W3C home > Mailing lists > Public > public-interledger@w3.org > October 2015

Re: Private 'Distributed Ledgers' Miss the Point of a Blockchain

From: Jae Kwon <jae@tendermint.com>
Date: Wed, 28 Oct 2015 18:46:09 -0700
Message-ID: <CAMXKmU5MgSB5d+xrAJBSG74DU5YGhNeznUxg4gCXsCQO0F5FNA@mail.gmail.com>
To: Melvin Carvalho <melvincarvalho@gmail.com>
Cc: Interledger Community Group <public-interledger@w3.org>
>
> The consensus algorithm "Paxos" was one such implementation created in
> 1989, followed shortly thereafter by "Raft.” It’s hard to believe that such
> innovations would take 25 years to be discovered in the hyper-efficient
> world of database design.


Those are fault-tolerant systems, but not Byzantine fault-tolerant
systems.  Byzantine fault-tolerant algorithms have been around since the
80s, and PBFT from 1999 was the first such practical system.  The problem
with PBFT is that it's difficult to understand and implement.

Bitcoin's core value proposition is that it's a superbly simple Byzantine
fault-tolerant ledger where security (mining power) from untrusted parties
can be aggregated together.

The only decentralized ledger that can supplant the need for Bitcoin's
mining process and provide the immutability of Bitcoin are those based on
classical Byzantine fault-tolerant algorithms like PBFT.  Hyperledger
almost implemented one.  Tendermint is the most complete implementation
today, though it isn't exactly PBFT. (See
tendermint.com/posts/tendermint-vs-pbft/).  These systems can be faster,
more secure, more efficient, and more scalable -- better by so many metrics
-- than Bitcoin mining blockchains.  The only problem is that the
technology isn't understood well enough by the blockchain world, and
there's a lot of wild false claims that others who attempt to solve the
consensus problems by other (usually flawed) means.  There's a reason why
Byzantine fault-tolerant consensus is cited as the most confusing field of
computer science.

There's no reason why such a non-PoW BFT consensus blockchain can't be
permissionless either.  Tendermint's open source reference implementation
was a crack at that.  It's a shame that it never launched as a
cryptocurrency ;)

On Wed, Oct 28, 2015 at 5:11 PM, Melvin Carvalho <melvincarvalho@gmail.com>
wrote:

> Bitcoin may have had the most successful marketing campaign of any web
> technology in recent years. But now there’s a new buzzword making waves
> throughout the financial industry: “distributed ledger.”
>
> As with many buzzwords, distributed ledger technology has come to mean
> many things. Some say it's a tool to enable transparency by ensuring that
> all members of a group receive cryptographically secured messages about
> participants’ activities. Others suggest that these ledgers will notarize
> communications. Some are even bold enough to predict that distributed
> ledgers will end the madness of managing multiple database and
> reconciliation structures.
>
> While the industry is still working out the details of just what
> distributed ledgers actually are, the major options seem to have a few
> things in common.
>
> Distributed ledgers have primarily claimed to supplant the need for
> Bitcoin's mining <https://en.bitcoin.it/wiki/Mining> process by
> introducing trust requirements among participants. These ledgers also
> promise users the immutability of Bitcoin without the need for expensive
> mining operations. Unfortunately, most of these claims demonstrate a
> significant lack of understanding about the efficiencies of a blockchain
> <http://www.americanbanker.com/bankthink/get-ready-for-the-rise-of-the-blockchain-1073843-1.html>
> .
>
> One immediate red flag that should jump out at any fan of market
> efficiency is that the technology powering distributed ledgers predates
> blockchains by well over 20 years.
>
> The consensus algorithm "Paxos" was one such implementation created in
> 1989, followed shortly thereafter by "Raft.” It’s hard to believe that such
> innovations would take 25 years to be discovered in the hyper-efficient
> world of database design.
>
> Moreover, the inefficiency of interbank settlement services has little to
> do with technology. The primary reason inter-bank settlement takes days to
> clear comes down to regulations.
>
> These regulations exist for a number of reasons. Most of them are designed
> to reduce risk via lengthy, often manual processes. These delays allow
> banks to reverse transactions if needed.
>
> Proponents of distributed ledgers argue that they can displace centralized
> providers such as SWIFT, ACH and CHIPS by moving money faster. But
> distributed ledgers are only more efficient insofar as they are able to
> circumvent the overhead of regulatory requirements.
>
> There’s no doubt that blockchain technology will facilitate disruptive
> innovations in finance, much as the Internet facilitated easy public access
> to information. But a world of private ledgers sounds eerily similar to a
> range of “private Internets.”
>
> In addition, many of the features offered by distributed ledgers are
> already available on other systems.
>
> Encrypted e-mail, for example, is a system in which cryptographic receipts
> are time-stamped and logged by all participating parties. The SWIFT network
> itself facilitates the exchange of identity-verified and auditable
> messages. Participating institutions have thus appeared largely satisfied
> with the reliability of SWIFT’s notarization and auditability. And no
> institutions have been clamoring to share logs of their activity on this
> network with direct competitors.
>
> Blockchain technology is useful not because it offers efficiency in a
> world of message-passing but because it uses a complex process to settle
> value between untrusted parties. But distributed ledgers do not offer users
> the ability to easily convert their tokens and messages into fungible units
> of value. Nor do distributed ledgers escrow value between parties that
> don't trust each other.
>
> If a ledger is not a public resource, it will have the pressures incumbent
> to existing settlement systems plus the overhead of maintaining a shared
> database among competitors. What efficiency will remain thereafter remains
> dubious.
>
> If distributed ledgers do offer the ability for banks to improve
> efficiency in their processes, it will likely be because they've afforded
> banks permission to innovate—not because they’re able to provide settlement
> to underserved notarization clients.
>
>
> http://www.americanbanker.com/bankthink/private-distributed-ledgers-miss-the-point-of-a-blockchain-1077435-1.html
>
> Interesting (tho perhaps controversial) article, maybe relevant to
> discussions here.
>
Received on Thursday, 29 October 2015 15:05:00 UTC

This archive was generated by hypermail 2.3.1 : Thursday, 29 October 2015 15:05:01 UTC