Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Why aren’t distributed systems engineers working on blockchain technology?

eng.paxos.com

Post reply on HN
Print view View original

11–20 of 204 posts

Re: Why aren’t distributed systems engineers working on blockchain technology?

#11

Post by jchanimal »

Distributed consensus algorithms are super interesting, but most distributed databases run in a trusted context. So instead of spending time optimizing a trustless system, database engineers optimize commit protocols like Calvin[1], because high-throughput scalable consistency solves so many business problems.

When you have a general purpose globally consistent distributed database, many of the problems that look like blockchain algorithm problems turn out to be standard application programming tasks. Eg a distributed ledger is just a table in a distributed database.

[1] https://fauna.com/blog/distributed-consistency-at-scale-span...

Re: Why aren’t distributed systems engineers working on blockchain technology?

#12

Post by lotyrin »

I don't really get this article. The transaction rate limit and the wastefulness and limitations of PoW as a fault-tolerance mechanism are brought up but then just hand-waved away.

If we had discussed proof of stake or theorized about how some of these other hurdles like transaction rate could be leaped then sure. Otherwise it's just a list of things any distributed system has to solve for, demonstration that blockchain solves for them (inefficiently) with the undertone of "I bought in on the hype, and you can too".

Re: Why aren’t distributed systems engineers working on blockchain technology?

#13

Post by random3 »

The goal of the article is to get distributed systems engineers excited about blockchain and invite them to explore positions at Paxos. Ironically none of the open positions are actually for distributed systems engineers :)

I think it's funny the author chose to center the article around that assumption. IMO most engineers related to distributed computing (not necessarily themselves distributed systems engineers) are actually quite aware of the technology and many of the challenges.

Re: Why aren’t distributed systems engineers working on blockchain technology?

#14

Post by JackFr »

tostitos1979 wrote: post #8
random023987 wrote: post #4

Proof-of-work blockchains were developed to solve a problem that only exists in a small percentage of distributed engineering use-cases. Unless you need to build a "censorship-free" peer-to-peer ledger distributed among untrusted, anonymous participants, the complexity costs and latency and computation costs simply aren't worth it. It's the same reason that Ford doesn't build a tunnel under the US/Mexico border to fe…

Very good points. In complex financial transactions (at least within regulatory domains), my understanding is that rollbacks can and do happen. I initially thought across borders/regulatory domains, blockchain will shine. I haven't seen a killer application just yet. It is a bit frustrating because the tech is indeed interesting .. we just don't have multi-billion dollar real (as opposed to imagined) use-cases beyond…

https://www.oii.ox.ac.uk/blog/the-blockchain-paradox-why-dis...

"Let me explain why. In economic organization, we must distinguish between enforcing rules and making rules. Laws are rules enforced by state bureaucracy and made by a legislature. The SWIFT Protocol is a set of rules enforced by SWIFTNet (a centralized computational system) and made, ultimately, by SWIFT’s Board of Directors. The Bitcoin Protocol is a set of rules enforced by the Bitcoin Network (a distributed network of computers) made by — whom exactly? Who makes the rules matters at least as much as who enforces them. Blockchain technology may provide for completely impartial rule-enforcement, but that is of little comfort if the rules themselves are changed. This rule-making is what we refer to as governance."

Re: Why aren’t distributed systems engineers working on blockchain technology?

#15

Post by vasilipupkin »

tostitos1979 wrote: post #8
random023987 wrote: post #4

Proof-of-work blockchains were developed to solve a problem that only exists in a small percentage of distributed engineering use-cases. Unless you need to build a "censorship-free" peer-to-peer ledger distributed among untrusted, anonymous participants, the complexity costs and latency and computation costs simply aren't worth it. It's the same reason that Ford doesn't build a tunnel under the US/Mexico border to fe…

Very good points. In complex financial transactions (at least within regulatory domains), my understanding is that rollbacks can and do happen. I initially thought across borders/regulatory domains, blockchain will shine. I haven't seen a killer application just yet. It is a bit frustrating because the tech is indeed interesting .. we just don't have multi-billion dollar real (as opposed to imagined) use-cases beyond…

In my opinion, the ICO is the blockchain killer app. You don't really need a distributed ledger except in case of digital currencies, which can be used to facilitate transactions in specific apps

Re: Why aren’t distributed systems engineers working on blockchain technology?

#17

Post by DINKDINK »

andrewstuart2 wrote: post #9
random023987 wrote: post #4

Proof-of-work blockchains were developed to solve a problem that only exists in a small percentage of distributed engineering use-cases. Unless you need to build a "censorship-free" peer-to-peer ledger distributed among untrusted, anonymous participants, the complexity costs and latency and computation costs simply aren't worth it. It's the same reason that Ford doesn't build a tunnel under the US/Mexico border to fe…

Proof-of-work is simply not sustainable (energy-wise), IMO, for the foundation of a global system. The stability of whole ecosystems is predicated upon those ecosystems being attractive to people who want to make money by competing to solve hard problems by brute force. That's a lot of energy going into a system that can commit a new block of transactions once every 10 minutes on average. It solves a problem in an in…

>Proof-of-work is simply not sustainable (energy-wise), IMO, for the foundation of a global system.

What have you found that is cheaper to proof-of-work that works on a global system?[1]

[1] "Nothing is Cheaper than Proof of Work" http://www.truthcoin.info/blog/pow-cheapest/

Re: Why aren’t distributed systems engineers working on blockchain technology?

#18

Post by davidgerard »

Because it turns out the good bits are the ledger of transactions made tamper-evident in a Merkle tree, and we had that with Git in 2005 and it works great thanks.

The bit Git doesn't do: distributed consensus via Proof of Work, because it turns out that a bit of trust is fabulously more efficient.

I predict any good and useful product labeled "Blockchain" will be the cryptographically tamper-evident ledger of transactions, and not any of the stuff that makes a blockchain different to Git.

(Researching the book, I had one developer admit that his "blockchain product" was pretty much a simplified Git in terms of what it offered ;-) )

Re: Why aren’t distributed systems engineers working on blockchain technology?

#19

Post by random023987 »

vasilipupkin wrote: post #15
tostitos1979 wrote: post #8

Earlier quoted context omitted.

Very good points. In complex financial transactions (at least within regulatory domains), my understanding is that rollbacks can and do happen. I initially thought across borders/regulatory domains, blockchain will shine. I haven't seen a killer application just yet. It is a bit frustrating because the tech is indeed interesting .. we just don't have multi-billion dollar real (as opposed to imagined) use-cases beyond…

In my opinion, the ICO is the blockchain killer app. You don't really need a distributed ledger except in case of digital currencies, which can be used to facilitate transactions in specific apps

> In my opinion, the ICO is the blockchain killer app.

There are so many examples, from Canadian Tire Money https://en.wikipedia.org/wiki/Canadian_Tire_money to Microsoft Points https://en.wikipedia.org/wiki/Microsoft_Points showing you can create private currencies granting the bearer certain privileges without a blockchain.

Re: Why aren’t distributed systems engineers working on blockchain technology?

#20

Post by jimktrains2 »

andrewstuart2 wrote: post #9
random023987 wrote: post #4

Proof-of-work blockchains were developed to solve a problem that only exists in a small percentage of distributed engineering use-cases. Unless you need to build a "censorship-free" peer-to-peer ledger distributed among untrusted, anonymous participants, the complexity costs and latency and computation costs simply aren't worth it. It's the same reason that Ford doesn't build a tunnel under the US/Mexico border to fe…

Proof-of-work is simply not sustainable (energy-wise), IMO, for the foundation of a global system. The stability of whole ecosystems is predicated upon those ecosystems being attractive to people who want to make money by competing to solve hard problems by brute force. That's a lot of energy going into a system that can commit a new block of transactions once every 10 minutes on average. It solves a problem in an in…

> Proof-of-work is simply not sustainable (energy-wise), IMO, for the foundation of a global system.

Visa, MasterCard, AmEX, &c along with each individual bank, not to mention the intermediaries and gateways all of them use also consume a tremendous amount of power. It's not as if our current system uses a negligible amount of power, not to mention the number of steps and entities a transaction needs in order to be finalized.

Post reply on HN