Quote from: sturle on February 14, 2016, 09:01:40 PM

Quote from: kano on February 14, 2016, 06:34:52 PM

Quote from: sturle on February 14, 2016, 06:21:59 PM

950 of the last 1000 blocks (what they actually test for) is not necessarily produced by 95% of the hashrate, and certainly not 95% of the network (most nodes don't mine). Again, I refer to Organofcorti's blog for the explanation and details.

That's discussing 75% where the XT morons tried to make it lower and screwed up coz they didn't work out the side effect on probabilities when dropping it so low ...

Do you think 75 is some kind of special number, and with 95% it is impossible for e.g. 94.9% to produce 95% of the blocks for a while?

No, it's just way less likely, that's the point it seems you've missed.
That's the whole point of using 95% vs 75%
Poisson distribution statistics aren't linear ...

Quote from: sturle on February 14, 2016, 09:01:40 PM

Quote from: kano on February 14, 2016, 06:34:52 PM

Edit: see point 4.

See my previous posts. I wrote a summary of point 4 above:

Quote from: sturle on February 14, 2016, 09:01:40 PM

Quote from: sturle on February 14, 2016, 12:15:55 PM

I made no argument against 95%, btw. However 95% of the last 1000 blocks will be triggered by less than 95% (see explanation by Organofcorti). 95% of one difficulty period (2016 blocks, fixed measurement points) is much more likely to be a certain 95%. At 95% of the last 1000 this somewhat premature activation doesn't carry a high risk of a prolonged fork, so it probably doesn't matter as long as there is a cut-off date where the soft fork will fail if not achieved.

No, you've made the clear mistake of taking probabilities and saying something "WILL" happen.
Obviously something can happen, but if you decide a low probability is an issue, then you can never do a fork, since you can never reach 100% certainly that every miner has switched.
You choose a high confidence interval.
75% is low.