[Cryptography] Electronic currency revived after 20-year hiatus

[Natanael]

Den 17 aug. 2016 06:48 skrev "Ray Dillinger" <bear at sonic.net>:
>
>
>
> On 08/16/2016 04:49 PM, Allen wrote:
>
> > Blockchains that can handle 20 billion transaction a day will be here
> > within a year.
>
> That is an interesting assertion.  I do not believe it.
>
> 20G transactions.  Assume 100 bytes per, and that's terribly generous
> in terms of how small you think you can get them.  2T bytes per day.
> You have a data plan that will allow you to download 2T bytes daily?
> You think you'll have one by the end of a year? You have a hard drive
> that can store 700 Tbytes for even one year's worth of block chain
> history? You think you'll have one by the end of a year?
>
> The only parties with that kind of connection and storage will
> perforce act as Trusted Parties to all other users.  That means
> they'll either Gox the users over and over, or they'll become
> just as heavily regulated, insured, and monitored as any
> conventional bank.  A banker is a banker is a banker.
>
> The vast majority of people are already letting strangers hold
> their keys and trusting that they'll get their bitcoins back.
> That means the online-wallet people and the exchanges function
> as banks.  And these banks are Trusted. When they embezzle
> (or even if they just get robbed), it leaves people broke.
>
> Ending the existence of Trusted parties and their ability to Gox
> people, if you recall, was sort of the point of the block chain.
>
> It failed.

Just like how Bitcoin solved the unsolvable Byzantine General's problem by
ignoring the original problem statement and going for a probabilistic
approximation, Bitcoin's got a solution here too.

You know Bitcoin's smart contracts? And how it supports multisignature
transactions, and timeouts? And that it's got "payment channels", where two
people can merge all their transactions they performed during a timeperiod
into just two (a commitment transaction and a settlement transaction on the
blockchain, where the two parties simultaneously keeps a non-final
settlement transaction ready in memory until they're done?

Now combine them.

You create a "wallet" where you commit money to a payment channel against a
Lightning Network routing node (server) for a time period, where that
server can't steal from you (because you never signed a transaction that
leaves the server with all your money). The network of Lightning Network
nodes has payment channels as their backbone links too.

So it kind of becomes an email for instant high volume transactions, with a
blockchain to settle on from time to time - you tell your server who on
what server to send X coins to, and sign the corresponding transaction to
enable it to transfer that sum. All the involved payment channels are
updated accordingly to reflect that transfer of money, updating the
non-final settlement transactions being held in memory.

Even freezing funds would fail, because after a certain amount of time of
non-cooperation from the server it will be possible to just withdraw it
singlehandedly.

You could question if they wouldn't just reverse a set of transactions that
pay out money and try to run with what's been committed to them. Of course
there's a number of reasons, based on game theory;

The total sum that the server will be able to claim for itself at any
period in time will be relatively small compared to the sums involved (any
attempts at inflating it would trigger a large number of warning systems
that flag unusual behavior). Thus the reward and incentive is small.

Any sign of reversing any transaction is provable thanks to settlement
transactions being signed and mutually updated (you just show the world the
signed transaction that the server reversed through sending a prior one to
the blockchain). Thus there's accountability, and high risk of getting
caught.

Because malice can be detected so quickly, the timeframe for attacks given
the automated systems involved is limited to mere seconds before you get
flagged.

The above points combined with that trust takes time to build, and that
fees for processing transactions is likely to be more profitable (analogous
to the Bitcoin mining situation), the incentive is to follow the protocol
and proceess transactions as told by the users.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20160817/f9a2d21f/attachment.html>