MIME-Version: 1.0
From: Devrandom <c1.bitcoin@niftybox.net>
Date: Wed, 01 Nov 2017 05:48:27 +0000
Message-ID: <CAB0O3SVjhG19R61B78hFCPwfwWemTXj=tOsvgAgoNbjFYXXAtg@mail.gmail.com>
To: "bitcoin-dev@lists.linuxfoundation.org"
	<bitcoin-dev@lists.linuxfoundation.org>
Content-Type: multipart/alternative; boundary="001a1141025c521d73055ce56c87"
Subject: [bitcoin-dev] Introducing a POW through a soft-fork
Precedence: list

--001a1141025c521d73055ce56c87
Content-Type: text/plain; charset="UTF-8"

Hi all,

Feedback is welcome on the draft below.  In particular, I want to see if
there is interest in further development of the idea and also interested in
any attack vectors or undesirable dynamics.

(Formatted version available here:
https://github.com/devrandom/btc-papers/blob/master/aux-pow.md )

# Soft-fork Introduction of a New POW

## Motivation:

- Mitigate mining centralization pressures by introducing a POW that does
not have economies of scale
- Introduce an intermediary confirmation point, reducing the impact of
mining power fluctuations

Note however that choice of a suitable POW will require deep analysis.
Some pitfalls include: botnet mining, POWs that seem ASIC resistant but are
not, unexpected/covert optimization.

In particular, unexpected/covert optimizations, such as ASCIBOOST, present
a potential centralizing and destabilizing force.

## Design

### Aux POW intermediate block

Auxiliary POW blocks are introduced between normal blocks - i.e. the chain
alternates between the two POWs.
Each aux-POW block points to the previous normal block and contains
transactions just like a normal block.
Each normal block points to the previous aux-POW block and must contain all
transactions from the aux-POW block.
Block space is not increased.

The new intermediate block and the pointers are introduced via a soft-fork
restriction.

### Reward for aux POW miners

The reward for the aux POW smoothly increases from zero to a target value
(e.g. 1/2 of the total reward) over time.
The reward is transferred via a soft-fork restriction requiring a coinbase
output to an address published in the
aux-POW block.

### Aux POW difficulty adjustment

Difficulty adjustments remain independent for the two POWs.

The difficulty of the aux POW is adjusted based on the average time between
normal block found
to aux block found.

Further details are dependent on the specific POW.

### Heaviest chain rule change

This is a semi-hard change, because non-upgraded nodes can get on the wrong
chain in case of attack.  However,
it might be possible to construct an alert system that notifies
non-upgraded nodes of an upcoming rule change.
All blocks are still valid, so this is not a hardforking change.

The heaviest chain definition changes from sum of `difficulty` to sum of:

    mainDifficulty ^ x * auxDifficulty ^ y

where we start at:

    x = 1; y = 0

and end at values of x and y that are related to the target relative
rewards.  For example, if the target rewards
are equally distributed, we will want ot end up at:

    x = 1/2; y = 1/2

so that both POWs have equal weight.  If the aux POW is to become dominant,
x should end small relative to y.


## Questions and Answers

- What should be the parameters if we want the aux POW to have equal
weight? A: 1/2 of the reward should be transferred
to aux miners and x = 1/2, y = 1/2.

- What should be the parameters if we want to deprecate the main POW?  A:
most of the reward should be transferred to
aux miners and x = 0, y = 1.  The main difficulty will tend to zero, and
aux miners will just trivially generate the
main block immediately after finding an aux block, with identical content.

- Wasted bandwidth to transfer transactions twice?  A: this can be
optimized by skipping transactions already
transferred.

- Why would miners agree to soft-fork away some of their reward?  A: they
would agree if they believe that
the coins will increase in value due to improved security properties.

## Open Questions

- After a block of one type is found, we can naively assume that POW will
become idle while a block of the other type is being mined.  In practice,
the spare capacity can be used to find alternative ("attacking") blocks or
mine other coins.  Is that a problem?
- Is selfish mining amplified by this scheme for miners that have both
types of hardware?

## POW candidates

- SHA256 (i.e. use same POW, but introduce an intermediate block for faster
confirmation)
- Proof of Space and Time (Bram Cohen)
- Equihash
- Ethash

## Next Steps

- evaluate POW candidates
- evaluate difficulty adjustment rules
- simulate miner behavior to identify if there are incentives for
detrimental behavior patterns (e.g. block withholding / selfish mining)
- Protocol details

## Credits

Bram Cohen came up with a similar idea back in March:
https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2017-March/013744.html

--001a1141025c521d73055ce56c87
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr"><span style=3D"color:rgb(33,33,33);font-size:13px">Hi all,=
</span><div style=3D"color:rgb(33,33,33);font-size:13px"><br></div><div sty=
le=3D"color:rgb(33,33,33);font-size:13px">Feedback is welcome on the draft =
below.=C2=A0 In particular, I want to see if there is interest in further d=
evelopment of the idea and also interested in any attack vectors or undesir=
able dynamics.</div><div style=3D"color:rgb(33,33,33);font-size:13px"><br><=
/div><div style=3D"color:rgb(33,33,33);font-size:13px">(Formatted version a=
vailable here:=C2=A0<a href=3D"https://github.com/devrandom/btc-papers/blob=
/master/aux-pow.md">https://github.com/devrandom/btc-papers/blob/master/aux=
-pow.md</a> )</div><div style=3D"color:rgb(33,33,33);font-size:13px"><br></=
div><div style=3D"color:rgb(33,33,33);font-size:13px"><div># Soft-fork Intr=
oduction of a New POW</div><div><br></div><div>## Motivation:<br></div><div=
><br></div><div>- Mitigate mining centralization pressures by introducing a=
 POW that does not have economies of scale</div><div>- Introduce an interme=
diary confirmation point, reducing the impact of mining power fluctuations<=
/div><div><br></div><div>Note however that choice of a suitable POW will re=
quire deep analysis.=C2=A0 Some pitfalls include: botnet mining, POWs that =
seem ASIC resistant but are not, unexpected/covert optimization.</div><div>=
<br></div><div>In particular, unexpected/covert optimizations, such as ASCI=
BOOST, present a potential centralizing and destabilizing force.</div><div>=
<br></div><div>## Design</div><div><br></div><div>### Aux POW intermediate =
block</div><div><br></div><div>Auxiliary POW blocks are introduced between =
normal blocks - i.e. the chain alternates between the two POWs.</div><div>E=
ach aux-POW block points to the previous normal block and contains transact=
ions just like a normal block.</div><div>Each normal block points to the pr=
evious aux-POW block and must contain all transactions from the aux-POW blo=
ck.</div><div>Block space is not increased.</div><div><br></div><div>The ne=
w intermediate block and the pointers are introduced via a soft-fork restri=
ction.</div><div><br></div><div>### Reward for aux POW miners</div><div><br=
></div><div>The reward for the aux POW smoothly increases from zero to a ta=
rget value (e.g. 1/2 of the total reward) over time.</div><div>The reward i=
s transferred via a soft-fork restriction requiring a coinbase output to an=
 address published in the</div><div>aux-POW block.</div><div><br></div><div=
>### Aux POW difficulty adjustment</div><div><br></div><div>Difficulty adju=
stments remain independent for the two POWs.</div><div><br></div><div>The d=
ifficulty of the aux POW is adjusted based on the average time between norm=
al block found</div><div>to aux block found.</div><div><br></div><div>Furth=
er details are dependent on the specific POW.</div><div><br></div><div>### =
Heaviest chain rule change</div><div><br></div><div>This is a semi-hard cha=
nge, because non-upgraded nodes can get on the wrong chain in case of attac=
k.=C2=A0 However,</div><div>it might be possible to construct an alert syst=
em that notifies non-upgraded nodes of an upcoming rule change.</div><div>A=
ll blocks are still valid, so this is not a hardforking change.</div><div><=
br></div><div>The heaviest chain definition changes from sum of `difficulty=
` to sum of:</div><div><br></div><div>=C2=A0 =C2=A0 mainDifficulty ^ x * au=
xDifficulty ^ y</div><div><br></div><div>where we start at:</div><div><br><=
/div><div>=C2=A0 =C2=A0 x =3D 1; y =3D 0</div><div><br></div><div>and end a=
t values of x and y that are related to the target relative rewards.=C2=A0 =
For example, if the target rewards</div><div>are equally distributed, we wi=
ll want ot end up at:</div><div><br></div><div>=C2=A0 =C2=A0 x =3D 1/2; y =
=3D 1/2</div><div><br></div><div>so that both POWs have equal weight.=C2=A0=
 If the aux POW is to become dominant, x should end small relative to y.</d=
iv><div><br></div><div><br></div><div>## Questions and Answers</div><div><b=
r></div><div>- What should be the parameters if we want the aux POW to have=
 equal weight? A: 1/2 of the reward should be transferred</div><div>to aux =
miners and x =3D 1/2, y =3D 1/2.</div><div><br></div><div>- What should be =
the parameters if we want to deprecate the main POW?=C2=A0 A: most of the r=
eward should be transferred to</div><div>aux miners and x =3D 0, y =3D 1.=
=C2=A0 The main difficulty will tend to zero, and aux miners will just triv=
ially generate the</div><div>main block immediately after finding an aux bl=
ock, with identical content.</div><div><br></div><div>- Wasted bandwidth to=
 transfer transactions twice?=C2=A0 A: this can be optimized by skipping tr=
ansactions already</div><div>transferred.</div><div><br></div><div>- Why wo=
uld miners agree to soft-fork away some of their reward?=C2=A0 A: they woul=
d agree if they believe that</div><div>the coins will increase in value due=
 to improved security properties.</div><div><br></div><div>## Open Question=
s</div><div><br></div><div>- After a block of one type is found, we can nai=
vely assume that POW will become idle while a block of the other type is be=
ing mined.=C2=A0 In practice, the spare capacity can be used to find altern=
ative (&quot;attacking&quot;) blocks or mine other coins.=C2=A0 Is that a p=
roblem?</div><div>- Is selfish mining amplified by this scheme for miners t=
hat have both types of hardware?</div><div><br></div><div>## POW candidates=
</div><div><br></div><div>- SHA256 (i.e. use same POW, but introduce an int=
ermediate block for faster confirmation)</div><div>- Proof of Space and Tim=
e (Bram Cohen)</div><div>- Equihash</div><div>- Ethash</div><div><br></div>=
<div>## Next Steps</div><div><br></div><div>- evaluate POW candidates</div>=
<div>- evaluate difficulty adjustment rules</div><div>- simulate miner beha=
vior to identify if there are incentives for detrimental behavior patterns =
(e.g. block withholding / selfish mining)</div><div>- Protocol details</div=
><div><br></div><div>## Credits</div><div><br></div><div>Bram Cohen came up=
 with a similar idea back in March:</div><div><a href=3D"https://lists.linu=
xfoundation.org/pipermail/bitcoin-dev/2017-March/013744.html" target=3D"_bl=
ank">https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2017-March/013=
744.html</a></div></div></div>

--001a1141025c521d73055ce56c87--