[Cryptography] Standards Trolls: Re: Bitcoin is a disaster.

Ray Dillinger bear at sonic.net
Tue Jan 12 16:06:27 EST 2021


I'm pretty strongly of the opinion that you're not going to get 
scalability in any system in which one transaction no matter how
trivial results in communication to every node.

In looking for a scalable solution we're not looking for a way to get a
global ledger of transactions updated more efficiently; we're looking
for a way to not need a global ledger of transactions. 

TheWholeDamnUniverseTM is not an appropriately manageable unit to form
a distributed consensus around for the sake of a single transaction.
Universal agrements and distributed consensus about huge sets of facts
are cryptographically REALLY COOL!!! But not practical.  We need to
relax the consensus requirement and find a way to reach _enough_
_agreement_ _to_ _perform_ _transactions_ without reaching agreement
about TheWholeDamnUniverseTM. 

For a single transaction you need consensus that the coins you're
getting actually exist and actually belong to the person you're getting
them from.  You don't need a history of the universe for that; you need
a history of those particular coins and evidence that the act of
minting them was valid according to protocol.

If everybody just has an infinitely-divisible "balance" that they're
transferring some fraction of, that means you need evidence that the
counterparty's "balance" is valid, implying you need to know about
every fractional coin that went into every transaction that resulted in
that user's current "balance", which becomes the problem of agreeing on
TheWholeDamnUniverseTM. 

But if instead the universe consists of "tokens" in some standard
denominations such as 1/2/5 times some power of 10, or 1 times some
power of 2, then you only have to agree on the history of the
particular tokens being transferred, and that's an agreement that
doesn't require the cooperation of every!last!one! of a million, or a
billion, users.

If the tokens themselves are temporary - that is, if they are minted,
they go through some number of transactions, and then they are melted -
then the history of a given token can be short and require
communicating about only a few transactions. A $20 bill, for example,
only changes hands about 600 times before it is removed from
circulation and destroyed.  Agreeing about a few hundred transactions
seems a lot more reasonable than agreeing about TheWholeDamnUniverseTM.

				Bear




More information about the cryptography mailing list