[crankylinuxuser]

I would think that having more transparency (eg: moderation/removal log) is much more appropriate than "he looks good to me".

I've already seen a few times where content demeaning YC companies mysteriously got disappeared... and then summarily blamed on automated removal. Who's right? No clue. But being able to see that log as it happens would be a significant good faith action.

[Beldin]

This leads to am interesting security question: how to achieve real transparency?

I mean HN could publish a ranking algorithm and claim that that is what they use. But then we still wouldn't know

(1) if that is really what's running in the background,

(2) if that is really receiving user inputs that lead to the observed outcome.

I'm guessing for most people here the programming assignment "create a fake log including these real inputs so that this story is suddenly dropped according to the ranking algorithm" isn't that hard.

I think that in the end, you end up trusting something. (Eg. Is the log fake or real?) IF that is so, you might as well design the system around a predetermined root of trust... here, the moderators.

[crankylinuxuser]

I also think that "real transparency" is something one can approach. Just because complete transparency is nigh impossible doesn't mean the steps towards it is worthless. And there is always a lower layer one can point at that is opaque... right down to the silicon.

[zelphirkalt]

Perhaps both are correct. You can automatically remove "content demeaning YC companies".