Quote from: xyzzy099 on February 02, 2016, 07:14:45 PM

Quote from: jbreher on February 02, 2016, 07:06:40 PM

Quote from: xyzzy099 on February 02, 2016, 06:51:36 PM

If you are really interested in why 75% is not enough, the definitive answer is found here:

http://organofcorti.blogspot.com/2015/08/bip101-implementation-flaws.html

Because the stated 75% criteria produces a negligible yet discernible chance that we get a false trigger at an actual 67% adoption rate due to variance? Yawn. Troll harder.

Uhm, no. He shows pretty conclusively that

Quote

5. Summary
As it stands, the BIP101 has implementation flaws that could cause BIP101 activation with a significantly sub-supermajority, or (in the presence of fake BIP101 voters) a minority. It is almost certain that if BIP101 is activated, it will be with a sub-supermajority, or even a minority.

It also allows true proportions of fake voters to be sufficiently low that it becomes quite possible for one large mining pool or a couple of smaller ones in collusion contributing fake BIP101 votes to cause premature BIP101 activation.

Emphasis is mine. If you want to present math that disproves OrganofCorti's, feel free. Calling me a troll for pointing out OrganOfCorti's excellent blog post is just childish.

It is not so much the math, but interpreting what it means in real terms. OrganOfCorti's post seems to focus upon mathematics, yours seems to focus upon hysteria. Shall we analyze this together?

The damning part of the analysis is specific to BIP 101, and assumes 'vote spoofing' upon the part of nefarious actors. Such is fine, and important in regards to analyzing the BIP 101 situation. However, you present it as if it is universally-applicable to any 75% proposal. 'Flaw #3' is inapplicable to any situation where the tabulation of 75% is strictly based upon hash power.

So yes, it seems to me that you are trolling.

Even so, there is an admitted latent issue with the order of magnitude in the comments, unaddressed for months.

Quote

Anonymous31 August 2015 at 03:46
> The number of failure attempts before a success occurs in trials of this type is called a geometrically distributed random variable, and can be used to find the probability of some arbitrary true proportion resulting in more than 749 blocks of a sequential 1,000, after that true proportion has been present for some number of blocks.

This is incorrect, as overlapping sequences are extremely correlated. Treating overlapping sequences as independent trials will massively overestimate the chances of success. The expected time for a 0.7 proportion of hashrate to result in a 0.75 proportion of blocks is closer to 300,000.

http://bitco.in/forum/threads/triggering-the-bip101-fork-early-with-less-than-75-miners.13/

Reply

Organ Ofcorti31 August 2015 at 16:18
Yes and I feel a bit silly about missing that! I realised it after a redditor commented: https://www.reddit.com/r/Bitcoin/comments/3ilwq1/bip101_implementation_flaws/cuhy71q

I'll be posting an update after I get the weekly stats out. I haven't had time to figure out an analytical approach, but I'll generate some nice plots based on simulations.

Or more importantly, the amount of time for variance to result in a false trigger is important. The other analysis in the quoted comment above puts the chance as "TL;DR: At anything less than 70% of steady hashrate, triggering a fork would take at least 6 years, and gets exponentially less likely as miner share decreases."

Even the bulk of Core devs are seeming to claim that it will be necessary in the not-too-distant future to increase the block size. Just not now. For some unstated reason. If variance results in a trigger after Core would have already increased the block size anyhow, then the trigger is a non-event. No chain fork results. So what?