As for why we don't publish a moderation log of every change we make, there are a number of reasons for that, but one is that I fear it would actually not be good for fostering trust. The reason is that it would provide a lot more scope for a tiny minority of litigious users (not to mention opportunistic self-promoters) to lob accusations that we would then have to either answer or leave unanswered. That's a Hobson's choice, because not answering is not an option—absence of an answer would be taken as "confession! see? they can't even answer!" and allow this concocted sort of scandal to burn unchecked. But answering isn't much better, because it takes a huge amount of time and energy.
For example, the story about HN that appeared at https://news.ycombinator.com/item?id=30066969 was completely false. But just writing up a suitable demonstration of that, as well as answering users' many legitimate questions that arose, literally took me the entire day. Pseudo-scandals like this are cheap to stir up and costly to refute. This makes us vulnerable to a kind of DoS attack, not because we're doing anything that isn't defensible to the community—our most important practice is never, ever to do that, and when we do screw up, to admit the mistake and fix it—but rather because time and energy are so limited. It would be easy to peg us at 100% just dealing with this alone, and that's not even considering the soul-destroying impact on morale.
That would be a bad trade. It would prioritize the demands of a small minority of low-trust users, some of whom (alas!) are determined for whatever reason not to believe a thing we say, while neglecting the interests of the vast majority of the community here, who aren't interested in such drama and just want HN to be a good site to read. It would starve us of resources to make the site better and please the community as a whole.
It's true that I don't know for sure that this would be the outcome of publishing a complete moderation log, but I don't see how we could run the experiment in a low-risk way. If there were a way to take such a step, see whether the fears came true and reverse the decision if so, then I might be up for it. There's an optimistic scenario in which it would make the community happier, enable people to answer their own questions, and free up more time to invest in HN goodness. I just feel like I know enough about internet dynamics not to find that scenario very likely, and it doesn't seem like a reversible thing to try.
That doesn't mean we don't care about transparency—we do! But over the years, we've struck a workable balance in a different way: by being happy to answer questions about specific cases, and about HN in general. We try always to do that, and we always tell the truth when asked (partly because it feels better, and partly it would be dumb not to, since the good faith of the community is literally the only value that HN has). That balance has worked well over the years, although it does mean that I spend a lot of my day answering the same classes of question over and over again. That's probably just baked into the job though.
The reason this approach to transparency works is that the community knows us, we've been fortunate to build up trust with most users over the years, and we try our best never to jeopardize that trust. When we answer questions, most people are willing to provisionally believe we're answering in good faith, subject to making up their own minds based on available evidence. That's all we need for this to work. Meanwhile, low-trust (and/or hostile) users aren't about to trust us even if we did publish everything—they'd just dig new trenches around "well, but you could be falsifying the log".
If anyone has a question that I haven't answered here, take a look at these past explanations: https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que.... If your question still isn't answered, I'd like to hear it and will do my best to answer.