Let me paraphrase- "You are wrong with 90% success of system because I can find one extreme case"  just kidding

I don't know how this flip (5:5) get "weak consesus" if it's 100% no consensus. It is impossible to shuffle answer better. It's network bug not something that we should base our discussion at.  In most cases it looks like that:



"no consensus" saved this node in 5 times out of 6 (random SS). - I'm taking about 4 out of 4 score (4 good answers and 1 bad flip).

edit: I've checked and this 1 out of 6 that "no consensus" did not saved this node (4 out of 5 score) was in fact his fault. Flip was easy.

I even seen nodes with score 2 good out of 2 (means that he received 3 bad flips and still passed with 100% score)

Weak consensus is when flip has <75% (or 70%) - https://scan.idena.io/flip?flip=bafkreigelddbvctxovmbtznouq4ppmhpjl6xdkdsgpn3rsr7tohg3owwza this one has weak consensus with 2 wrong answers and 4 good.

No consensus flip is a flip with score ~60% like this one https://scan.idena.io/flip?flip=bafkreifo3zjlkgxknk6z6a3u55rbhqfnsec25dohkfx6i5wy5z5q44buz4 5/8 = 62,5% - no consensus
https://scan.idena.io/flip?flip=bafkreialkwcdcrqwcj6xwsxttxycij6hrgsirrl2dwjvawuuz7g5o7shyy - 57%
I don't know why 50% get "weak consensus". In fact there is no way of getting worse score. Report that to Andrew. I'm sure he will address this issue and fix in next node upgrade.




With weak consensus you are getting 0.5 point even with wrong answer. So you can have 3 good flips and 3 "extreme cases" fail all of them (going random you should pass at least 1) and still get 75%.


There is a reason why each flips pass 2 separate test. Consensus and reports.
Reported Flip (f.e wrong keywords) can still be good to check if you are human or not (that's what consensus is for).


I've send you my telegram id.