Harmony Protocol — Genius or Nonsense?

Harmony Protocol is a project which seems to be gaining a lot of hype within private crypto communities — let’s take a deeper look into the fundamentals of this project:

The following objective is drawn from the 2019 section of the roadmap within Harmony’s whitepaper — (http://simple-rules.com/tech):

“Deploying to 100k nodes with 100k tx/sec and 1s latency”

Whilst this is surely an ambitious goal there are serious obstacles to attaining it; as I will outline bellow.

A cluster of 100k nodes will cost a tremendous amount of money. For reference, Google operate using 150,000 servers globally — let us conservatively estimate that the cost of establishing a single node is $1000 — this puts Harmony Protocol’s hardware costs at $100M — and there is no evidence that they have given consideration as to how they plan to supply this capital.

The paper embedded above states that:

“OmniLedger is the first DL that securely scales out to offer a visa-level throughput and latency of seconds while preserving full decentralization and protecting against a Byzantine adversary.”

If you Google “Visa-level throughput”, you will find: “12th May 2017 — Visa handles on average around 2,000 transactions per second (tps)”.

Harmony Protocol are simultaneously stating that Visa = 2,000 TPS, and that they can somehow achieve 100,000 TPS based on the exact same protocol.

It is stated within the whitepaper that:

“Harmony will also serve as a high volume data marketplace, and optimize its machine learning performance. We follow blockchain-based machine learning marketplaces to build a new decentralized economy based on data.”

Now, imagine the 100,000 proposed nodes / computers holding data; the best and fastest framework currently available in the world from which you can gather data from computers is Apache Spark: their website (https://spark.apache.org/faq.html) states:

“Many organizations run Spark on clusters of thousands of nodes. The largest cluster we know has 8000 of them.”

If there is currently nothing better or faster than Apache Spark, and the largest known install is 8000 machines — how will Harmony Protocol deploy on 100,000 nodes by 2019, and what is going to co-ordinate the data?

If VISA can’t do it, and Google, who have a $130B valuation currently use 150,000 servers: ask yourself — does Harmony Protocol appear to be grounded in reality? Will they be able to deliver their roadmap?