Why a soft fork is harder for miners than a hard fork

[johnyj]

This is because the miners do not have a choice in the case of a soft fork. They can not run the old client to mine, since those mined blocks will be orphaned by the new clients which supposedly have larger hash power and do not accept old blocks. As a result 100% of the miners have to upgrade, they don't have a choice

In a hard fork however, old miners can still run their small fork and see how that works (of course it will die eventually due to less support, but it is their own choice to run a minority fork), their mined blocks will not be orphaned, so it is not so hard as a soft fork

http://bitcoin.stackexchange.com/questions/30817/what-is-a-soft-fork
Look at the button of this page, it is very clearly explained by Charlie Lee

So I think a soft fork will be much more difficult to get miner approval since it leaves them with no choice

[Lauda]

Are the miners the only ones who matter? Really? In reality it is more like this:

SegWit soft fork

• Clients that do not upgrade are unable to validate 'new format' signatures
• All interactions between new and old clients possible
• Services that do not upgrade function normally

2MB hard fork

• Clients that do not upgrade are completely cut off
• Services that do not upgrade stop functioning
• Interacting between new and old clients impossible

You could argue that the people who get 'cut off' could remain on that chain, but they're at a huge risk of 51% attacks (a singular example).

[johnyj]

Are the miners the only ones who matter? Really? In reality it is more like this:

SegWit soft fork

• Clients that do not upgrade are unable to validate 'new format' signatures
• All interactions between new and old clients possible
• Services that do not upgrade function normally

2MB hard fork

• Clients that do not upgrade are completely cut off
• Services that do not upgrade stop functioning
• Interacting between new and old clients impossible

You could argue that the people who get 'cut off' could remain on that chain, but they're at a huge risk of 51% attacks (a singular example).

I think miners are the one that matters most, my reality is like this:

SegWit soft fork

• Miners that do not upgrade are unable to validate 'new format' signatures, and their blocks mined will be orphaned
• All interactions between new and old clients possible, so a roll back is also possible, e.g. some SW nodes can roll back to Core thus leave a small number of SW nodes to be attacked easily
• Services that do not upgrade function normally, so they don't know what is happening

2MB hard fork

• Miners do not upgrade can have their fork as they wish
• Services that do not upgrade will stop functioning, so they will know it is time to upgrade
• Interacting between new and old clients impossible, so that people clearly understand what they are running

You see, listing part of the facts does not help, since each solution has its merits and weakness. In the end, if miners say no, nothing will be implemented. You can start a new coin without hash power, and see how much that coin will worth

Anyway, to have 100% of miners to switch at the same time is extremely difficult, better let them fork and make the choice by themselves, when some of the minority miners see that their chain is get obsolete, they will be convinced that upgrade is necessary

[Lauda]

I think miners are the one that matters most, my reality is like this:
-snip-

Your reality is wrong. Did you not see the last consensus threshold set for a soft fork? There will be no miners left behind or there might be 1-2 small ones. Attacking SW nodes has not different impact than attacking normal nodes. Services do not care about soft forks.
Hard fork: Useless fork. Services will be damaged in this 'they will know..' due to a time frame in which they do not function. If you really think that cutting off people is a green point that I'm actually going to stop discussing and most likely ignore and of your future input. This is a consensus algorithm, not total war.

You see, listing part of the facts does not help, since each solution has its merits and weakness. In the end, if miners say no, nothing will be implemented. You can start a new coin without hash power, and see how much that coin will worth

It was a summary, not a technical explanation.

[countryfree]

As a result 100% of the miners have to upgrade, they don't have a choice

Let's be clear: that's the goal! We want all miners going in the same direction. I understand it won't be easy to convince everybody, but all miners shall understand that not going the way of the fork (soft or hard) is taking the risk to be left behind.

[johnyj]

As a result 100% of the miners have to upgrade, they don't have a choice

Let's be clear: that's the goal! We want all miners going in the same direction. I understand it won't be easy to convince everybody, but all miners shall understand that not going the way of the fork (soft or hard) is taking the risk to be left behind.

This is a good point, we really need 100% consensus, but if you try to force it to happen, what kind of consequence will that be? The appearance of Bitcoin Classic already shows that forcing people will always generate unintended consequence. This is similar to that you beat your wife to force her to listen to your idea, the result is a divorce

[johnyj]

I think miners are the one that matters most, my reality is like this:
-snip-

Your reality is wrong. Did you not see the last consensus threshold set for a soft fork? There will be no miners left behind or there might be 1-2 small ones. Attacking SW nodes has not different impact than attacking normal nodes. Services do not care about soft forks.
Hard fork: Useless fork. Services will be damaged in this 'they will know..' due to a time frame in which they do not function. If you really think that cutting off people is a green point that I'm actually going to stop discussing and most likely ignore and of your future input. This is a consensus algorithm, not total war.

You see, listing part of the facts does not help, since each solution has its merits and weakness. In the end, if miners say no, nothing will be implemented. You can start a new coin without hash power, and see how much that coin will worth

It was a summary, not a technical explanation.

I just want to be clear about the facts about soft fork

It's funny that Pieter claim that a soft fork is one of the merits of the SW solution. I highly doubt that he does not know that a soft fork will force all the miners to upgrade at the same time. If he knows about this fact but intentionally hide it, trying to fool the miners, then this action is even more questionable

So SW will not do 4MB, only 1.7MB; soft fork will not allow miners run old version, they must upgrade, that's two false claims. I can't think of other reason than marketing when people are making false claims

[Bitcoinpro]

Are the miners the only ones who matter? Really? In reality it is more like this:

SegWit soft fork

• Clients that do not upgrade are unable to validate 'new format' signatures
• All interactions between new and old clients possible
• Services that do not upgrade function normally

2MB hard fork

• Clients that do not upgrade are completely cut off
• Services that do not upgrade stop functioning
• Interacting between new and old clients impossible

You could argue that the people who get 'cut off' could remain on that chain, but they're at a huge risk of 51% attacks (a singular example).

I think miners are the one that matters most, my reality is like this:

SegWit soft fork

• Miners that do not upgrade are unable to validate 'new format' signatures, and their blocks mined will be orphaned
• All interactions between new and old clients possible, so a roll back is also possible, e.g. some SW nodes can roll back to Core thus leave a small number of SW nodes to be attacked easily
• Services that do not upgrade function normally, so they don't know what is happening

2MB hard fork

• Miners do not upgrade can have their fork as they wish
• Services that do not upgrade will stop functioning, so they will know it is time to upgrade
• Interacting between new and old clients impossible, so that people clearly understand what they are running

You see, listing part of the facts does not help, since each solution has its merits and weakness. In the end, if miners say no, nothing will be implemented. You can start a new coin without hash power, and see how much that coin will worth

Anyway, to have 100% of miners to switch at the same time is extremely difficult, better let them fork and make the choice by themselves, when some of the minority miners see that their chain is get obsolete, they will be convinced that upgrade is necessary

that pretty much sums it up

[Lauda]

It's funny that Pieter claim that a soft fork is one of the merits of the SW solution. I highly doubt that he does not know that a soft fork will force all the miners to upgrade at the same time. If he knows about this fact but intentionally hide it, trying to fool the miners, then this action is even more questionable

If you have a high consensus threshold then the soft fork does not handle miners much differently than a hard fork. Essentially if 5% don't upgrade, with a soft fork in this case they would know instantly because of orphaned blocks. In the case of a hard fork they could mine worthless coins for a while without realizing that the network forked.

So SW will not do 4MB, only 1.7MB; soft fork will not allow miners run old version, they must upgrade, that's two false claims. I can't think of other reason than marketing when people are making false claims

It can do 4 MB in theory. In real world usage is it going to be around ~1.6MB and will grow upwards (there were calculations recently re-done by aj on the mailing list).