Questioning the Value of Cosmos
Wondering what the expected value baked into Cosmos ATOMs is today?
“The Cosmos blockchain has the expected value of running 40,000 TPS and beyond”
Comparing Cosmos to a real world system
You have to pretty much compare what Cosmos’ interconnected blockchain can be used for in today’s real world for applications hosted on the Cloud. This is really difficult and unfair to compare with, but it’s a comparison that I’m very familiar with.
Think of Cosmos Validators as a set of computers tied together with some sort of agreed upon relationships or protocol (e.g. TCP/IP, Practical Byzantine Fault Tolerance, etc.) and that it can serve the higher network layers, to serve users a specific purpose. For what purpose? Well, it’s merely passing on messages as “transactions,” changing a state like increasing and decreasing ownership of ATOMs in wallets, recording other blockchain transactions, etc. So, it’s like a global database, and dumbing this down further, it’s like an accounting ledger system that tracks changes of states in everyone’s accounts. The closest I can think of in comparison with today’s Cloud offerings is AWS’s Simple Storage System (or S3 for short). Storing files of ledgers that records transactional changes, simple as that. But it has database-like features, to perform Create, Read, Update and Delete (or CRUD) operations so that you can read what the latest state looks like (e.g. how much ATOM is left in a Cosmos wallet). Technically there’s no Delete in the blockchain immutability construct, but that’s beyond the point of this article.
Let’s start with transaction costs
So, how much does it cost to run a transaction?
At the time of this writing, each block proposer is rewarded about 3.69 ATOMs, and each block is proposed every 6.855 seconds (see StakingRewards.com for ATOM).
And the price of ATOMs per CoinMarketCap.com (it fluctuates a lot!) averages at $4.13 at the time of this initial draft.
How much do transactions cost in the public Cloud today?
Now, let’s compare with AWS S3 pricing. Imagine you’re recording a company’s accounting practices, and you will be Creating (or POST-ing) new batches of files into S3, and later asking to Read a LIST of transactions from time to time, and retrieving it as needed. There’s also Reading (or GET-ing) the ledger to figure out what’s in the ledger.
I’d imagine there’s no more than an average of 1 POST/PUT to record a new transaction and an average of 5 requests to GET the file over its lifetime, and approximately 4 LIST-ings an entire block’s transactions for audits and searching the file locations in an S3 bucket. That’ll be $0.005 x 1 POST + $0.0004 x 5 GETs (and the ledger file is really small, I’ll approximate this to 1MB). There’s also data transfer, or bandwidth, costs that are really small but it adds up, ranging from $0.05 to $0.09 per GB depending on volume. Every 1,000 transactions could be 1GB in total, so it’s another $0.05. This means every 1,000 transactions equals to $0.057
… every 1,000 transactions on AWS costs $0.057
How much does Cosmos transactions cost today?
That’s the wrong question to ask. Cosmos is designed to scale horizontally, hence it could scale to an infinite number of transactions per second (or TPS). The right question is, how many TPS is expected of Cosmos based on the price today?
There IS a finite TPS for Cosmos
Cosmos, or more accurately Cosmos Hub which powered by the ATOM tokens, is one of many possible hubs that’ll be surrounded by validators, incentivized by receiving block proposal rewards. Cosmos runs on top of the Tendermint Core consensus engine, and in theory it could run at 10,000 transactions per block proposed (see Tendermint’s blog on A-to-Z of Blockchain Consensus). But according to Paradigm’s writeup and analysis, with 64 validators, the throughput drops to 4,000 TPS. Hence at best, with 100 validators for Cosmos Hub powered by ATOM tokens, it’s 4,000 TPS. Since one can spin up as many hubs as possible, some powered by ATOMs, others could be powered by a different token, hopefully most will be powered by ATOMs. Hence there’s no real ceiling on the TPS for ATOM-powered Hubs.
Let’s reverse into the TPS equation
Calculating Cost per TPS
Assuming each Cosmos block can retain as much data as the above accounting example, we can reach parity in this calculation. The retrieval of historical data (the GETs and LISTs) is FREE once you run a validator node (with or without stakes). You can query the state of the Cosmos blocks at any point in time. The only true cost is recording the transaction (the POSTs/PUTs). Since we can run at max 4,000 TPS, fitting into 1 new block every 6.855 seconds, there’s 27,420 transactions per block.
Each block costs (in rewards to validators) 3.69 ATOMs. At USD $2 (updated Sep 2019) per ATOM, that’s $7.38 per block, or 27,420 transactions. That means each 1,000 transactions costs $7.38 ÷ 27,420 x 1,000 = USD $0.27. With a single Cosmos Hub today, every 1,000 transactions on Cosmos costs $0.27.
… every 1,000 transactions on Cosmos costs $0.27
and ☝️that’s for One (1) Cosmos Hub
Cosmos is nearly 5 times more expensive to run than AWS
Yes, that’s true… for now. Actually that’s not a bad number. That means speculators are baking in that there could be 5 Cosmos Hubs when it gets to a wider adoption. Who knows, there could be improvements on the chain that helps increase block size (hence increasing the number of transactions possible per second) or speed-up block confirmations, both of which will bring Cosmos to parity with today’s Cloud offerings.
Cosmos is still in its experimental stage, there’s no other blockchains connected it other than itself to stake ATOMs to validators, and withdraw from them, passing ATOMs between wallets, etc. It’s serving it’s own purpose until the inter-blockchain communication (IBC) protocol, similar to UDP or TCP for blockchains, is completed to connect multiple blockchains to the Cosmos Hub and any other future hubs.
Hence to answer the question on TPS, Cosmos Hub has the expected value of running 20,000 TPS and beyond, if the ATOM price increases beyond what it is in September 2019, and if Tendermint Core can scale even further.
DISCLAIMER: The author is the Head of Blockchain Research at KysenPool. This disclaimer informs readers that the views, thoughts, and opinions expressed in the text belong solely to the author, and not necessarily to the author’s employer, organization, committee or other group or individual.