Research

[ben]

Hello,

this is a general research thread, since I don't want to post into slack channel which is dedicated to more operational development. I'll be mostly focusing on blocksize, scaling, accounting and assets. Note: this represents my own view on the ColoredCoin exchange architecture, and relation to Bitcoin.

The scaling problem

What are some of the biggest open problems in this new emerging field? One is scaling and the other is market-mechanisms on- and off-blockchain. I will argue that both are strongly related.

It could be helpful to specify some assumptions in current exchange and transaction model (Bitcoin, ColoredCoins and Lykke). The transaction verification and transmission in Bitcoin comes at a cost. If one assumes its risk-free the (historical) cost of a transaction can be seen here: http://i.imgur.com/DrbyUim.png, see also http://i.imgur.com/Qm6jf4v.png The cost of a global Bitcoin based transaction is very low. So for example this transaction e.g. [1], transmitted ca. 50 million USD for 1.60$. That beats all traditional banking. Perhaps banks can't even process transactions internally at this cost (0.1 basis points would still be 500$). The question is however what the longterm properties of the sytem is (even if one only considers Bitcoin as plain digital cash ignoring regulatory and consensus issues). Currently Bitcoin is based on subsidies, which in somes avoids the transaction fee problem. Each block pays 25 BTC to the node who did the verification, while the fees are usually only a small fraction of that. As long as fees/block-reward is very small, the economics are supported by inflation and I'd argue the long-term implications unknown.

Value per transfer

The blocksize debate, which is a significant development in Bitcoin itself, is in my opinion misdirected. Fees for transactions should be determined ideally by some kind of demand and supply relationship, but Bitcoin does not offer such a mechanism yet. It is difficult to determine the potential effects of capacity increases.

[1] https://blockchain.info/tx/2c2ef42751873...2fbaee2d60

Instead of focusing only on the 1 megabyte per block, and one can consider the value per average transaction (value-transfer ratio). Each transaction is typically 250 bytes large which therefore allows 4000 transactions per block. Currently in all discussions and efforts it is assumed that the average value per transaction is a constant.

Scaling with assets

There is not necessarily a restriction to a constant value transferred given fixed transaction capacity. If e.g. there would be a consensus on some threshold for instance the value transfer capacity can increase indefinitely. This is because one byte can potentially encode any value, not just native Bitcoin. This fact to my knowledge has not been yet described, except maybe by Hal Finney in 2010. It implies that the Bitcoin blockchain potentially can process any value amount (100 trillion USD for example), if the encoded transport is not Bitcoin itself. There are some issue with this however, since Bitcoin's consensus is not based on derivative assets, but the native currency.

In other words, even with the known system that is Bitcoin's algorithm, it is possible for such an algorithm to support value transport in the volume of the entire world economy if bytes can encode not just native Bitcoin. There are several implications of this observation. One is that a ledger which allows to encode assets in a better way could in that sense be more valuable, since the underlying currency encodes higher value transferred and therefore higher fees earned. While the Bitcoin system is flexible enough to allow for ColoredCoins it is not primarily designed for assets.

What could such a derivative asset be? Besides ColoredCoins as a forward for fiat money, it could also be a forward for Bitcoin itself. So for example 50% of all transactions could be forwards of Bitcoin delivery, and the other 50% native transactions. Given 4000 transactions per 10 minutes, this would imply 2000 Bitcoin transactions. If the average Bitcoin transaction is 100$, then this implies 200'000$ value transferred for the Bitcoin part. The other 2000 transactions could e.g. by delivery of 1M$ Bitcoins each, that is 2B$ transferred in one single block. Technically this is entirely possible.

[ben]

One analogy of the Bitcoin blockchain is a ledger. Since the beginning of finance it was always connected to how information is stored in records. The accounting, beginning in the 15th century, was a paper based record of transactions. By way of analogy Bitcoin is a distributed ledger which records transactions. This ledger is controlled by many party which reach agreement on this record - what enter it, and what the rules of these transactions are. The consensus stops abuse of the record system and allows anyone to participate freely. What is currently not well understood is the relationship to assets and potential bridges to the existing financial system ("bridges"). I want to go into some detail of Bitcoin's operation to explain the relevance of consensus for assets and bridges. Bridges are necessary when Bitcoin is converted to fiat currency (besides pure cash transactions via Localbitcoins for example).

How the Bitcoin ledger stores records is related to how it processes transactions. Processing transaction happens within a script system. One way to think about this is to imagine a ledger with some intelligence. The most basic operations for a ledger are:

* store something in it
* retrieve something from it
* process some transaction which moves some cash from A to B
* process some transaction which moves some asset from A to B

A more complicated case is anything that relates to time or several people. For instance:

* lock funds and make them available to no one, until time T in the future
* lock funds and only make them available if several parties agree on the spend conditions (in the paper world we have directors of corporations, trust managers or even parliaments discussing how communal funds should be spent)

[ben]

The Lightning network describes (in theory mostly) a mechanism how to provide some kind of back-end settlement, where most transactions happen off-chain, but "defaults" get settled on-chain. Transactions could happen in private, by attaching a default contract to it, so that in case of a dispute on-chain settlement happens. This is an interesting idea, which I believe still needs a lot of work. One interesting scenario is where organisations mostly do on-chain transactions this way, and scaling happens through these organsations. This to some degree already occurs in Bitcoin with web-wallets and exchanges.