Post
Topic
Board Bitcoin Discussion
Re: Why Bitcoin Core Developers won't compromise
by
dinofelis
on 18/05/2017, 04:45:56 UTC
Strictly speaking you are very wrong.
The blocksize was capped at 1MB and what you are referring to is the "soft cap"
and is neither a hardfork or a softfork. You are very wrong that a block made larger
than the soft cap would have been invalidated, in fact, some miners made larger blocks
while other still mined at the lower soft cap levels.

Ah, I thought that it was a mutually respected rule, sorry then.  I didn't know that during the period when the "soft cap" was 500 KB, there were miners making larger blocks.  Why did miners restrict themselves then to 500 KB ?  You are saying that it is pure ignorance/incompetence on their side ?

Quote
You are wrong in many of your statements and your interpretation of words and
events are incorrect. A softfork needs 95% miner consensus, so the other miners
that are "FORCED" are 5% acceptable loss which quickly follows.

No, a soft fork doesn't need 95% miner consensus.  That is an extra mechanism that has been applied to past soft forks in bitcoin, but the dynamics of mining doesn't make that necessary.  As we saw, on litecoin, it was set at 75%.

A soft fork, by definition, is a change in *de facto protocol* that makes potentially correct blocks under the old protocol invalid under the new protocol, but is such that all potentially correct blocks under the new protocol are also correct under the old one.  For instance, black-listing an address is an example of a soft fork.  Rejecting any block that contains a transaction with that address, while before, such a block would be accepted, is typically a soft fork.

A hard fork, by definition, is a change in the *de facto protocol* that makes potentially correct blocks under the new protocol invalid under the old protocol.  There are two variants, here: bilateral (that is, the correct blocks under the old protocol will also be invalid under the new one) or unilateral (the correct blocks under the old protocol are still considered valid by the new protocol).

A soft fork, and an unilateral hard fork, are each-other's time reversals, so any undone soft fork is an unilateral hard fork (for instance, removing an address from a black list is an unilateral hard fork).

As to the needs for miner consensus for it to work dynamically, depending on the kinds of fork, we have 3 situations:

A) soft fork:
if a majority of miners applies a soft fork, then the minority of miners is forced to follow, because all old protocol blocks they will make, will end up being orphaned, as they accept the forked blocks themselves, and hence cannot build an own chain.  A minority of soft forking miners will always mine themselves on "old blocks" and hence orphan themselves out of business.    So with a soft fork, there's always only one chain, and the majority imposes its will on the minority.


B) bilateral hard fork:
Whatever the ratio of miners picking one or the other protocol, two independent chains emerge, all users get "double coins" and if exchanges allow so (list both chains), users can vote with their money ; miner revenue vs difficulty will be such, that the hash rate will adapt to the market valuations of both chains.   Bitcoin has an extra problem here for the minority chain, which is the slow difficulty adjustment, making the minority chain grow very slowly.
In this case, there's no forcing minorities: each has its own chain.  The market votes.

C) unilateral hard fork:
Only when the forkers have majority, they can split off.  If they are minority, they get orphaned because they don't build on each other fast enough.  So if the old protocol is majority, they cannot fork away.  If the old protocol is minority, they can fork away, but the old protocol can remain as an independent fork.  We are then in situation B, but the danger is that if ever the market votes for the old protocol, the new chain will be taken over at a certain point, and be entirely orphaned (even after, say, a month or so) and all transactions in the new coin are gone.  Unilateral forks are hence extremely dangerous.

In all of this, non mining nodes count for zilch.  But in B and C, USERS in the market count a lot (whether they have a full node or not).