The EOS blockchain may not be able to live up to all that it has promised investors, according to a recent study.
A press release for the study states that a benchmark test of the EOS blockchain was "publicly requested" by a number of companies participating in the blockchain industry. The study was completed by blockchain testing platform Whiteblock and led by Brent Xu, project lead at ConsenSys. However, the press release states this was "the first independent benchmark testing of the EOS network."
In July of this year, hedge fund billionaire Michael Novogratz, whose venture capital firm has invested $325 million in EOS creator Block.one, caused a stir when he announced, without evidence, that EOS was "already doing 5,000 transactions a second. It should be doing 50,000 transactions per second in a few months."
In contrast, the new study suggests that not only does the EOS platform not live up to the high standards set forth by Novogratz (who does not speak for EOS as a whole), but that it is actually much slower than previously thought.
The study claims:
"[T]he transaction throughput in the system does not exceed 250 TPS even in optimal settings with 0 ms of latency and 0% packet loss. During tests with real world conditions of 50 ms of round-trip latency and 0.01% packet loss, performance dropped below 50 TPS putting the system in close proximity to the performance that exists in Ethereum."
The study suggests that EOS is maximizing its transaction throughput by bypassing difficult transactions that would cause network congestion and only completing transactions that are easy to process. This method essentially leaves unprocessed transactions in a heap and does not alleviate congestion, but rather ignores it.
In addition to throughput, the authors call into question the transparency of the EOS network because there is no mechanism or protocol in place that validates the EOS software the block producers are using. There are also issues concerning the governance of the EOS platform, which uses a voting mechanism that relies on social consensus rather than "algorithmic [Byzantine Fault Tolerant] consensus."
The study even goes as far to say that because transactions on EOS are not cryptographically validated that the platform cannot even be called a blockchain and that it should be referred to as a "non-autonomous homogeneous distributed database system"
It concludes by saying:
"From this comprehensive analysis of the EOS system, it has become apparent that in order for EOS to be able to successfully act as a foundational base layer protocol, it needs to re-architect a significant portion of its infrastructure. EOS can potentially act as a side chain appended to other more foundationally secure networks, though the system would need to be rebuilt in order to address the problems detailed in this report."
On November 4, EOS block producer EOS Dublin published a short response to the study in a Medium blog post. In it, the block producer expresses annoyance at what it describes as "inaccuracies of both the approach and method of the study…which is just plain sloppy."
More importantly, the response is quick to remind readers that the study was paid for and conducted by ConsenSys, which, of course, is "highly invested in a competing technology."
Neal Roche, CEO of Whiteblock, told ETHNews: "We understand that it's a fair question to ask about the independence of the study if it is partially funded by a competitor like ConsenSys. To address this we have had the paper peer reviewed and also published our test methodology and results." Cognizant that the report had initiated a dialogue, he added, "We are also planning a livestream event with an expert moderator and inviting an EOS expert to the discussion."
Whether the findings in the study are accurate or sloppy, EOS is a relatively new platform that still has a lot of growing to do. EOS Dublin, for one, feels the study was premature, stating, "EOS is 5 months old. We have just started to learn how to crawl. Just watch in awe at the next 5 months."
Update (11/7/2018): This article has been revised to incorporate a response received from Whiteblock after initial publication.