Is the transaction throughput really infinitely scalable as you claim? What about stress tests?

I can’t find any trustworthy numbers of what kind of TPS nano can handle. Please explain.

The claims are based on the fact that there is no limiting block time, so scalability depends on hardware, more specifically on network performance. Obviously, the only thing that matters for any currency is the performance obtained in a real-time environment. The development team identified bottlenecks early 2018 and have been working hard to put improvements in place. Recent or upcoming advancements such as Vote by Hash, Lazy Bootstrapping, 2-phase face voting, changed PoW algorithm and structured network overlay are all designed to reduce voting traffic on the network, which as a result will increase the number of transactions per second (TPS) the network can handle. Once these features are added, dev team look forward to conducting another stress test.

It’s also important to understand what one really means by TPS. Is it max capacity during a short period or how it can be sustained for a longer period? Also, there is a new measurement possible now and that is CPS (confirmed blocks per second). That makes more sense because those blocks are final and CPS will naturally be lower than TPS during heavy traffic. You also have to be clear what you mean when talking about these numbers because for a transaction to be spent again both a send block and a receive block has to be confirmed. Thus some argue that the real TPS is half of the maximum CPS. At the same time, a send block is enough to change the owner of funds as they can stay “unpocketed” forever. I prefer confirmed blocks because that can’t be misinterpreted.