I haven't had time to read the intervening posts, but here are a couple of followups to posts before I decided to leave this thread yesterday.
Ok so im reading the mini blockchain whitepaper. It appears to be a marginal improvement in some ways but it doesnt address the fundemental difficulty in scaling blockchains. The scalability problem doesn’t have anything to do with the size of the blockchain. It comes from the fact that each actor's transactions must be verified by all other network participants. Its the same math as network effects, except its a negative network effect.
Suppose actors make 1 transaction per minute.
1 actors = 0 verifications because he doesnt need to verify his own transactions.
2 actors = 2 transactions per minute. 2 * 2 actors = 4 transactions verifications. They dont need to verify their own so 4 - 2 = 2.
3 actors = 3 transactions per minute. 3 * 3 = 9 verifications. they dont need to verify their own so 9 - 3 = 6.
4 actors = (4*4)-4=12
5 actors = (5*5)-5=20
ect...
0,2,6,12,20,30,42,56,72
This very quickly gets out of hand when you consider that there is a cost associated with verifying a transaction. even if that cost is infinitesimal.
Actors (users) are not verifying nodes, so you math is slightly incorrect in that respect.
However, your point remains valid that transactions scale O(NxN) by Metcalf or Reed's law. And verifying nodes probably don't scale by N actors.
However, Metcalf's law doesn't tell you the frequencies at which actors do transactions. Visa is currently at about 6000 transactions per second, so this can be verified with a single Intel CPU, so no problem for verifying nodes.
Scaling up to 6 billion people and micro transactions (more frequent transactions) might present a scaling problem. I've looked at Lamport signatures schemes that can verify 100,000+ transactions per second on a single i7 cpu. Since verifying nodes tend to be pools with considerably more resources (amortized over a large amount of hashrate), then a 10 - 100 cpu farm (or a Tilera 64 core cpu) is not unfathomable without destroying decentralization of pools.
Long-term the solution is simple. An ASIC for verification will scale sufficiently to 6 billion and micro transactions.
In short, no problem! Mini-block chain addresses the problem of block chain size and its impact on decentralization of mining. Cryptonite does not include anonymity however.
The
real scaling issue.
He was arguing against a constant % rate of perpetual mining rewards (a.k.a. debasement). He conflated this with the exponential growth of the value and adoption of the coin.
This is not at all what I read. I understood him to mean he favored a fixed constant block reward (ie 1 coin) as opposed to the exponential growth from a percent inflation (ie 1% total coins per x time).
The "1 coin per" scenario keeps rewards flowing while effectively reducing the value of the reward via the greater total coins while the percent ends up ramping up the coin supply on an exponential curve.
Was it not clear?
I see now that he was referring to an exponential growth of the money supply, and not an exponential growth of the rewards relative to the money supply.
I see no divergence problem with an exponential growth of the money supply if you are distributing it to yourselves (i.e. the broad usership). The money supply has been growing exponentially since the beginning of human history.
For one thing exponentially growing money supply is necessary to be consistent with the issuance of debt, because otherwise there is no means of paying the aggregately accumulated exponential growth in money needed due to interest compounding.
If we want to consider divergent outcomes and resilient strategies, note that any strategy might have a divergent outcome, e.g. even a feedback loop might carry with it some game theory for manipulation.
Note the alternative of a fixed nominal rate of emission means asymptotically the money supply stops growing (an exponential decline) which also has pathological divergence, e.g. you have to use transaction fees to fund mining which I had pointed out
has pathological failure modes.
(so much for your blah, blah)
