W3C home > Mailing lists > Public > public-blockchain@w3.org > September 2016

Re: Open Timestamps

From: Peter Todd <pete@petertodd.org>
Date: Thu, 29 Sep 2016 20:35:51 -0400
To: "S. Matthew English" <s.matthew.english@gmail.com>
Cc: Mountie Lee <mountie@paygate.net>, Wayne Vaughan <wayne@tierion.com>, Melvin Carvalho <melvincarvalho@gmail.com>, Blockchain CG <public-blockchain@w3.org>
Message-ID: <20160930003551.GA10692@fedora-21-dvm>
On Mon, Sep 26, 2016 at 02:39:30PM +0200, S. Matthew English wrote:
> here's an article I wrote recently about timestamping in bitcoin:
> https://cointelegraph.com/news/timestamp-hacking-debunking-the-myth-of-precision-timestamps

I think your article both misunderstands and overstates the problem.

> You might have heard that one of the properties a Blockchain possesses is the
> ability to “prove certain data exists at a certain moment of time” or that it
> somehow “provides proof that some data existed at a specific time”. The
> problem with these claims is that they are demonstrably false.

Suppose I claim that as of Jan 1st, 2016, 12:01am - a specific time t - the
following string existed:

    "The Times 03/Jan/2009 Chancellor on brink of second bailout for banks"

As my evidence, I provide Bitcoin block 0, and the hundreds of thousands of
blocks that follow it.

Is that claim false? I mean, maybe all the Bitcoin miners had their clocks
wrong for seven consecutive years. Or maybe you're looking at a chain that
someone made with tens of millions of dollars worth of electricity. But I can
make those kinds of claims of *any* timestamping solution, so that's not
terribly interesting.

Sure, you can't naively say that a Bitcoin block header timestamp means that
data existed as of the exact time in the block. But you have exceptionally high
assurance that the header existed as of that date + a month, certainely as good
if not better assurance than you can get by any traditional means like
publishing notices in news papers.

Saying otherwise - particularly saying that Chainpoint's claims are
"demonstratable false" is fear-mongering; a more useful and accurate article
would accurately explain the limitations of the tech and why it's non-trivial
to prove things to within more than a day or so. Chainpoint may in fact be
guilty of that false precision - I know my OpenTimestamps Client is, and I
mention it in the "Known Issues" section of its README(1), with a link to
further discussion on the topic - but let's not fall into the trap of making
our debunking's themselves hype.

1) It's a tricky problem to fix, because just rounding off to the nearest day
risks people getting confused by the interactions of timezone conversions
across the international date line.

https://petertodd.org 'peter'[:-1]@petertodd.org

Received on Friday, 30 September 2016 00:36:29 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:08:26 UTC