Have you noticed that you pay a fee every time you make a cryptocurrency transaction? In this article we’ll look at these transaction fees and the essential role they play in Bitcoin.
Fees reward miners
In Bitcoin, transactions issued by users on the network are assembled into blocks that are confirmed by miners. The result is a blockchain that contains all the transactions made and represents the system’s record of ownership. To ensure the validity of this chain, miners are rewarded with Bitcoins for their activity, which makes Bitcoin an economic system.
Part of the miners’ reward comes from money creation (sometimes called “subsidy”), which is determined by a policy set in 2009. This money creation was 50 Bitcoins per block in 2009 and has since been halved every 210,000 blocks, or about every 4 years, during events called halvings. The reward today is 6.25 bitcoins per block.
In the long term, this monetary creation is intended to tend towards zero, and should pass under the satoshi (i.e. be effectively zero unless the protocol is changed) in 2140. This subsidy should gradually be replaced by what constitutes the other part of the block reward: the transaction fees. This evolution has been planned from the beginning, as evidenced by the Bitcoin white paper in which Satoshi Nakamoto writes:
“Once a predetermined number of coins have been put into circulation, the incentive can be funded entirely by the transaction fee and no longer require inflation.”Satoshi Nakamoto, October 31, 2008
The transaction fee system is fully integrated into Bitcoin and plays a significant role in the economics of mining. By convention, every transaction has an implicit fee, which is defined as the difference between the total value of the inputs and the total value of the outputs. This difference can be zero (free transaction), but it is always accounted for.
This mechanism (present in Bitcoin since the beginning) is intended to avoid unnecessary transaction and block costs. When building a block, the miner observes how much each transaction pays in fees by calculating the difference between the inputs and outputs. He then collects all the transaction fees of the block by adding the total to the reward transaction (coinbase). Finally, it validates the block by adding the corresponding proof of work and receives the bitcoins from the fees that are added to the newly created bitcoins by the protocol.
Transaction fees therefore help to reward miners for their service. In particular, they serve to partially reward the purely technical service related to the system, which requires costs. Coupled with money creation, they compensate for the cost of electricity, hardware infrastructure (ASIC, storage, bandwidth, etc.) and software maintenance, but also for the risk of invalidating mined blocks (orphaning risk). As money creation decreases, the fees will increasingly be used to fund all mining operations.
The fee market
As everyone knows, Bitcoin does not have unlimited transactional capacity. In addition to the natural limit that technically discourages miners from processing too many transactions, there is a consensus rule that artificially restricts the transactional capacity of the network. This was originally a block size limit and prevented blocks from being larger than 1 MB. Since the activation of SegWit in August 2017, the capacity of BTC blocks is limited to 4 million weight units, weight being a specific quantity, which corresponds to (at most) 12 transfers per second.
This artificial limit is intended to keep the cost of managing a node at a given level and thus preserve the decentralization of the network (or rather its possibility of decentralization). Indeed, the less data processed, the less expensive the hardware to run a node, mining or non-mining, is.
In addition, this artificial transactional limit has the effect of making block space scarcer (inelastic supply) and thereby encouraging the growth of the “fee market” as demand increases. During periods of congestion, users compete for this scarce block space, which creates an auction effect on transaction fees. Users then have the choice to increase their fees or wait longer for confirmation of their transactions.
Since the limiting technical factor is the size or weight of the data, fees are usually calculated based on the size or weight of the transaction, not the value it moves. Initially, the fee rate was measured in satoshis per byte (sat/o); since SegWit it is measured in satoshis per virtual byte (sat/ov). Miners select transactions according to this rate in order to be as profitable as possible.
The fee is therefore used to discriminate access to the block space. Thanks to the transcational capacity artificially limited by the protocol, this increases the level of fees and allows the miners’ income to increase, which finances the security of the network.
Resistance to censorship
However, the function of fees is not limited to replacing monetary creation for incentive and discriminating access to block space: they play a much more important role in resisting Bitcoin censorship.
In Bitcoin, censorship can mean two things:
- Passive censorship, which consists of refusing to confirm a certain group of transactions for reasons that are not economically rational.
- Active censorship, which consists of preventing the transactions in question from being confirmed by making any block containing one invalid.
Censorship is not impossible in Bitcoin: each miner is free to include a transaction in his or her block or not, and to accept a block or not, in a completely subjective way. But Bitcoin is said to be censorship-resistant, i.e. it has the property of making it difficult to hinder transactions.
Passive censorship is not a major concern, except for the confirmation delay caused by such a refusal. Even if this refusal is general, an honest minor will eventually confirm the transaction after some time. Active censorship, on the other hand, is much more problematic, as it actually prevents transactions from taking place and thus damages Bitcoin’s value proposition.
Money creation plays a negligible role in combating censorship, whether passive or active.
- On the one hand, the reward for money creation is the same for all miners, so it does not influence their economic choice of whether to include a transaction or not, and thus has no influence against passive censorship.
- On the other hand, a drop in utility and price as a result of active censorship is not a sufficient reason, because the censor is by nature an entity that could be described as “irrational”: its goal is to control the chain by deciding which transactions are allowed and which are not. Typically, this will be a state seeking to impose regulation on the blockchain, in order to preserve its power to levy. So the censor doesn’t care about destroying Bitcoin’s usefulness in the process; its goal is control.
It is, as has been said, the transaction fees that serve an essential purpose in resisting censorship. Not only do these fees combat passive censorship by encouraging as many miners as possible to confirm the transaction, but they also intervene in the case of active censorship.
This role of fees in resisting censorship has been clearly described by Eric Voskuil in his Property of Resisting Censorship. Here’s how he thinks they might restore the situation in the face of active censorship:
- First, a censor acquires more than half of the network’s computing power in order to apply active censorship to a group of transactions. Typically, this is a state that otherwise raises revenue through control over its currency, through taxation and money creation. The censor can mine itself or forcefully impose restrictions on existing miners.
- The initiators of censored transactions, seeing that their transactions are not confirmed, increase their fees. This is a natural behavior that is already observed during periods of network congestion.
- This increase creates an extra charge between authorized transactions and total transactions, which encourages honest miners to deploy more computing power over time.
- Once enough computing power is deemed sufficient (which could be done via flagging), the honest miners begin to confirm censored transactions. Since their computing power is the majority, their chain becomes the longest and the censor’s chain is invalidated. In this way, censorship is defeated, at least temporarily.
The role of transaction fees is thus to compensate for the political risks taken by miners to confirm censored transactions. Indeed, if such censorship were to be practiced, one can imagine that the state in question would have drastically regulated or banned mining beforehand, and would have done everything in its power to attack the world’s major mining operations head-on.
It seems that Satoshi never made this essential role clear. At the time, he only explained why his system was economically secure against double-spending and how it solved the Byzantine general problem in a robust way, which was already a great evolution from previous systems. He did not, however, describe why the system could resist censorship (i.e., partial or total blocking by a hostile entity) and seemed to rely on the voluntary goodwill of “honest” miners, even thinking that there would “probably always be nodes willing to process transactions for free.
However, this mechanism has been built into the protocol from the beginning and is, in my opinion, essential to Bitcoin’s survival.
Fees limit Bitcoin’s usefulness
The final aspect of fees that is too often overlooked is their (negative) propensity to limit Bitcoin’s utility. This is because Bitcoin is not infinitely scalable, no matter what artificial transactional capacity is chosen, and therefore a selection in transactions must always be made according to the fees they pay.
What is important to understand is that fees exclude uses of Bitcoin as their average level increases. As mentioned, the fees increase with the demand for the block space, which has a limited supply. Thus, over time, the demand for block space increases and the bidding for that space rises accordingly, so that uses requiring a fairly low level of fees are excluded from Bitcoin.
In the case of BTC, there is a low limit to the transactional capacity, and the average fee level tends to be very high in relation to the number of transactions made. Today the network nodes do not relay transactions that pay a fee rate below 1 sat/ov (about 4 euro cents for a traditional transfer) and the transaction memory area is regularly “overloaded”, so the median fee fluctuates between 1 and 25 € per transaction.
Thus, blockchain transactions are currently not very suitable for everyday payments, such as buying a coffee for example. If a person consumes one coffee a day for a price of €2 and still persists in using the BTC chain to buy that coffee, paying €1 in fees each time, that would be a 50% fee. Even if that coffee is purchased in the underground economy and the merchant gets a direct discount due to the absence of VAT and other taxes, one can imagine that the situation is hardly advantageous, as the tax avoidance and possible bank fees would probably not be enough to offset the Bitcoin fees (not to mention the risk cost for the merchant).
Some will argue that there are overlay solutions such as the Lightning network or sidechain on which the fees are very low. However, even with Lightning, the system has a limit, directly related to the cost of settlement on the chain.
Let’s assume that in the not-too-distant future Bitcoin gains popularity and that this success brings the median transaction fee to €100 per transaction to open or close a channel (in reality the costs are distinct, but we’ll simplify). Then there will quickly be a threshold where it will no longer be profitable for an individual to use Lightning.
Indeed, even if a person used Lightning abundantly, he would have to open or close a channel from time to time, if only because of technical incidents and rebalancing. If the person made a profit of €50,000 per year from his activities on Lightning and made 10 channel openings/closures per year, then the cost of the channel fees would be 2% of his income (not counting the fees on Lightning), which would be a significant cost.
All of this puts into perspective the wild predictions one hears here and there about the price of bitcoin. Indeed, Bitcoin’s potential utility is not unlimited, and its price, which is the rational consequence of that utility, cannot therefore go “to the moon.
Bitcoin is still useful, however small its transactional capacity and however enormous its costs. It is not suitable for very small transactions, but it is perfectly suitable for large international transfers, and it is always suitable for opening channels on the Lightning network.
Thus, transaction fees are a very important aspect of Bitcoin. Besides helping to pay miners for their technical service, they serve to manage the block space optimally and to ensure that the protocol is censorship-proof. Keep this in mind the next time you pay a mining fee.