The Collective Action Problem of Mining Fees

[Note: This was originally published on October 7, 2014 at]

Contrary to conventional wisdom, usage fees for many cryptocurrencies including Bitcoin, Litecoin and Dogecoin is voluntary. In fact, when Bitcoin was first released, there were no direct usage fees. The fee that users visibly see included within wallet software (such as BitcoinQT, Electrum, and Hive) is in fact, arbitrarily set by the wallet code and can – in most cases – manually be reduced to zero. That is to say, users can broadcast transactions (UTXOs) to the network for “free” and eventually a mining pool (assuming it the transaction amount is higher than the dust limit) will pick them up and include them into a block. In the meantime these transactions will float around in the mempool, sometimes for several hours waiting to be picked up and confirmed. In theory, the higher the fee a user includes, the faster it will likely be included into a block because mining pools have an incentive to package it.

Why can’t all users permanently “free ride” off of what effectively is a tragedy of the commons (Sybil protection via hashrate and access to the mempool)? There is another variable: blocks have limited space (currently set at 1 MB), are a scarce resource and thus by definition are not a public good. Fees enable users to effectively bid on this private good rationed by the labor force of miners (I describe this at length in Chapter 2). In the future, miners may actually begin to set fees themselves, as the core protocol development team is planning to “float” the fees at some point.

In practice today, “voluntary” fees (donations) represent about 0.3% of the wages a miner earns each day. How do network operators (miners) get paid then? Through block reward subsidies (inflation) which are awarded throughout the day. Nearly all hash-based proof-of-work coins use a similar method of payment in which a coinbase transaction is paid out at intervals – for Bitcoin it is roughly every 10 minutes, Litecoin every 2.5 minutes and Dogecoin every minute. As a reward for services rendered, a fixed amount of payment is sent to the miner who broadcasts the “correct” block first which is then added on top of the previous block of the longest chain (those who broadcast a different block after the fact are said to have worked on “orphaned” blocks).

In theory, based on section 6 of the original 2008 whitepaper, transaction fees are supposed to eventually replace this subsidy:

The incentive can also be funded with transaction fees. If the output value of a transaction is less than its input value, the difference is a transaction fee that is added to the incentive value of the block containing the transaction. Once a predetermined number of coins have entered circulation, the incentive can transition entirely to transaction fees and be completely inflation free.

A June 2014 paper published by Kerem Kaskaloglu attempts to illustrate the “ideal scenario” (shown below) of the seamless switch from block rewards (seigniorage subsidy) to transaction fees (donations).

bitcoin-blockreward-timelineAccording to the current narrative, a combination of increased transactional volume and higher mandatory (or perhaps marketed-based) fees will purportedly pay for the labor force to stick around in the coming years. Why is this important? Because the network currently operates almost entirely on subsidies and thus with each block reward halving, the labor force is essentially given notice that their wages have decreased 50% and many could leave for more profitable ventures. While this is not an immediate concern in October 2014 for a network like Bitcoin – which does not have a closely competing SHA-based chain to worry about attack from (yet) – other altchain designers need to be cognizant of the economic incentive model they are building before they launch a new coin.

Again this is an empirical matter, so it cannot be known a priori whether or not the transition from a subsidy to a fee will happen for Bitcoin or other coins. However what we do know is that based on the history of altcoins up to the date of this writing, an increase in fees is the exception rather than the rule. Very few altcoins have seen a markedly significant increase in usage fees over the life of their chain.

The two charts below illustrate this:

bitcoin-litecoin-transaction-feedogecoin-transaction-feesThe first chart shows the “voluntary” transaction fees from the previous 6 months for both Litecoin and Bitcoin. It is almost entirely flat which suggests a number of things including the fact that there probably has not been an increase in usage of either blockchain itself (other indices such as Bitcoin Days Destroyed and Total Volume Output would help narrow the amount of UTXOs on the move and to what extent). One exception is Counterparty, which on some days represents 3% of the Bitcoin network and whose assets may represent increasingly larger commercial transactions that are not fully measured yet.

The second chart shows the same time period but for Dogecoin. In this case, over the past 2 months there has been a steady increase in fees paid to the network, effectively doubling the spring and summer rates. At this time it is unclear why this is the case; almost all tipping is done off-chain in trusted third parties (so fees are usually not assessed) and the default fee for most Dogecoin clients are set at 1 dogecoin. We may learn later that it was due to the AuxPOW merge mine with Litecoin, asset issuance via Dogeparty, or perhaps this is a statistical outlier altogether.

Nevertheless, while it cannot be said for certain, it is unlikely that the voluntary fee mechanism will fully provide the type of income to incentivize the Bitcoin labor force to continue providing its services (hashing) because it is a collective action problem (Note: Robert Sams recently touched on this issue in a new article). After all, why would most or all users one day in the future collectively start to pay (potentially) several orders of magnitude more to have the same exact service (recall that the total network reward to miners for their services today ranges from $20 to $40 per transaction)? In practice the fees alone may not be enough.

While the jury is still out on the longevity of all altcoins, one argument is that some are substitute goods. If this is the case, when conditions change (such as an increase in fees) users may move over to what they perceive as a similar, cheaper service. Or in this case, a similar chain. What does a substitute good mean? One definition explains that:

This means a good’s demand is increased when the price of another good is increased. Conversely, the demand for a good is decreased when the price of another good is decreased.

Or in other words, if the fee for using the Bitcoin network increases and a user perceives that Litecoin, Dogecoin or another network is a substitute good, then they may switch over and use the other, cheaper network for their transactional needs. The increased demand of that network token then may in turn lead to higher market prices of that token. In theory this could lead to some kind of market equilibrium among chains, though in practice there are and likely will be other factors at play.

In summation, there are no real “fees” in Bitcoin as they are entirely voluntary and should probably be called “donations” (until miners require that a fee be included). These donations are entirely arbitrary and are probably not comparable to fees on other payment networks (such as Visa) which are mandatory and holistic (the interchange fee pays for all of Visa’s expenditures, no donations required). This is further described in Chapter 3 and the full costs of today’s subsidies are visibly illustrated in the Cost Per Transaction chart. For more information on interchange fees, Richard Brown recently wrote a highly recommended piece for readers.

The next post will discuss why token prices do not double for a token after a block reward halving.

Leave a Reply

Your email address will not be published. Required fields are marked *