It's easier to write a bad comment than a good one, so I don't think that downvoting bad comments should be more difficult than it is now. HN software already has little nudges that make it easier to gain karma than to lose it:

- You can earn points but never lose them by submitting URLs for discussion.

- You can't lose more than 4 points per post, no matter how many times it is downvoted. There is no upper limit to points gained from a single upvoted comment.

- There is only a 24 hour time window in which comments can be downvoted, but they can still be upvoted even months later.

The expertise problem is harder. Sometimes I see a post that has more misconceptions than sentences, but is grammatically correct and written in a confident tone. Only people who actually understand the subject matter will spot the problems quickly. It's tedious to write introductory level corrections for commonly repeated mistakes. For some of them I have considered keeping a mini-FAQ locally and pasting standard responses to standard misconceptions, but HN also discourages repeating identical posts.

If I downvote a misinformed post the original poster won't know where they went wrong, and people who see a grayed post won't necessarily understand what's wrong with it either. The saving grace of the downvote is that it affects comment sorting order, so at least misconception-heavy posts that are downvoted won't dominate the upper reaches of the discussion page.

What’s the criteria for a “bad comment”? And how do you objectively evaluate it?

I lurk and post once in a while. What I’ve noticed is that rarely is something downvoted for being factually incorrect (relatively speaking). You just have to offer a perspective that doesn’t conform with the hive mind and the downvoting essentially amounts to censorship of opinions.

Most of the comments I downvote the official comment guidelines in the section "In Comments" here: https://news.ycombinator.com/newsguidelines.html

When I see heated back-and-forth discussion chains I'll often find myself downvoting several comments and upvoting none of them, because sarcastic quippy retorts to bad comments are also bad comments.

I'll also downvote comments for factual inaccuracy, particularly if I don't have the time to patiently correct them. This only partially mitigates the damage and doesn't really help the original author to learn. When someone writes a comment with just a single big mistake, and they don't seem to have a history of willfully repeating it, that's when I am most likely to write a lengthy reply with explanation and citations.

Anything which is Reddit level of quality in terms of contributing value. For example, low hanging puns that add no actual substantial information.

I'd appreciate some examples. I'm not saying it doesn't happen, but in my experience downvoted comments tend to have something beyond a counterpoint or difference of opinion; their content is usually irrelevant or unnecessarily snarky or somehow hostile.

I don't really have the time to go digging around for examples, but it's common enough that one of the site rules is to not complain about downvotes because it frequently happens for no reason.

The comment author may think there was no reason but almost every time I see a comment complaining about being downvoted, I see a reason among those I listed above.

It also depends on time hours.

I once made a post that criticized the Anglocentric perspective of an article. — it was initially upvoted, but then became downvoted when UTC+2 went to sleep, and then upvoted again when UTC+2 woke up. — one can take a particular guess as to why.

For me the fuzzy litmus is whether it "promotes knowledge or aids the discussion".

Despite whether I agree with it, I upvote anything that I learn something new from.

Factual incorrectness is often downvoted, but usually only if there is a reply that points it out.

I think downvoting is an anti-pattern because it's a negative interaction, and I think negative interactions shouldn't be encouraged if we value high quality discourse.

If downvoting bad comments (for any definition of bad) was harder or even outright impossible you'd still have an alternative: upvote an other, better comment to have it shadow what you consider a less valuable contribution. And that's a positive interaction.

Whenever I find a comment that I think is on-topic and somewhat constructive, but I still fundamentally disagree with the take, I generally go hunting for a comment that embraces my point of view (or write one myself) and upvote that.

I very rarely downvote comments on this site, and generally only do it for comments that I consider to be completely off topic, flamebait or stupid "meme" references. That being said I do comment quite a bit and I admit that I'm always a bit frustrated when I take the time to write what I consider to be a constructive (if sometimes controversial) comment and I get downvoted reddit-style. It feels disrespectful, almost. I probably care too much about that stuff, I acknowledge that.

How do you promote signal over noise without being able to attenuate (downvote) the noise? Trying to outshout the noise results in FB style cacophony.

The main problem with FB IMO is the community, not the moderation tools. I highly doubt that adding downvotes to FB would vastly improve the average quality of the contributions, actually it may well make it worse.

> "If I downvote a misinformed post the original poster won't know where they went wrong, and people who see a grayed post won't necessarily understand what's wrong with it either."

that's ok, the onus is on the downvoted poster to figure it out for themselves. usually that requires seeing a pattern over the course of a few downvoted posts rather than pinpointing it exactly from a given post.

most downvoted posts break one or more hn rules, but a few are downvoted ideologically, so a little resilience is required. strictly unpopular opinions can often be more tactfully rephrased and at least be accepted for discussion, if not agreement. this is a pretty good life skill to develop generally anyways.

I find it sort-of ironic that at the time I am writing this reply, _your_ comment has been downvoted and I'm not sure why.

Presumably the onus is on you to figure that out, but good luck, since I'm not sure either and usually I can tell.

yes, sometimes the reason is truly inscrutible. that’s where the resilience part comes in. =)

Not everything on HN is user-facing. For example, one comment can only sink to -4 points, but further downvotes on that comment still contribute to the user being muted. I've experienced that myself.

SO and HN are a bit different though. When you go to SO, you want to know what is right. For HN, you want to read something interesting.

“one its major flaws is that people speak authoritatively about things they don't know about”

See: any thread about aeronautics, education, or (historical) warfare.

... and machine learning and medical research and ...

I personally notice this most frequently on the topic of human nutrition. Most days I don't bother to read the comments on ANY nutrition article because they're too depressingly wrong.

... and ... and ... and ...

"Interesting, making downvotes costly seems like a good system."

Say, I like the idea. No doubt people have discussed this to death elsewhere. You can definitely see downvotes used as simple disagreement or for punishing unpopular speech (rather than being wrong, let's say). That's a bummer but is a side effect of modern times.

Let's say that a issuing a downvote resulted in a 15 minute break from posting or voting. You gotta want to do it.

" people speak authoritatively about things they don't know about."

I'm afraid that's baked deeply into the DNA of computer programmers.

While not a perfect solution, perhaps you could tag various conversations by subject and keeping track of which topics a user tends to discuss. If a user mostly talks about a few subjects, have their comments in conversations with those tags be prioritized by the ranking algorithm and deprioritized outside of it. Obviously just because you spend a lot of time talking about something doesn't mean you actually know much about it, but it's a start. Perhaps you could improve further by keeping track of users' karma within a topic (presumably those speaking from ignorance will be less upvoted), but this runs the risk of promoting views that are popular instead of informed.

I think if people want the next level of quality on technical discussion, there needs to be a gatekeeping mechanism which is concerned with technical competency.

I would also expect such a place to have a magnitude or two fewer people.

I've tried that in my own community but it's a real struggle because in my experience newbies drive the conversation much more than pros so in my case gatekeeping was creating a silent community of highly skilled lurkers.

With no gatekeeping, there's lots of activity and pros feel like participating once in a while with deep discussions sparked from seemingly simple questions, but on the other hand a lot of pros feel alienated and leave if the average level of competency is too low.

I've been toying with the idea of pro:newbie ratios but haven't implemented anything yet.

An approach I think is good is for pros to help visitors solve problems, using messages of interest to other pros who can also participate and it can really brainstorm in unforseen ways.

Pros learn by watching other people's problems get addressed, and contribute when it falls under their domain expertise. They interact in a positive way to form the personality of the messageboard. Which should be a reference point for an actual newbie who wants to participate long-term, but it should be expected that most visitors will not return after their problem is solved. The message board becomes a curated/community archive for solutions.

Where the subject matter and the discussion are contained within the board.

That's for a problem-solving message board.

Now HN is a tech news aggregator with comments which make it a board of a different kind, with way more diverse comments to deal with than any one news site.

And very many news sites individually are challenged to adjust to the recognized lower quality of comments the last few years by trying experimental approaches or no longer accepting comments at all.

The news is always elsewhere, only the comments are here and that makes it debatable.

Maybe there's such a thing as a problem-creating message board?

You would need to provide professional indemnity insurance cover. Most policies likely don’t cover advising the whole internet.

HN comments can't be killed by downvotes and the karma loss on the commenter is capped at -4. As forums-with-downvoting go, HN's version is fairly benign.

I think downvoting should cost something. Maybe not a whole point, but something.

If shutting someone else up (and let's not mince words here... that is exactly what you are doing) isn't worth even a token cost to you, maybe you should reconsider whether shutting them up is truly justified in the first place.

They are doing a pretty good keeping the spammers away, considering how easy it is to create a new account.

>making downvotes costly seems like a good system

Habr.com makes both up and downvotes costly, and they have an evolved system checks and balances. The best I´ve seen. They can even fight off Kremlin bots successfully, this is major feat in their region/language of choice.

Maybe downvotes should cost some karma? might be a trigger to think before downvoting