It is ironic that CBOE will launch futures on bitcoin (the currency) on Sunday of the very week that bitcoin (the blockchain) demonstrated how clearly it is broken. As a medium of exchange, it does not even compare favorably at this point with a bank wire: transactions can cost more, take longer, and be less certain. As of Friday, the average transaction fee was $15, the average transaction time over the last seven days has been nearly two hours, and the backlog (“mempool”) of unconfirmed transactions is at all-time highs and probably worth over $3 billion. In contrast, a Ripple exec reported moving ~$14 billion on Thursday in 15 minutes at a cost of less than $1. There is no obvious fix for bitcoin because the professional miners who now control the network are not aligned with users in seeking lower transaction fees.
Ripple achieves vastly better network performance through dropping the usual chain requirement for global consensus, where all users reach agreement, and operating a permissioned protocol where users restrict the ability to verify transactions to those with whom they have the most trust and engagement; and, crucially, users can change the voteweight of these “validators” and therefore marginalize those who perform poorly. The result is a more efficient and adaptive protocol for transaction verification where control resides with users rather than, in the case of bitcoin, with miners (or, as for chains using proof-of-stake rather than bitcoin’s proof of work approach to consensus-formation, with coinholders). The Ripple blockchain confirms typical transactions in under five seconds and can handle over 1,000 transactions per second (or tps) versus 7-15 for bitcoin.
Blockchains will achieve their transformational potential as a database technology only if transaction fees can be reduced to make very low-value (or VLV) payments economic. VLV payments matter for two reasons. First, they can be integrated into a business process so that, for example, marketplaces can pay vendors at time-of-service (think Uber and a driver in, say, Mexico) and content providers can evolve revenue models beyond sharing advertising proceeds with platforms (think replacing the “please donate” messages you sometimes see beside a crypto address underneath YouTube videos with pay-to-view access control). Second, larger payments can be packetized so as to diversify routing (by path and in time) and hence reduce settlement risk.
Bitcoin is irredeemably broken as a medium of exchange
In March this year, Miguel Vias, part of whose job is to bring institutional grade liquidity to one of the digital assets XRP, commented, “what you’ve seen … is the smart crypto-money flow out of bitcoin and into other digital assets right as the fiat money has flowed in.” He adds that the crypto-space is well aware of the challenges for the bitcoin blockchain around “scaleability issues, the fact that confirmation times are slow, and there is chatter about a hard fork because the community can’t quite seem to figure out what’s best for itself.” The network performance over the last few days, and particularly the congested mempool (backlog of unconfirmed transactions), puts an exclamation point on these challenges. As of Friday morning, there were over 200,000 unconfirmed transactions and average confirmation times over the last seven days are nearly two hours. In November, this metric reached ~7 hours. Furthermore, the average transaction fee is nearly $15. Bitcoin is clearly broken as a medium of exchange. At this point, bank wires can be cheaper, faster, and more certain. The deeper question is whether the situation is redeemable, and the thesis of this note is that it is not.
Our thesis is that bitcoin as a medium of exchange cannot be rescued because of the way the protocol works (and the emergence of better consensus-formation algorithms since the original paper on distributed ledger technology or “DLT” was published in 2008), and the lack of a governance structure to allow for orderly adaptation. As context, the defining distinction between a traditional database and a blockchain database (recall that blockchains are databases and, in the special case where they track asset-ownership, ledgers) is that transaction-confirmation is handled by a voting algorithm rather than a central trusted administrator. The distinctive characteristics of blockchains emerge from differences in how this voting algorithm operates to generate consensus around which database updates (aka transactions) are confirmed: who votes, how much weighting attaches to each participant’s vote, and how the chain is guarded against “Sybil” attacks (where a malicious user fraudulently casts multiple votes).
Bitcoin uses a “proof of work” (or POW) protocol in which the voteweight accorded to any participant depends on their ability to prove computational effort through solving a cryptographic puzzle. This protects against Sybil attacks (no work, no vote) and was inspired by the “hashcash” idea developed in 1992 (so well before bitcoin and not, despite its name, originally envisioned in a currency context) for limiting e-mail spam and other frivolous use of a shared resource by requiring a user to compute a moderately hard, but not intractable, function in order to gain access; the functions are called hash functions and the ability to compute with them referred to as hash power. In a bitcoin context, doing this computational work to confirm transactions is referred to as “mining” and, as Ripple CTO Stefan Thomas points out, part of the governance problem for the Bitcoin blockchain arises as hash power has concentrated in the hands of “professional” miners rather than being distributed among general users as originally intended. The result is that protocol-changes require the agreement of these miners whose financial interest, unlike that of general users, is not aligned with the lowest possible transaction fees.
To address this, the crypto-community is experimenting with blockchains that use different consensus-formation algorithms such as proof-of-stake (or POS) where vote-weight is linked to wealth as measured by holdings of the chain’s native token or “coin” in which the network fee is denominated. An advantage of the approach over POW is that it is deterministic so that blocks of transactions are confirmed in uniform intervals whereas POW is inherently probabilistic (the time taken to solve the cryptographic puzzle is variable) giving rise to unevenness in block-confirmation times. Stefan Thomas quantified this by looking at block confirmation times over a day earlier in the year, and his network metric results are striking. However, the POS protocol rewards coin-holders over those who contribute to the community through transacting although some blockchains, such as NEM, look to address this with a modified POS protocol that assigns vote weight based on a hybrid of a participant’s account balance and transaction volumes.
The natural question is how Ripple is able to confirm transactions so much more quickly (lowest average delay) and with so much more certainty (lowest standard deviation in both absolute terms and in terms of the coefficient of variation measuring the ratio of the standard deviation to the mean). The answer is that the protocol does not insist on a global consensus of all blockchain users but rather operates a “permissioned” system where users restrict the power to verify transactions to those with whom they have the most trust and engagement and can change the vote-weight of these “validators” depending on performance.
This obviates the need for other protection against Sybil attacks, whether through POW or POS, since validators who cast fraudulent votes find that users withdraw trust (by reducing their voteweighting) so that, in the extreme, they are marginalized. In other words, network control is in the hands of users who care about low transaction fees rather than, in the case of bitcoin, miners who care about high transaction fees (or, as in some other chains, coinholders who care about store of value over medium of exchange).
RenMac: Analyzing macro factors that impact the investment world.