[Bitcoin-development] Solution for Block Size Increase

Nicolas DORIER nicolas.dorier at gmail.com
Thu May 7 23:14:22 UTC 2015

Executive Summary:

I explain the objectives that we should aim to reach agreement without
drama, controversy, and relief the core devs from the central banker role.
(As Jeff Garzik pointed out)
Knowing the objectives, I propose a solution based on the objectives that
can be agreed on tomorrow, would permanently fix the block size problem
without controversy and would be immediately applicable.

The objectives:

There is consensus on the fact that nobody wants the core developers to be
seen as central bankers.
There is also consensus that more decentralization is better than less.
(assuming there is no cost to it)

This means you should reject all arguments based on economical, political
and ideological principles about what Bitcoin should become. This includes:

1) Whether Bitcoin should be storage of value or suitable for coffee
2) Whether we need a fee market, block scarcity, and how much of it,
3) Whether we need to periodically increase block size via some voodoo
formula which speculate on future bandwidth and cost of storage,

Taking decisions based on such reasons is what central bankers do, and you
don’t want to be bankers. This follow that decisions should be taken only
for technical and decentralization considerations. (more about
decentralization after)

Scarcity will evolve without you taking any decisions about it, for the
only reason that storage and bandwidth is not free, nor a transaction,
thanks to increased propagation time.
This backed in scarcity will evolve automatically as storage, bandwidth,
encoding, evolve without anybody taking any decision, nor making any
speculation on the future.

Sadly, deciding how much decentralization should be in the system by
tweaking the block size limit is also an economic decision that should not
have its place between the core devs. This follow :

4) Core devs should not decide about the amount of suitable
decentralization by tweaking block size limit,

Still, removing the limit altogether is a no-no, what would happen if a
block of 100 GB is created? Immediately the network would be decentralized,
not only for miners but also for bitcoin service providers. Also, core devs
might have technical consideration on bitcoin core which impose a temporary
limit until the bug resolved.

The solution:

So here is a proposal that address all my points, and, I think, would get a
reasonable consensus. It can be published tomorrow without any controversy,
would be agreed in one year, and can be safely reiterated every year.
Developers will also not have to play politics nor central banker. (well,
it sounds to good to be true, I waiting for being wrong)

The solution is to use block voting. For each block, a miner gives the size
of the block he would like to have at the next deadline (for example, 30
may 2015). The rational choice for them is just enough to clear the memory
pool, maybe a little less if he believes fee pressure is beneficial for
him, maybe a little more if he believes he should leave some room for
increased use.
At the deadline, we take the median of the votes and implement it as a new
block size limit. Reiterate for the next year.

Objectives reached:

   - No central banking decisions on devs shoulder,
   - Votes can start tomorrow,
   - Implementation has only to be ready in one year, (no kick-in-the-can)
   - Will increase as demand is growing,
   - Will increase as network capacity and storage is growing,
   - Bitcoin becomes what miners want, not what core devs and politician
   - Implementation reasonably easy,
   - Will get miner consensus, no impact on existing bitcoin services,


   - Effect on bitcoin core stability (core devs might have a valid
   technical reason to impose a limit)
   - Maybe a better statistical function is possible

Additional input for the debate:

Some people were debating whether miners are altruist or act rationally. We
should always expect them to act rationally, but we should not forget the
peculiarity of TCP backoff game: While it is in the best interest of
players to NOT reemit TCP packet with a backoff if the ACK is not received,
everybody does it. (Because of the fallacy that changing a TCP
implementation is costless)

Often, when we think a real life situation is a prisoner dilemma problem,
it turns out that the incentives where just incorrectly modeled.

Core devs, thanks for all your work, but please step out of the banker's
role and focus on where you are the best, I speak as an entrepreneur that
doesn't want decisions about bitcoin to be taken by who has the biggest.
If the decision of the hard limit is taken for other than purely technical
decisions, ie, for the maximization of whatever metric, it will clearly put
you in banker's shoes. As an entrepreneur, I have other things to speculate
than who gets the biggest gun in the core team.
Please consider my solution,

Nicolas Dorier,
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.linuxfoundation.org/pipermail/bitcoin-dev/attachments/20150508/6a845fc6/attachment.html>

More information about the bitcoin-dev mailing list