[tmccrmck]

Perhaps we should revisit the idea of a political detox?

I for one have been noticing the signal-to-noise ratio on HN has been lowering over the past year. Discussions have become less about the content of the article and more about the title or some other related fact. Technical discussions are becoming increasingly derailed by bike-shedding, ideology, and flame wars. Hell, I bet even the average word count per comment is down.

In signal processing, when noise becomes too high we implement a filter. It doesn't have to be too harsh of a filter, perhaps a band-stop filter which passes most frequencies unaltered but attenuates those in a specific range to very low levels.

[throwawayknecht]

The last political detox week occurred at the height of first evidence of social media's influence on the election, and the same week as YC partner Peter Thiel built out his influence on Donald Trump's transition team.

It would be implausibly coincidental for the second attempt to happen at the same time Facebook and Twitter are being directly investigated for their role in electoral propaganda and Thiel is a frontrunner for the PIAB position.

[dang]

That didn't work, because the question of how to define 'political' is itself highly political. We knew that before we tried the detox experiment but boy did we know it afterwards. And it wouldn't be in HN's mandate (like, not at all) to try to limit it to technical stories. Our core audience, even if it feels like a minority sometimes, is readers who are interested in both technical and non-technical things.

> the signal-to-noise ratio on HN has been lowering over the past year

People have been saying that since the beginning. Is it true? Maybe, but there are strong biases (e.g. it always feels like things are getting worse) and it's hard to measure.

[_dps]

I agree that defining "political" is itself partly political. Having said that, and just as my 2 cents, I think the experiment was not run long enough to be conclusive.

People always complain with changes. I think it's reasonable to believe that people who actively want politics on HN would be on the noisier end of the spectrum of complainers. Any attempt to get these conversations off HN was bound, initially, to cause a loud and pointed reaction from one subset of users.

But people also adapt and forget. I think there's a real chance that after, say, 6-8 weeks, HN would have established a "new normal" and people would know that it's not a place to discuss the outrage of the week, just as it's already known it's not a place to discuss the latest celebrity-dancing-baking-show scandals.

I'll reiterate that it's a very tricky problem. But I don't agree that there was conclusive evidence that the political detox "didn't work" -- it was canceled long before a steady state was reached.

[mercer]

While defining 'political' might be too difficult, isn't detecting 'noise' on a comment-level quite a bit easier?

There are still many comments that, even when they often end up 'dead', spawn entire pointless discussions or just clutter up the rest of the discussion in the time that they're up. Quite possibly they affect the general conversation in ways that are difficult to measure.

The thing is that many of these comments appear to me as either immediately and obviously noise, or obviously so after a cursory check of the poster's comment history. In fact, I'm very often surprised that you or your colleague(s) leave a "don't do this here" message, when after clicking through to the commenter it's obvious that they're probably going to keep doing this until they're banned (and trivially create a new account).

I'm wondering if perhaps a stricter zero-tolerance to low-quality posts, or at least repeat offenders, might not be necessary to offset the inevitable growth of 'noisy' accounts.

I'm very impressed with the moderation here, to be clear, so I'm sure that it's more difficult than I think it is, or that I'm missing something. But I guess I'm just how often I come across an active, unbanned account that clearly hasn't been commenting in good faith for quite a while. Considering how little these accounts are 'worth' (as in, if there's any association with the username, it's negative, and karma is low), and how cheaply new ones can be create, a slightly swifter 'banhammer' might not do much harm?