One often discussed topic lately has been the carbon footprint of blockchain ecosystems. Proof-of-work systems are enormously successful in securing transactions and preventing fraud and collusion, but they consume an enormous amount of energy. Current estimates are that a single bitcoin transaction has a carbon footprint of over 350 kilograms (772 pounds) and a single Ethereum transaction consumes about 39 kg.

Compared to centralized transaction processing systems, these are hundreds of times less efficient. The same carbon footprint for a single Ethereum transaction would power more than 80,000 transactions on a centralized credit card payment network. If the desire is to replace all banking systems and small payments with blockchains, we are in trouble. That would be true even after a shift to proof of stake, which is thought to be 99% more efficient. In my view, it is unlikely that blockchains will replace most banking and credit card transactions today. As a result, the comparisons in carbon footprint don’t make sense either. As I have previously argued, blockchain transactions offer no benefits and many costs compared with most simple consumer transactions – or even many smaller enterprise transactions, like payroll.

Where blockchain-based transactions have a compelling advantage over alternatives is when they are put in the context of the total carbon required for a complex multi-party transaction that’s part of a larger business process. In these scenarios, the alternative isn’t an efficient credit card transaction, it’s the cost and complexity of having a human being in the loop – and the carbon-cost of the working time involved.

The best example I have is a typical business purchase order. There are many estimates out there, but on average, it seems that large enterprises typically spend between $50 and $100 to raise a single purchase order or pay an invoice. Almost none of this is related to the information technology cost. Instead, the big driver is the cost of human time required to verify that the purchase order or invoice complied with the company’s rules for payments and any existing contracts. In other words: we’re talking about labor here. At current average rates in professional and business services, we’re talking about one to two hours of work.

The real comparison, then, isn’t between an Ethereum transaction and a credit card transaction, it’s between an Ethereum transaction and an hour of labor. In the U.S., a typical employee (using California data from Buffer.com) requires about 4,000 kg carbon to cover the office space and commute each year. Spreading that across 2,000 working hours (I’m rounding up and down to keep the math simple.) means we can assign about 2 kg of carbon to each working hour here in the Golden State.

All of this is, for the most part, highly speculative. Estimates of carbon footprints are built on SWAGs (Scientific Wild-*** Guess), multiplied together and scaled up. Errors and incorrect assumptions get magnified in these types of projections, which can lead to results that are very far off from reality. Big questions, like the amount of energy in the mix that is renewable, are not easy to answer because they are so variable and few authoritative data sources. A shift to more renewable energy will further reduce the footprint of these systems.

The bottom line is that it is impossible to give a precise answer, but what this little thought exercise does show, is that just comparing one transaction to another by system isn’t useful. You cannot compare a business purchase order with shared rules and multiple participants to a credit card transaction.

Trustworthy data is costly — from a human time or computation standpoint. We can use blockchain technology to reduce the carbon footprint of our industrial world while eliminating operational complexity and redundant human work.

The views reflected in this article are the views of the author and do not necessarily reflect the views of the global EY organization or its member firms.