Bitcoin Forum

With all this talk of the block-size, could we not create a new test net and have the system simulated 5000 transactions a second to determine how best to implement an eternal block-size algorithm now?

Things we could test for:

• 1. Does a flexible transaction rate determined block-size cause too much consolidation of power? (only a few people being able to host the entire chain)
• 2. Will the transaction fees be enough once the transaction rate is the same as Visa, MasterCard and PayPal combined?

By building various "testnets" and then simulate millions of clients pushing a huge amount of transactions, we can determine what our best course of action will be.

We must do this, if we want Bitcoin to one day be a $Trillion+ economy it must be able to handle more than VISA, MasterCard and PayPal combined, if the network cannot do this we can never hope to replace them. Why should we replace them? Because we want to be the currency of the internet right? Lets dream big.

If Bitcoin can't scale past 100 transactions a second we are going to be doomed to a competing crypto-currency in the future.

Actually, rather than a testnet, I'm working on doing a Bitcoin network simulator, that works out how fast blocks and transactions propagate, as well as allowing for models of miner behavior.

This sounds like a really good idea! Will you be putting it up on Github at some point, I would love to poke around. :)

I've also been working on a simulator, written in javascript. It uses d3's force directed graphing and setTimeout for event management. As a result the code has become messy trying to compensate for VM's poor consistency and some weird ordering problems, making locks somewhat necessary. This is not a scientific simulator, I'm actually embarrassed to release it. :o

• latency/processing simulation
• simulates block propagation, height changes with color changes (wait a few seconds it'll show)
• peer discovery
• bootstrap nodes

If you want to check it out: http://pastehtml.com/raw/cur2u33ss.html

I've gotten bored with it because of the many code changes that are necessary to get this much more accurate. But it's a cute framework for anyone who wants to hack it.