Re:“In 2018 the blockchain/decentralization story fell apart”
by EGreg on Tuesday January 01, 2019 @07:04PM
(#18803231)
Full disclosure I started the intercoin.org project. So I am speaking from about 2 years experience working on and thinking hard about these problems, such as doublespending and distributed hash timestamping.
There have been very interesting PARALLELIZABLE distributed byzantine fault tolerant systems being built, like MaidSAFE and Holochain. They are based around Distributed Hash Tables and have no centralized bottlenecks unlike the miners in proof of work. This is much older than cryptocurrency, we are talking 2002 - Kademlia and Merkle Trees, used in eg BitTorrent.
That is the key. Any technology, like a blockchain, where the entire network has to store all the data, scales ridiculously poorly. To coin a phrase, it’s EMBARASSINGLY UNSCALABLE.
You can call the opposite of that sharding, but I will call it PARALLELIZABLE. The good news is future is simply to take EXISTING systems (eg the Web) and INCREMENTALLY make them end to end encrypted and make the backend into a DHT where each activity or token is watched by SOME but not all computers.
The current Web topology is scalable but TERRIBLE for security, as evidenced by the steady stream of hacks and leaks and untrustworthy behavior by platforms. We throw up our hands like after every school shooting, as if there is nothing we can do. We can, and blockchains are not necessary.
More details can be found here, including mathematical results from 2009 on the probability of a successful double-spend attack as a function of how many computers watch each token. The results are perhaps surprising:
https://forum.intercoin.org/t/intercoin-technology-the-ledge...
If you want to discuss architecture of DLTs in detail, you are welcome to post in our forum. (The guy arguing in that particular thread is the chief crytographer of Ripple, he signed up to debate ideas of the architecture.)
Anyway the short story is that any “blockchain” system with global consensus simply can’t scale, not enough to support actual transactions. Has a single ETHEREUM token been used for its intended purpose in daily transactions? (Cryptokitties is the closest and it nearly brought the whole system to a halt.) It’s all been relegated to speculation on upcoming things, and this is what killed the dream. The exchanges are centralized databases. The actual ledger they post to when you cash out can handle 10 transactions per second regardless of the number of computers on the network. The same is NOT true of almost any other distributed protocol (email, web, etc.)
There have been very interesting PARALLELIZABLE distributed byzantine fault tolerant systems being built, like MaidSAFE and Holochain. They are based around Distributed Hash Tables and have no centralized bottlenecks unlike the miners in proof of work. This is much older than cryptocurrency, we are talking 2002 - Kademlia and Merkle Trees, used in eg BitTorrent.
That is the key. Any technology, like a blockchain, where the entire network has to store all the data, scales ridiculously poorly. To coin a phrase, it’s EMBARASSINGLY UNSCALABLE.
You can call the opposite of that sharding, but I will call it PARALLELIZABLE. The good news is future is simply to take EXISTING systems (eg the Web) and INCREMENTALLY make them end to end encrypted and make the backend into a DHT where each activity or token is watched by SOME but not all computers.
The current Web topology is scalable but TERRIBLE for security, as evidenced by the steady stream of hacks and leaks and untrustworthy behavior by platforms. We throw up our hands like after every school shooting, as if there is nothing we can do. We can, and blockchains are not necessary.
More details can be found here, including mathematical results from 2009 on the probability of a successful double-spend attack as a function of how many computers watch each token. The results are perhaps surprising: https://forum.intercoin.org/t/intercoin-technology-the-ledge...
Debates about consensus: https://forum.intercoin.org/t/intercoin-technology-consensus
If you want to discuss architecture of DLTs in detail, you are welcome to post in our forum. (The guy arguing in that particular thread is the chief crytographer of Ripple, he signed up to debate ideas of the architecture.)
Anyway the short story is that any “blockchain” system with global consensus simply can’t scale, not enough to support actual transactions. Has a single ETHEREUM token been used for its intended purpose in daily transactions? (Cryptokitties is the closest and it nearly brought the whole system to a halt.) It’s all been relegated to speculation on upcoming things, and this is what killed the dream. The exchanges are centralized databases. The actual ledger they post to when you cash out can handle 10 transactions per second regardless of the number of computers on the network. The same is NOT true of almost any other distributed protocol (email, web, etc.)