|
April 01, 2017, 02:55:17 AM Last edit: April 01, 2017, 03:25:51 AM by franky1 |
|
i think its best people simply learn real consensus and the symbiotic relationship between nodes pools services.
nodes set the rules, pools collate data(transactions) to create blocks in a formation that can be accepted by nodes, users/services see the tx/block details using nodes(directly or indirectly).
doing things like bypassing node consensus is a backdoor exploit, which some think is useful but is actually a open hole which can be exploited. (plans to us this exploit to make it even easier in future to slide new things in unnoticed later(trojan possibility risks))
so nodes have the most important role. they set the acceptable rules pools need to follow and allows services and users to access and view data. its important that nodes remain diverse, decentralised to not have a single point of weakness.
by having a dozen+ node implementation(brands) means if one had a bug it doesnt cause big issues to the entire network (BU 2017 assert bug(negligible network drama))
by having one node implementation(brand) means if it had a bug it does cause big issues to the entire network because it affects so many. (core v0.8 2013 Berkeley/leveldb bug(large network drama))
not only does diverse nodes help reduce bugs affecting everyone. but it also makes it less easy for corporations to change the rules without consent. (blockstream cannot just activate segwit tomorrow because although they deny it. it does require community consensus, even if they did go soft)
pools will not produce blocks that are not acceptable to node consensus, else it would get rejected in about 3 seconds. and they just wasted their time (dynamic block with 1.000250 bytes reject 3 seconds)
devs should have the mindset of offering a option. and letting the combined node/user/pool/services to choose it or not. no consensus bypassing, no mandatory activations without consensus. no threats to kill pools or change things to less secure mechanisms
obviously if there were true bugs(network danger risk level, rather than feature level), then users would update their nodes to the implementation that has fixed the bug.
|