From: Matt Whitlock <bip@mattwhitlock.name>
To: bitcoin-development@lists.sourceforge.net
Date: Fri, 12 Jun 2015 14:22:36 -0400
Message-ID: <23144512.HX7dfatEFr@crushinator>
User-Agent: KMail/4.14.9 (Linux/3.18.12-gentoo; KDE/4.14.9; x86_64; ; )
In-Reply-To: <20150612181153.GB19199@muck>
References: <20150612181153.GB19199@muck>
MIME-Version: 1.0
Content-Transfer-Encoding: 7Bit
Content-Type: text/plain; charset="us-ascii"
Subject: Re: [Bitcoin-development] User vote in blocksize through fees
Precedence: list

Why should miners only be able to vote for "double the limit" or "halve" the limit? If you're going to use bits, I think you need to use two bits:

	0 0 = no preference ("wildcard" vote)
	0 1 = vote for the limit to remain the same
	1 0 = vote for the limit to be halved
	1 1 = vote for the limit to be doubled

User transactions would follow the same usage. In particular, a user vote of "0 0" (no preference) could be included in a block casting any vote, but a block voting "0 0" (no preference) could only contain transactions voting "0 0" as well.

Incidentally, I love this idea, as it addresses a concern I immediately had with Jeff's proposal, which is that it hands control exclusively to the miners. And your proposal here fixes that shortcoming in a economically powerful way: miners lose out on fees if they don't represent the wishes of the users.


On Friday, 12 June 2015, at 2:11 pm, Peter Todd wrote:
> Jeff Garzik recently proposed that the upper blocksize limit be removed
> entirely, with a "soft" limit being enforced via miner vote, recorded by
> hashing power.
> 
> This mechanism within the protocol for users to have any influence over
> the miner vote. We can add that back by providing a way for transactions
> themselves to set a flag determining whether or not they can be included
> in a block casting a specific vote.
> 
> We can simplify Garzik's vote to say that one of the nVersion bits
> either votes for the blocksize to be increased, or decreased, by some
> fixed ratio (e.g 2x or 1/2x) the next interval. Then we can use a
> nVersion bit in transactions themselves, also voting for an increase or
> decrease. Transactions may only be included in blocks with an
> indentical vote, thus providing miners with a monetary incentive via
> fees to vote according to user wishes.
> 
> Of course, to cast a "don't care" vote we can either define an
> additional bit, or sign the transaction with both versions. Equally we
> can even have different versions with different fees, broadcast via a
> mechanism such as replace-by-fee.
> 
> 
> See also John Dillon's proposal for proof-of-stake blocksize voting:
> 
> https://www.mail-archive.com/bitcoin-development@lists.sourceforge.net/msg02323.html
> 
> -- 
> 'peter'[:-1]@petertodd.org
> 0000000000000000127ab1d576dc851f374424f1269c4700ccaba2c42d97e778