Re: [bitcoin-dev] Miners forced to run non-core code in order to get segwit activated

From: Mark Friedenbach <mark@friedenbach•org>
To: Erik Aronesty <erik@q32•com>
Cc: Bitcoin Dev <bitcoin-dev@lists•linuxfoundation.org>
Subject: Re: [bitcoin-dev] Miners forced to run non-core code in order to get segwit activated
Date: Tue, 20 Jun 2017 19:11:15 -0700	[thread overview]
Message-ID: <CAOG=w-tPgJ5baaNiZC5rTs_y=eV7AU+F=aGaH+uObqaB-VgL2w@mail.gmail.com> (raw)
In-Reply-To: <CAJowKgJcp7d30LsrDZ5iR6-k9Ncz0N90pxs2GmJkuG1qYDG6GQ@mail.gmail.com>

80% have set "NYA" in their coinbase string. We have no idea what that
means. People are equating it to BIP 91 -- but BIP 91 did not exist at
the time of the New York agreement, and differs from the actual text
of the NYA in substantive ways. The "Segwit2MB" that existed at the
time of the NYA, and which was explicitly referenced by the text is
the proposal by Sergio Demian Lerner that was made to this mailing
list on 31 March. The text of the NYA grants no authority for
upgrading this proposal while remaining compliant with the agreement.
This is without even considering the fact that in the days after the
NYA there was disagreement among those who signed it as to what it
meant.

I feel it is a very dangerous and unwarranted assumption people are
making that what we are seeing now is either 80% support for BIP-91 or
for the code in the btc1 repo.

On Tue, Jun 20, 2017 at 6:36 PM, Erik Aronesty <erik@q32•com> wrote:
> # Jacob Eliosoff:
>
>>  will start orphaning non-bit-1 blocks before Aug 1, and we avoid a split.
>
> Correct.  There are 2 short activation periods in BIP91 either of which
> would avoid a split.
>
> # Gregory Maxwell:
>
>> unclear to me _exactly_ what it would need to implement to be consistent.
>
> This is the relevant pull req to core:
>
> https://github.com/bitcoin/bitcoin/pull/10444
>
> Seems OK.  It's technically running now on testnet5.   I think it (or a
> -bip148 option) should be merged as soon as feasible.
>
>> previously debunked "XT" and "Classic" hysteria.
>
> apples vs oranges, imo.   segwit is not a contentious feature.   the
> "bundling" in segwit2x is, but that's not the issue here.   the issue is we
> are indirectly requiring miners that strongly support segwit to install
> consensus protocol changes outside of bitcoin's standard reference.   80% of
> them have signaled they will do so.   these are uncharted waters.
>
>
> On Tue, Jun 20, 2017 at 6:57 PM, Jacob Eliosoff via bitcoin-dev
> <bitcoin-dev@lists•linuxfoundation.org> wrote:
>>
>> I could be wrong, but the latest BIP91 implementation (also included in
>> Segwit2x) cuts the activation period to 336 blocks (2.33 days).  (This has
>> been updated at
>> https://github.com/bitcoin/bips/blob/master/bip-0091.mediawiki.)  So if 80%
>> of hashpower is actually running that code and signaling on bit 4 by July 25
>> or so, then those 80+% will start orphaning non-bit-1 blocks before Aug 1,
>> and we avoid a split.
>>
>> There may still be a few non-bit-1 blocks that get orphaned after Aug 1,
>> because they're mined by old BIP141 nodes.  But it seems like very few
>> miners won't be signaling either Segwit2x *or* BIP141 by then...
>>
>> Make sense?
>>
>>
>> On Tue, Jun 20, 2017 at 6:48 PM, Mark Friedenbach <mark@friedenbach•org>
>> wrote:
>>>
>>> Why do you say activation by August 1st is likely? That would require an
>>> entire difficulty adjustment period with >=95% bit1 signaling. That seems a
>>> tall order to organize in the scant few weeks remaining.
>>>
>>> On Jun 20, 2017, at 3:29 PM, Jacob Eliosoff via bitcoin-dev
>>> <bitcoin-dev@lists•linuxfoundation.org> wrote:
>>>
>>> If segwit is activated before Aug 1, as now seems likely, there will be
>>> no split that day.  But if activation is via Segwit2x (also likely), and at
>>> least some nodes do & some don't follow through with the HF 3mo later
>>> (again, likely), agreed w/ Greg that *then* we'll see a split - probably in
>>> Sep/Oct.  How those two chains will match up and how the split will play out
>>> is anyone's guess...
>>>
>>>
>>>
>>> On Jun 20, 2017 6:16 PM, "Hampus Sjöberg via bitcoin-dev"
>>> <bitcoin-dev@lists•linuxfoundation.org> wrote:
>>>
>>> > Ironically, it looks like most of the segwit2x signaling miners are
>>> > faking it (because they're not signaling segwit which it requires).
>>> > It'll be unfortunate if some aren't faking it and start orphaning
>>> > their own blocks because they are failing to signal segwit.
>>>
>>> Well, they're doing some kind of "pre-signaling" in the coinbase at the
>>> moment, because the segwit2x project is still in alpha-phase according to
>>> the timeline. They're just showing commitment.
>>> I'm sure they will begin signaling on version bit 4/BIP91 as well as
>>> actually running a segwit2x node when the time comes.
>>>
>>>
>>> > As far as prevent a chain split goes, all those things
>>> > (148/91/segwit2x(per today)) effectively guarantee a chainsplit-- so I
>>> > don't think that holds.
>>>
>>> Segwit2x/BIP91/BIP148 will orphan miners that do not run a Segwit2x (or
>>> BIP148) node, because they wouldn't have the new consensus rule of requiring
>>> all blocks to signal for segwit.
>>> I don't believe there would be any long lasting chainsplit though
>>> (because of the ~80% hashrate support on segwit2x), perhaps 2-3 blocks if we
>>> get unlucky.
>>>
>>> Hampus
>>>
>>> 2017-06-20 23:49 GMT+02:00 Gregory Maxwell via bitcoin-dev
>>> <bitcoin-dev@lists•linuxfoundation.org>:
>>>>
>>>> On Tue, Jun 20, 2017 at 3:44 PM, Erik Aronesty via bitcoin-dev
>>>> <bitcoin-dev@lists•linuxfoundation.org> wrote:
>>>> > Because a large percentage of miners are indifferent, right now miners
>>>> > have
>>>> > to choose between BIP148 and Segwit2x if they want to activate Segwit.
>>>>
>>>> Miners can simply continuing signaling segwit, which will leave them
>>>> at least soft-fork compatible with BIP148 and BIP91 (and god knows
>>>> what "segwit2x" is since they keep changing the actual definition and
>>>> do not have a specification; but last I saw the near-term behavior the
>>>> same as BIP91 but with a radically reduced activation window, so the
>>>> story would be the same there in the near term).
>>>>
>>>> Ironically, it looks like most of the segwit2x signaling miners are
>>>> faking it (because they're not signaling segwit which it requires).
>>>> It'll be unfortunate if some aren't faking it and start orphaning
>>>> their own blocks because they are failing to signal segwit.
>>>>
>>>> I don't think the rejection of segwit2x from Bitcoin's developers
>>>> could be any more resolute than what we've already seen:
>>>> https://en.bitcoin.it/wiki/Segwit_support
>>>>
>>>> On Tue, Jun 20, 2017 at 5:22 PM, Mark Friedenbach via bitcoin-dev
>>>> <bitcoin-dev@lists•linuxfoundation.org> wrote:
>>>> > I think it is very naïve to assume that any shift would be temporary.
>>>> > We have a hard enough time getting miners to proactively upgrade to
>>>> > recent versions of the reference bitcoin daemon. If miners interpret
>>>> > the situation as being forced to run non-reference software in order
>>>> > to prevent a chain split because a lack of support from Bitcoin Core,
>>>> > that could be a one-way street.
>>>>
>>>> I think this is somewhat naive and sounds a lot like the repeat of the
>>>> previously debunked "XT" and "Classic" hysteria.
>>>>
>>>> There is a reason that segwit2x is pretty much unanimously rejected by
>>>> the technical community.  And just like with XT/Classic/Unlimited
>>>> you'll continue to see a strong correlation with people who are
>>>> unwilling and unable to keep updating the software at an acceptable
>>>> level of quality-- esp. because the very founding on their fork is
>>>> predicated on discarding those properties.
>>>>
>>>> If miners want to go off and create an altcoin-- welp, thats something
>>>> they can always do,  and nothing about that will force anyone to go
>>>> along with it.
>>>>
>>>> As far as prevent a chain split goes, all those things
>>>> (148/91/segwit2x(per today)) effectively guarantee a chainsplit-- so I
>>>> don't think that holds.
>>>> _______________________________________________
>>>> bitcoin-dev mailing list
>>>> bitcoin-dev@lists•linuxfoundation.org
>>>> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
>>>
>>>
>>>
>>> _______________________________________________
>>> bitcoin-dev mailing list
>>> bitcoin-dev@lists•linuxfoundation.org
>>> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
>>>
>>>
>>> _______________________________________________
>>> bitcoin-dev mailing list
>>> bitcoin-dev@lists•linuxfoundation.org
>>> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
>>
>>
>>
>> _______________________________________________
>> bitcoin-dev mailing list
>> bitcoin-dev@lists•linuxfoundation.org
>> https://lists.linuxfoundation.org/mailman/listinfo/bitcoin-dev
>>
>