Update to Double-Spend Deprecation Window Proposal

This update strengthens the incentive not to confirm double-spends after time T (30 seconds). To grow and stabilize adoption, it is necessary to influence the miner of the block after a deprecated block, who in turn is concerned with the block after that. Accordingly, the disincentive is changed from a simple delay to a temporary chain work penalty, which can be negative. Hal Finney first suggested this in 2011.

The penalty is graduated in two steps based on the respend gap, for reasons explained within. I believe it is the minimum required to achieve the intended result.

Double-Spend Deprecation Window https://github.com/dgenr8/out-there/blob/master/ds-dep-win.md

This is an incredibly dangerous and foolish proposal that opens up the Bitcoin network to serious vulnerabilities, both from attackers outside the network, as well as miners trying to gain an advantage over their competition.

Ultimately it's flawed for the same root problem that proof-of-stake proposals suffer from: the p2p network just isn't a reliable broadcast medium. Seeing a transaction is not a guarantee that any other node has seen it; not seeing a transaction is not a guarantee other nodes have not seen a spend.

You can measure "propagation times" and other metrics all you want, but you're measuring a network that isn't under attack; Bitcoin must be robust against attacks, and it must not create incentives to launch them. Institutionalising the punishment of miners being they did not have perfect connectivity - an unattainable goal in a trust less, decentralised system - is athema to the goals of having a decentralised systmem and will only lead to smaller mining operations being punished for being the victim of attacks on their network connectivity that are only made profitable by this proposal.

Equally your proposal actually makes it *easier* to pull off apparently single-confirm double-spend attacks - any miner who ignores a block containing the apparent double-spend is just as likely to be aiding an attacker trying to get a 1-conf transaction double-spent. This forces *everyone* to waiting *longer* before accepting a transaction because now even a single-confirmation is no longer good evidence of an accepted transaction. In an ecosystem where hardly anyone relies on zeroconf anyway your putting a much larger group of people at risk who weren't at risk before.

Frankly if this idea gets traction it should serve as a warning to all miners that it's time they adopt replace-by-fee to set a clear precedent that they have no obligations other than the same economic self-interest- not vague notions of "honesty" - that makes Bitcoin work in the first place.

BTW you quote Hal Finney and Satoshi in your proposal to try to lend support to it. Don't do that - appealing to authority is a surefire way to get people to ignore you. Its particularly bad when the authorities being appealed too haven't participated in consensus research for years; you're referencing stuff from a time when Bitcoin was barely understood.

On 8 February 2015 21:38:55 GMT-06:00, Tom Harding <tomh@thinlink•com> wrote:

>
>This update strengthens the incentive not to confirm double-spends
>after
>time T (30 seconds).  To grow and stabilize adoption, it is necessary
>to
>influence the miner of the block after a deprecated block, who in turn
>is concerned with the block after that. Accordingly, the disincentive
>is
>changed from a simple delay to a temporary chain work penalty, which
>can
>be negative.  Hal Finney first suggested this in 2011.
>
>The penalty is graduated in two steps based on the respend gap, for
>reasons explained within.  I believe it is the minimum required to
>achieve the intended result.
>
>Double-Spend Deprecation Window
>https://github.com/dgenr8/out-there/blob/master/ds-dep-win.md
>
>
>------------------------------------------------------------------------------
>Dive into the World of Parallel Programming. The Go Parallel Website,
>sponsored by Intel and developed in partnership with Slashdot Media, is
>your
>hub for all things parallel software development, from weekly thought
>leadership blogs to news, videos, case studies, tutorials and more.
>Take a
>look and join the conversation now. http://goparallel.sourceforge.net/
>_______________________________________________
>Bitcoin-development mailing list
>Bitcoin-development@lists•sourceforge•net
>https://lists.sourceforge.net/lists/listinfo/bitcoin-development

Many thanks for the feedback Peter. Please if you would, see below

On 2/8/2015 10:32 PM, Peter Todd wrote:

> Seeing a transaction is not a guarantee that any other node has seen it; not seeing a transaction is not a guarantee other nodes have not seen a spend.

In no way does proposal rely on such assumptions. It develops local rules which result in a desirable outcome for the network as a whole, under the applicable statistics.

> you're measuring a network that isn't under attack; Bitcoin must be robust against attacks, and it must not create incentives to launch them.

Two specific attacks are addressed at some length. No one is keener than I to learn of new ones, or flaws in those treatments.

> Institutionalising the punishment of miners being they did not have perfect connectivity - an unattainable goal in a trust less, decentralised system - is athema to the goals of having a decentralised systmem and will only lead to smaller mining operations being punished for being the victim of attacks on their network connectivity that are only made profitable by this proposal.

Building from unavoidable imperfections is the necessary spirit when interfacing with physical reality. I would defer to miners whether these specific worries outweigh the benefits of helping to achieve a 30 second network, rather than a 10±10 minute network.

> Equally your proposal actually makes it *easier* to pull off apparently single-confirm double-spend attacks - any miner who ignores a block containing the apparent double-spend is just as likely to be aiding an attacker trying to get a 1-conf transaction double-spent. This forces *everyone* to waiting *longer* before accepting a transaction because now even a single-confirmation is no longer good evidence of an accepted transaction. In an ecosystem where hardly anyone relies on zeroconf anyway your putting a much larger group of people at risk who weren't at risk before.

I agree on one point -- it is necessary to let transactions mature for something on the order of 15 to 30 seconds before mining them, as discussed in proposal. I quite disagree regarding Finney (1-conf) attacks. In fact this proposal is the only one I've seen that actually stops most Finney attacks -- all those where the block comes more than 30 seconds after tx1.