Please read the paper before you comment • Hacker News Board

Re: Please read the paper before you comment

#291

Post by derefr » Tue, Aug 25, 2020, 3:30 PM UTC

derefr wrote: post #281

nabla9 wrote: post #218

Earlier quoted context omitted.

I have done several experiments in Reddit. I post a link with a plausible title and credible looking 404 to credible domain with made up url. In a big subreddit you easily get 30 top level comments before first comment says that link does not work or is fake. Discussion goes on and on without pause. People treat the title as a writing prompt. One of the best writing prompts is pseudo-philosophical techno-bro life or…

I feel like a closer approximation of what’s going on is that there are perennial discussion topics that people want to participate in when they see an opportunity (e.g. “battery technology continues to improve, but it’s rare that we see the results trickling down to consumer electronics.”) But just having a forum with “evergreen threads” where people talk continuously about perennial topics, doesn’t seem to work out…

Further thought:

Many people seem to mistake the social purpose of these “post a link and comment on it” sites as actually being purely for discussing the content of the link. It’s possible to use such sites this way, but it’s not incentivized by the UX compared to the alternative discussed above. You usually need stringent community moderation if you want people to focus on the news itself.

A free UX design idea: if you want to build “a discussion forum for news topics in a domain”, then rather than endlessly cloning the Reddit/HN model, try building a hybrid of Reddit with a traditional discussion forum. I.e.,

1. start with a traditional discussion forum (thread-list view with threads sorted by last-posted time);

2. make it so threads aren't “bumped” to the top from regular posts in those threads;

3. add a special “news link” type of post, that users can make, which does bump the thread.

(Bonus ideas: scrape the news-links and so render the news-link's content inline in the thread. Have automatic "threadmark" navigation between the spans of posts delimited by these news-item posts. Make clicking the thread in the thread-list view navigate to the newest news-item post. Require moderators to approve news-items before the bump triggers, to avoid "redundant" bumps, to in turn increase users' faith in bumps translating to a real renewal of discussion.)

I imagine such a UX would work much better as a way to explicitly run continuous long-running discussions of news topics, as refueled by news links. One core benefit is that what would on Reddit be “previous threads” would on this forum be “the same thread”, and so people would (hopefully!) feel much less of a reason to recapitulate the exact same posts.

Re: Please read the paper before you comment

#292

Post by goto11 » Tue, Aug 25, 2020, 3:36 PM UTC

Tangential I know, but how is the Knuth quote misunderstood? I have only ever seen it interpreted as you should only optimize for performance when you know a section of code is performance critical. Which is typically only the case for small parts of the code (3% according to Knuth).

Re: Please read the paper before you comment

#293

Post by freeflight » Tue, Aug 25, 2020, 3:40 PM UTC

goblin89 wrote: post #265

johnchristopher wrote: post #246

Earlier quoted context omitted.

You hacked the crowd to get them to rate headlines ! Can it be monetized ?

With recent advancements in ML, a comment thread from people who don’t read TFA could be used to back-generate a plausibly looking article.

Or just do it from the other end and have GPT create comments from submissions by GPT, the result is kinda scary in its realism [0]

[0] https://www.reddit.com/r/SubSimulatorGPT2/

Re: Please read the paper before you comment

#294

Post by freeflight » Tue, Aug 25, 2020, 3:47 PM UTC

thinkloop wrote: post #162

skrebbel wrote: post #155

Earlier quoted context omitted.

> Honestly comment sections on the internet are a cesspool A bit offtopic, but when YouTube first added comments, they quickly got into cesspool territory, exactly like you describe. However, these days, I find myself really enjoying YouTube comments quite often. Is this just me? Am I just watching more happy happy joy joy videos these days or did YT comments go from "the internet's awfulest" to "pretty neat" somehow…

It might have to do with there being real identities linked to them now from Google+, which many say was the main purpose of the project; not to mention the improved recommendation engine and upvotes, etc.

Looking at Facebook I would be willing to say that has literally nothing to do with it.

Facebook insists even more on having "real people" using their real names, often forcing them to vet themselves, yet at times it easily steals the crown in terms of "really bad user interactions" from YouTube comments.

Very likely much more related to the topic under which people are commenting: Non-political feel good fluff does not allow for much friction to be inserted, thus the tone of comments will be much more civilized and friendly.

But anything that goes into controversial, and particularly political, territory is bound to quickly escalate into a complete shit show, on YouTube and Facebook alike.

Which fits neatly into common communication strategies like focusing on similarities rather than on differences.

Re: Please read the paper before you comment

#295

Post by freeflight » Tue, Aug 25, 2020, 3:54 PM UTC

Fnoord wrote: post #275

Scarblac wrote: post #150

Earlier quoted context omitted.

It would be funny if it still let you comment, but it showed your test results besides your comment.

Or how about shadow banning the user on that specific article? Make it appear their comment was posted, only on their logged in account (and maybe IP address). Rest of the world wouldn't see it.

Works very well, but the massive lack of transparency will make it difficult to justify and ultimately creating paranoia and scissors in the heads of people.

I would not want to participate on such a platform forcing me to double check if my contributions were actually published or simply hidden away for some secret reason.

It's these pieces of accountability that go a long way to have users compliant with moderation.

Re: Please read the paper before you comment

#296

Post by IncRnd » Tue, Aug 25, 2020, 4:03 PM UTC

The article's page had a terrible viewing experience. With js turned off, visiting the page shows content split across 12 columns with a single row that took up just a quarter of the screen's real estate. Given all this, and the nondescriptive title, I was not about to turn on js.

Re: Please read the paper before you comment

#297

Post by michaelt » Tue, Aug 25, 2020, 4:14 PM UTC

wruza wrote: post #288

michaelt wrote: post #207

Earlier quoted context omitted.

Historically, societies have given a pretty shitty deal to people who can't vote, and a much better deal to people who can. After all, if you're a politician, you want to get re-elected - so appeasing voters is much more important than appeasing people who can't vote. And appeasing demographics where 90% of people can vote is more important than appeasing demographics where 50% of people can vote. So any type of test…

I'm not saying that you have to be cool and smart to vote. Just ask them how much percent a candidate is going to increase/decrease in their program to be sure that a voter is able to at least read and understand the damn booklet before putting their [V] into the bin. But a problem that you described could be addressed by counting invalid votes as 0.5 instead of 0. That would be fair — you didn't care to inform your…

> But a problem that you described could be addressed by counting invalid votes as 0.5 instead of 0.

Unfortunately that doesn't solve the problem - because spending money in the town with lots of people with a full vote will produce more votes than spending the same amount of money in the town with lots of people with half a vote.

> Just ask them how much percent a candidate is going to increase/decrease in their program

Everyone knows whether they feel happier, healthier, more successful, and whether they've got more money in their pocket today than they had 4 years ago.

Why is a candidate's policy on tariff rates on steel imports from Canada more important than that?

Re: Please read the paper before you comment

#298

Post by troughway » Tue, Aug 25, 2020, 4:14 PM UTC

Cogito wrote: post #286

nabla9 wrote: post #218

Earlier quoted context omitted.

I have done several experiments in Reddit. I post a link with a plausible title and credible looking 404 to credible domain with made up url. In a big subreddit you easily get 30 top level comments before first comment says that link does not work or is fake. Discussion goes on and on without pause. People treat the title as a writing prompt. One of the best writing prompts is pseudo-philosophical techno-bro life or…

Happens here too, as others have pointed out. Just yesterday an article that went in depth on all different kinds of atmospheric water harvesting was filled with comments on how this was 'debunked', despite some of the methods being used in practice (notably fog harvesting). Moreover, the article was discussing emerging changes in the technology and the hurdles they would face (it's almost always energy!) but none of…

HN would need to make a significant number of changes for it to restore its signal ratio, including:

1. Flagging sources of information that are known to use clickbait tactics, stupid popups and generally low effort content. Paywalls can also be labeled as such.

2. Niceties like showing original title next to the updated one, or page title next to the submission title. I want to know what the original title was so I can appropriately shame the individual into trying to clickbait a boring topic onto the front page.

3. Downvotes should need to be justified and reviewed. A “I disagree with this comment because it offends me” is not a valid reason to bury good commentary. Happens too frequently. Remove showdead etc altogether.

4. Limited number of new seats for posting/commenting. Read-only for general public. Remove accounts that are regularly submitting garbage to the front page (happens more frequently than you think).

This community has degraded over the years and it’s easily seen in most high volume commentaries. A lot of it is emotionally charged nonsense with people being “convinced” of things with nothing to show where they obtained such strong convictions.

Dan shouldn’t moderate, he should outright ban, and in the process push the low effort armchairing to Reddit and other sites.

HN is first and foremost a vehicle for YC ventures to post their stuff anyway.

While I can understand people are optimistic that this site is somewhat more “intelligent” or “knowledgeable” about particular subjects compared to other sites, the range of that subject matter is so narrow that in majority of categories you’re more likely to find higher quality, insight and volume of posts elsewhere. Namely Reddit.

Re: Please read the paper before you comment

#299

Post by tjalfi » Tue, Aug 25, 2020, 4:20 PM UTC

cgrealy wrote: post #48

Some years ago, Ars Technica ran a story on guns. https://arstechnica.com/science/2011/04/guns-in-the-home-lot... About halfway through a one-page article was this quote. "If you have read this far, please mention Bananas in your comment below. We're pretty sure 90% of the respondants to this story won't even read it first." Sure enough, there were about 3 pages of comments before someone mentioned bananas. Read the…

This reminds me of a prank I tried in high school.

There was a rumor that one teacher didn’t read submitted essays. I decided to test this by applying the jargon translator from Dilbert’s Desktop Games (https://en.m.wikipedia.org/wiki/Dilbert%27s_Desktop_Games) to my essay. The translator output was painful to read: for example, it replaces “mother” with “female parental unit”. I added a paragraph on the third page asking him to call me if he was still reading the essay. I never received a call but he later said that he had lost the essay so the results were inconclusive.

Re: Please read the paper before you comment

#300

Post by dspillett » Tue, Aug 25, 2020, 4:30 PM UTC

cgrealy wrote: post #48

Some years ago, Ars Technica ran a story on guns. https://arstechnica.com/science/2011/04/guns-in-the-home-lot... About halfway through a one-page article was this quote. "If you have read this far, please mention Bananas in your comment below. We're pretty sure 90% of the respondants to this story won't even read it first." Sure enough, there were about 3 pages of comments before someone mentioned bananas. Read the…

I've done similar with online groups like freecycle when selling or giving away things. Include "to prove you've actually read this post, mention the word elephant in your response" in the advert and have my mail agent delete everything to that address[1] which doesn't include the word.

Massively reduces the number of messages you get from time wasters who just mash "can I have this?" or "is this still available?" to get in first, who will then ghost you because they've taken an extra half second to decide they don't want the item after all[2] so you spend time writing replies that achieve nothing.

Probably means some genuinely interested people get redirected to trash too, but if the item is worth having there will be enough responders to make this not matter. If you see no responses after a time, maybe then check your trash to see if there are any there worth responding to.

[1] those services attract much spam, always have a specific address for each that you can turn off once the junk starts flowing in

[2] as it isn't the right colour, or the right size, or you can't deliver, or will deliver but can't conveniently offer collection, or they completely misread the subject line, ...