of overhead supporting those transactions resulting in high effective fees.
Do you think the fees will ever be low enough for casual use? Seems to me like like it's just a matter of time--though perhaps much more time than was foretasted by some.
Until then, there are oddball use cases, like Autonomi, which couldn't use a bank for this anyhow.
I've been watching MaidSafe/Autonomi for a while now and I'm excited to see how it goes for them. It's not a blockchain, it's not BitTorrent, but (if I've understood it) the Kademlia DHT is a bit of a fusion between the two.
It'll have its own unique social consequences and when I'm feeling optimistic I dare to hope that it'll be a step on the path to something better than what were doing today.
It's not what I see when I close my eyes and dream of the distributed infra we need. That dream is a partition tolerant one. But it's different enough that it might just solve some of the deeply entrenched problems that plague us.
Autonomi pretends that the network will store your data forever in exchange of just an initial payment. This is a bold claim, and I'm not sure how this can work in practice - the docs didn't give much details.
It does feel a bit unsustainable, but I'm not willing to write it off... blockchains manage a much more extreme version of the same proposition (every node stores your data forever). It'll just come down to whether the price stabilizes to a point that's low enough to be useful.
Intercoin.org isn’t the point of my comment, it isn’t even linked in there. Why would you zero in on it? It is your comment that appears stilted and following the same forced pattern of “crypto was mentioned, therefore … must… assume … grift … is the only reason for saying anything … and therefore must write the most negative adhominem possible to fulfill my duty of making sure decentralized technology has negative sentiment expressed”
I was sharing my experience with people in the decentralized space. After all the original post was by the founder of a very ambitious and helpful decentralized project, talking about how their DHT works. I am impressed by such founders and interviewed them. So I gave a bit more context as to what various projects use and linked to the interviews. All these people contributed to the decentralized routing schemes and consensus algorithms.
> Intercoin.org isn’t the point of my comment, it isn’t even linked in there. Why would you zero in on it?
Because you mentioned it. It would be sleazy even if it was a casual name drop for any other self promo. The crypto part was not what put me off in particular.
Hamming distance (which is what the entire article is about and somehow doesn't mention it, or edit distance, or anything along those lines) is still a metric space, and still follows all the standard triangle inequality properties.
If you convert your standard tree into a BK tree, suddenly all the assumptions about tree structures come back.