# Re: Bitcoin source files attached

100,000 block generating nodes is a good ballpark large-scale size to think about. &nbsp;Propagating a transaction across the whole network twice would consume a total of US$ 0.02 of bandwidth at today's prices. &nbsp;In practice, many would be burning off excess allocated bandwidth or unlimited plans with one of the cheaper backbones. &nbsp;There could be millions of SPV clients. &nbsp;They only matter in how many transactions they generate. &nbsp;If they pay 1 or 2 cents transaction fees, they pay for themselves. &nbsp;I've coded it so you can pay any optional amount of transaction fees you want. &nbsp;When the incentive subsidy eventually tapers off, it may be necessary to put a market-determined transaction fee on your transactions to make sure nodes process them promptly.

To think about what a really huge transaction load would look like, I look at the existing credit card network. &nbsp;I found some more estimates about how many transactions are online purchases. &nbsp;It's about 15 million tx per day for the entire e-commerce load of the Internet worldwide. &nbsp;At 1KB per transaction, that would be 15GB of bandwidth for each block generating node per day, or about two DVD movies worth. &nbsp;Seems do-able even with today's technology.

Important to remember, even if Bitcoin caught on at dot-com rates of growth, it would still take years to become any substantial fraction of all transactions. &nbsp;I believe hardware has already recently become strong enough to handle large scale, but if there's any doubt about that, bandwidth speeds, prices, disk space and computing power will be much greater by the time it's needed.

Satoshi

---

Source file: mmalmi-satoshi.html

External link: https://mmalmi.github.io/satoshi/#email-3
