Best (block nr % 2016) for hard fork activation?

Hi,

Question if you'll allow me. This is not about Gavin's latest hard fork proposal but in general about any hard (or soft) fork.

I was surprised to see a period expressed in human time instead of in block time:

> Blocks with timestamps greater than or equal to the triggering block's

timestamp plus 28 days (60*60*24*28 seconds) shall have the new limits.

But even more so I would expect there to be significant differences in effects on non-updated clients depending on the moment (expressed as block number) of applying the new rules. I see a few options, all relating to the 2016 blocks recalibration window.

1) the first block after difficulty adjustment. 2) the last block before the difficulty adjustment. 3) in the middle 4) n blocks before the adjustment with n the smallest number of blocks calculated such that the adjustment can just manage to do the maximum possible drop in difficulty.

One of the effects I'm thinking of would be in case of an evil contentious 75-25 hard fork. If that activates at 1) or 2) it will take an awful long time for the 25% chain to get to 2016 for the next adjustment all the while having 40 minutes block times. Option 4) sounds a lot better for the conservative chain. The attacking fork clearly has a choice to make it as hard as possible for them.

On the other hand when a non-contentious hard fork is rolled out, one could argue that it's actually best for everyone if the remaining 1% chain doesn't stand a chance of ever reaching 2016 blocks anymore (not even by a decent sized attacker trying to double spend on stragglers). Also causing all alarm bells to go off in the non-updated clients.

Have people thought through all the different scenarios yet?

And would it not make sense to define whatever the best choice is as mandatory for any hard fork proposal? BIP9? (Realising attackers won't necessarily follow BIPs anyway.)

Does something like this also play a role for soft forks?

I do realise that it's quite possible for the first few blocks, mined after the new rules become valid, to still be old style blocks. Thus maybe defeating the whole planning.

On Thu, Jan 28, 2016 at 9:31 PM, Jannes Faber via bitcoin-dev < bitcoin-dev@lists•linuxfoundation•org> wrote:

> Hi,
>
> Question if you'll allow me. This is not about Gavin's latest hard fork
> proposal but in general about any hard (or soft) fork.
>
> I was surprised to see a period expressed in human time instead of in
> block time:
>
> > Blocks with timestamps greater than or equal to the triggering block's
> timestamp plus 28 days (60*60*24*28 seconds) shall have the new limits.
>
>

Block timestamps are in the 80-byte block header, so activation is completely deterministic and can be determined from just the sequence of block headers. There are no edge cases to worry about.

But even more so I would expect there to be significant differences in

> effects on non-updated clients depending on the moment (expressed as block
> number) of applying the new rules. I see a few options, all relating to the
> 2016 blocks recalibration window.
>

It doesn't matter much where in the difficulty period the fork happens; if it happens in the middle, the lower-power fork's difficulty will adjust a little quicker.

Example: (check my math, I'm really good at screwing up at basic arithmetic):

Fork at block%2016: 25% hashpower will take 8 weeks to produce 2016 blocks, difficulty drops by 4.

Fork one-week (halfway) into difficulty period: 25% hashpower will take 4 weeks to adjust, difficulty drops by 5/2 = 2.5 It will then take another 3.2 weeks to get to the next difficult adjustment period and normal 10-minute blocks.

That's an unrealisitic scenario, though-- there will not be 25% of hash power on a minority fork. I wrote about why in a blog post today:

http://gavinandresen.ninja/minority-branches

If you assume a more realistic single-digit-percentage of hash power on the minority fork, then the numbers get silly (e.g. two or three months of an hour or three between blocks before a difficulty adjustment).

On Fri, Jan 29, 2016 at 5:39 PM, Gavin Andresen via bitcoin-dev <bitcoin-dev@lists•linuxfoundation•org> wrote:

> On Thu, Jan 28, 2016 at 9:31 PM, Jannes Faber via bitcoin-dev
> <bitcoin-dev@lists•linuxfoundation•org> wrote:
> It doesn't matter much where in the difficulty period the fork happens; if
> it happens in the middle, the lower-power fork's difficulty will adjust a
> little quicker.

The reason why BIP9 (versionbits) only checks for new activations during difficulty retargettings is a simple optimization to only check 1/2016 of the blocks. I suspect the check itself is not that costly for Bitcoin Core, which has all the block headers in memory anyway, but I don't think we should assume that will be the case for all implementations.

<BIP99 aside comment> As an aside, BIP99 never recommends a 75% mining signaling activation threshold: it recommends 95% for uncontroversial rule changes and no miner signaling at all for controversial hardforks. I still have to update BIP99 with some later changes I commented at Scaling Bitcoin HK like signaling hardfork activation with the "negative int32_t bit" so that old clients are forced to upgrade/decide. We could start deploying better ways to inform users about a hardfork event, but of course those changes cannot be applied to older software that is already deployed (but hopefully they will still notice something is weird is happening if the longest chain that keeps growing is invalid because it contained a block with a negative version in it). But I'm yet to see a single hardfork proposal that follows BIP99's recommendations besides the hardfork proposed in BIP99 itself, which should consist on a manageable list of very simple to deploy fixes like the timewarp fix forward-ported from Freicoin 0.8 for the BIP. I haven't seen much interest in growing that little list of "a few fixes nobody disagrees are bugs or sub-optimal design decisions, plus the changes are easy to implement both separately and as a whole" either. I cannot say I have seen any opposition at all to BIP99 as a hardfork either, but I naively expected people would ask me to implement more things for BIP99 besides https://github.com/bitcoin/bitcoin/compare/0.11...jtimon:hardfork-timewarp-0.11 or even contribute the patches themselves. For all that, I don't consider BIP99 a priority to work on and I plan to complete it at some point later, unless there's a time limit for a BIP to be in the "draft" state or something. If someone else considers completing BIP99 a priority, I'm happy to review and integrate things, though. Thanks again to all the reviewers and contributors to the BIP at this time and I'm sorry that it has been stuck for some time. Maybe the classification/recommendations should have been a BIP without code and the hardfork proposal itself should have been another one and that would have been clearer. I just wanted to have some code on my first BIP (and as said the plan is still to put more code at some point). </BIP99 aside comment>

On Fri, Jan 29, 2016 at 03:31:05AM +0100, Jannes Faber via bitcoin-dev wrote:

> On the other hand when a non-contentious hard fork is rolled out, one could
> argue that it's actually best for everyone if the remaining 1% chain
> doesn't stand a chance of ever reaching 2016 blocks anymore (not even by a
> decent sized attacker trying to double spend on stragglers). Also causing
> all alarm bells to go off in the non-updated clients.
>
> Have people thought through all the different scenarios yet?

I wrote up some of those risks in my "Soft Forks Are Safer Than Hard Forks" post the other week:

https://petertodd.org/2016/soft-forks-are-safer-than-hard-forks

I was writing mainly in terms of technical risks for deployment non-controversial forks; for controversial forks there's many more failure scenarios. In any case, on technical grounds alone it's obvious that hard-forks without very high - 95% or so - activation thresholds are quite dangerous.

In general, it should be remembered that high activation thresholds for hard-forks can always be soft-forked down after the fact. For instance, suppose we initially used 100% support over the past one month of blocks as a hard-fork threshold, but can't get more than 96% support. A soft-fork with the following rule can be implemented:

    If 95% of the past blocks vote yes, voting against the hard-fork is
    not allowed.

As soft-forks can be rolled out quite quickly, implementing this in the event that a hard-fork isn't getting sufficient support won't add much delay to the overall process; as it is a soft-fork, only miners need to adopt it for it to take effect.

For this reason I'd suggest any hard fork use 99%+ activation thresholds, measured over multi-week timespan. Hard-forks should not be controversial for good social/political reasons anyway, so there's little harm in most cases to at worst delaying the fork by two or three months if stragglers won't upgrade (in very rare cases like security issues there may be exceptions; blocksize is certainly not one of those cases).