EOSIO Throughput and Transactional Statistics of High-Scalability Blockchains
I’m writing this blog to respond to a paper on “Revisiting Transactional Statistics of High-scalability Blockchain”. In this paper they make the erroneous claim that “the current throughput is only 34 TPS for EOSIO”.
This particular paper gathers a lot of data; however, one must question the integrity of a paper that looks to redefine throughput as “only transactions they think have value”. The approach used in the paper attempts to redefine throughput from “what a system can do” to “what a system is doing that I value”. For example there are three potential definitions of “throughput”: how much water can flow through a pipe, actual water flowing through the pipe, and only counting water flowing to the houses you value.
The data drawn is useful for analyzing the demand for blockchain capacity and demonstrates what happens when there is an overabundance of a resource with relatively low demand. A blockchain with high fees (due to poor scalability) will naturally filter out low-value transactions. Scalability literally means the “ability to scale” and doesn’t care about whether that ability is actually utilized or how it is utilized. What we can conclude is that EOSIO has the ability to scale and that ability is independent of how users use it.
An interesting question that the paper didn’t investigate is “what percentage of Ethereum transactions are spam, gambling, or otherwise low-value transactions?” The authors of the paper have previously received funding from the Ethereum Foundation for research work they did to improve the security flaws that are inevitable with Ethereum’s gas model. Given their experience with Ethereum it seems that its exclusion from this analysis was intentional.
What we can conclude is that scarcity drives high transaction fees which improves the quality of the transactions that remain. A scalable blockchain technology such as EOSIO can dial in any quality of transaction they desire by setting a price floor and adjusting the resource allocation model.
How the media chooses to report on this paper will reveal whether or not they have integrity to differentiate technological capability and recognize EOSIO as being the most demonstrably scalable or whether they choose to corrupt the use of English language and the meaning of throughput to present the author’s summary as “EOS has a current throughput of 34TPS” or whether they wish to do real benchmarking and report that EOSIO has a sustained throughput of capability of 3000+ Transfers per Second in a global network.
All of that said, even with the authors poor definition of throughput, EOS is processing 2x the transactions throughput as Ethereum is even capable of if we assume no Ethereum transactions are spam or otherwise “valueless”.
Note: The paper also conflaites EOS and EOSIO by using data from an instance of EOSIO, the EOS network, to draw conclusions about the throughput of the software EOSIO. There are many EOSIO blockchains that are being used to transact real value and that have different resource models and fee systems.