I didn’t, and still don’t completely, understand the trade-offs between transaction fees vs. inflation in blockchain protocols (i.e., no transaction fees such as EOS).
Fortunately, Charles Hoskinson was interviewed by “Finance Magnates” on February 14, 2018, and discusses transaction fees vs. inflation starting at 51:30.
In my opinion, Cardano’s solutions to handle system costs are much better than EOS’s, especially when considering system costs over longer time periods. EOS, however, has a better marketing tool.
C.H. said that a developer on Cardano will be able to subsidize the transaction fees if it wants to do that for its customers. He also said that transaction fees over the long term will become increasingly small, and will fund infrastructure providers, treasury providers, or consensus providers.
C.H. also said that having no transaction fees in a protocol is a good marketing tool, but all token holders (including the ones that are not making transactions) are still paying fees through inflation (e.g., EOS creating more tokens). Inflation debases the currency, which can have a tremendous effect on the value of the token over time. Although he didn’t mention this, I believe DAPP developers on EOS need to buy or rent EOS tokens to allocate the necessary bandwidth and storage in proportion to the needs of their particular DAPP. Depending on the developer and DAPP, those costs would likely be shifted to the user.
In other words, if a protocol doesn’t have transaction fees, the fees are elsewhere in the protocol.
I tried to transcribe most of it Charles’ answer (please let me know if you find errors):
Charles Hoskinson (February 14, 2018). "You have to pay someone to run the system, that’s consensus, that’s governance, that’s development. […] If this is going to be an infrastructure with billions of people, the resources demanded to running such as system are enormous, so, you have to pay them somehow.
"How we pay them in Bitcoin is this through a combination of inflation and transaction fees. It is entirely possible to use a different mechanism and get rid of transaction fees altogether, strengthen up your naming system a bit, and then punt all that to inflation. That is what Dan Larimer has done with EOS and Steem and other projects that he has worked on. He seems to think that somehow transaction fees are a barrier to consumer adoption. […]
"If we build good tech, the fees can be exceedingly low, like less than a penny, in some cases .1 of a cent, but you still need some form of aggregation. Or else what you are going to do is your going to debase everybody’s money. Inflation is a tax on everyone. […] It’s like in Canada, they have free health care, but ‘no’ you pay taxes for that. The money has to come somewhere. So you don’t pay up front, but you pay on the back end and its to society. […]
"So it’s more a question of what is the minimum fee required, in terms of resources, and how do we make that smaller and smaller and smaller, so that’s a question of engineering and research, and then it’s a question of who should pay. And it’s entirely possible to build systems with subsidies. So, for example, the base layer of ADA can connect to the control layer, which is then used for a payment system and that is a subsidized smart contract by the business that wants to acquire customers and they cover all that.
"But to say that the protocol itself should be completely free to run and that somehow that is possible, whether you are IOTA or EOS, I think that comes from marketing and a misunderstanding of economics themselves. They are still going to need network traffic, you’re still going to need transaction processing, that’s CPU cycles, and you’re still going to need hard disk storage. That is energy and materials. The question is, who is paying for that? […]
"So what we are trying to do with Cardano is have a very conservative economic conversation and gradually open up. So first off you create domains were people can introduce subsidies, and the second you see how people are using the system and then you adjust the fee accordingly. So those fees are just arbitrary, you just put them in for DDOS control. […] We’re not even currently collecting them yet. […] Those fees are burnt […]
"So are we going to just have indefinite, persistent inflation to fill our treasury like the U.S. government does and just print money out of thin air, or we going to have a sum zero game where you pay something and that gets reallocated and you don’t have a debasement of the underlying money? Over an arc of time, debasement has a tremendous effect on the value of the dollar. If you look at my grandfather, his first house was $5,000, his first car was $300. […] The reason why he he was able to do that was the dollar was worth far more. If you have persistent inflation over a long arc, the buying power of your asset goes down, and that is just an economic fact.[…]
"We can alter the fees down. I don’t think they are a huge issue. We probably will reduce them by an order of magnitude. But they are there for DDOS control in the beginning, but then they are there for reallocation of funds to the infrastructure providers, whether that be the treasury providers, or it be the consensus providers, or any other role in the system that we desire to compensate that we think needs to be compensated as we grow to a global scale system.
"So I think it’s a more complicated discussion of whether it should be free or not free. [C.H. references minimum payments for credit card transactions and how credit card transactions fees are paid for by the consumer] The consumers are paying that fee; they are baking it into their price scheme, and they are creating purchasing minimums. For instance, if you buy a $5 thing you have to go buy some more just to complete that transaction.
"So I think it’s an oversimplification of a complex issue and I think it’s a lack of understanding of economics that are driving it. And people are purposely doing this because they want to gain marketing brownie points. It’s the same situation with transactions per second.