Ask PG: Why are submission titles manipulated without notification?
Some hours ago I added a Hacker News entry for a Microsoft bug, which was shown back then as:
| Ping of Death Reloaded (microsoft.com) 1 point by vog 9 minutes ago | discuss | edit | delete
Some hours later I revisited HN and my entry was shown with a different title: [1] [2]
| Vulnerability in ICMPv6 could allow Denial of Service (microsoft.com) 15 points by vog 7 hours ago | 11 comments
Also, the comments were filled with complaints about that new, crappy title:
https://news.ycombinator.com/item?id=6215368
taspeotis:
I read the linked page and all I could see was a vulnerability
in Windows. [...]
X-Istence:
Could it be added that this is just in Windows, in the title? [...]
scottlinux:
Vulnerability in how Windows does ICMPv6 [...]
wbl:
No, vulnerability in Windows allows Denial of Service.
This isn't a protocol flaw in ICMPv6.
To summarize:
1) Some HN admin is able to edit other peole's submission titles. (maybe okay)
2) There is no transparency about this kind of manipulation. (WTF?)
3) The new title still appears next to my nick name. (WTF?)
Result: I have never entered that crappy title, but receive the complaints!
Is this how HN is meant to work?
Sincerely,
Volker Grabsch
[1] Note that I'm aware that this is just the original title of the article. But that title was crap, which is why I chose a different one in the first place! As far as I can tell, my title was shorter and still more accurate than the article's title.
[2] To prevent more title manipulation, I'd like to notice that I submitted this entry unter the title "Ask PG: Why are submission titles manipulated without notification?"
149 comments
[ 20.0 ms ] story [ 350 ms ] threadIt's worse than that. Since you can't editorialize titles and some authors like to give their articles poetic, meaningless names like "A butterfly in the sky," when the actual article talks about a security exploit in Bitcoin, coded in Go and released by Wikileaks.
I often find myself ignoring interesting articles on HN only to go read them later on reddit with a much more descriptive, editorialized headline.
Right now there's a good example of this problem on the front page. The article about the Android/PRNG security flaw is aptly editorialized as "Google confirms Java and OpenSSL crypto PRNG on Android are broken," but the original article is titled "Some SecureRandom Thoughts," which is meaningless and I would never have clicked it under any circumstances. Thankfully the mods left this one alone.
The point of titles should be to give an informing mini-summary about the topic.
Let me know if you submit to YC, I might join!
It would be nice if every title accurately described what it was about so I could make better decisions about what to read now, what to read later, and what to pass over.
Thank you.
I revised the title for that link several times. My final edit lifted the style of the Ars Technica[1] headline, but specifically added the more technical "Java and OpenSSL crypto PRNG" that I worried would be incomprehensible to the layman...
Though, the mods did remove my trailing "." from text. ;-)
[1] https://news.ycombinator.com/item?id=6215972
I won't go as far as accusing HN for impersonation (if that's the name), but please, if you change original title, at least remove original submitter name.
If I have to hide my real identity here on HN because of such behaviour, this could be just a sign of more serious problem with this so called "moderation".
But it is not really karma related,.and more if people remember your nickname.
I happen to disagree with this policy and I think it's a somewhat overly rigid application of the guidelines. However, this is why it's happening, and it makes some sense from at least one perspective. And, after all, this is entirely pg's call to make.
Which included David Byrne's (slightly misinformed) thoughts on the Aaron Swartz case. As such I titled it something like "David Byrne on Aaron Swartz" - it was changed to "Civil Disobedience" (the title of the blog article).
Although the old title may be have been guilty of "editorialisation" in some sense, the new title gives no indication of why it would be of interest to HN readers.
I really think this policy does a considerable deal of harm.
It would take a change to the way things are presented, but what about having a title that is the same as the original article, and then another line -- a subtitle, if you will -- that (optionally) can contain the poster's rationale for contributing the link.
In this case it could be:
Civil Disobedience
<small>David Byrne on Aaron Swartz</small>
The issue as I see it is that the sum of parts is more than the whole; if less people of the calibre similar to yourself stop turning up then the site as whole starts to lose its value. Personally it's not enough to turn me away; maybe I have no taste :)
Having said that it is a subjective issue and opinions are like farts; everybody's stinks. Damned if you do, damned if you don't.
What the current guidelines actually say is: "Otherwise please use the original title, unless it is misleading or linkbait."
What they SHOULD say is something like "Don't editorialize the titles".
Perhaps we should add an information criterion? E.g.
"Otherwise please use the original title — unless it is linkbait, or misleading (or uninformative) about the content of the article."
It also avoids click baiting as most users probably can tell already by the url in how far a certain medium (...) uses sensationalist headlines.
If somebody thinks that there is a different view or a overlooked aspect in a linked submission that is not reflected by the headline it would probably be best to blog about it and resubmit the link to the blog.
The policy is public, it is consistently applied to any submission, even if you don't follow it the problem is fixed by an admin. I don't see any issue here that demands that kind of public outrage, let alone some kind of weird formal public letters.
I agree that it's a problem when the original title is lousy. Or when the original title is both lousy and longer than 80 chars.
> Result: I have never entered that crappy title, but receive the complaints!
It's frustrating that so many people made those comments - it's pretty obvious that an article from MS is going to be talking about how MS does stuff. If it helps I don't think the complaints are about you, but about MS. Because, you know, there are some companies which provoke that kind of reaction.
I agree that sometimes titles are changed to things that make no sense. That seems to be the rule, and it's consistently applied. That means that people do make comments about the title. This seems to work against PG's desire to avoid the "middlebrow dismissal".
As an aside: I'd love a plugin that helped with the text box. Something that highlighted 80 chars.
No it isn't. To quote one of the complaining comments:
| Sure it links to a Microsoft.com domain, but there are plenty of research projects hosted by Microsoft that are not necessarily on Windows
https://news.ycombinator.com/item?id=6215884
Ever heard of research.microsoft.com?
Yes.
The submission clearly says microsoft.com, not research.microsoft.com
Didn't you know that HN shows the research subdomain in submission titles?
EDIT: Sorry, this reply sounds a bit snarky.
EDIT2: I'm wrong! Thanks to TeMPOraL for pointing out my error about HNSearch and HN.
Oops. I didn't. Is there somewhere a written collection of rules to learn?
It's an annoying bug, it keeps getting raised as a problem, but it's never been dealt with.
'HN to show subdomains' is, I think, number 1 requested feature, but it never was officially done. Is there a whitelist of domains somewhere that get their subdomains shown?
EDIT
Except it doesn't.
https://news.ycombinator.com/item?id=4136016, https://news.ycombinator.com/item?id=6108151, et al.
Make sure to remember that a) lots of HN plugins fix that, and b) hnsearch is NOT HN, even though it looks similar.
The mods change titles. This is mostly to try to normalize posts, so things with super linkbait titles don't always do well, and things with terrible titles don't always do poorly.
You are copying and pasting a link into a box and writing a few words. You are not doing real work. How the post does should be relevant to its content, not the few words you typed into the box.
I don't get this impression. If anything, I've found that original article titles are far more likely to be link-baity than editorialized versions. In addition, It's rare that I've seen a bad article title improved the way you're suggesting.
If anything, I've had the impression that this system is somewhat automated and original article titles are used if they differ significantly from the title submitted (triggered after some threshold).
This is true, but idealistic. "How the post does should be relevant [...]" is not the same as "How the post does is relevant [...]".
Some lousy articles with click baity headlines get upvotes.
And it's particularly frustrating when a neutral informative headline is changed back to the uninformative or clickbaity original headline of the article.
It is simply unacceptable.
A. We all know it happens. B. It's replacing it with the actual title written by the author. It can't be confused with words written by the OP.
Personally no rule is perfect and simple things like just changing it can work. We all know or quickly learn the rule.
It could be worst like reddit style where the titles have collapsed and are consistently emotive/sensational/misleading to get the clicks.
Sidenote: Why would someone down vote this parent comment?
I didn't (I can't downvote) but I would guess it's because it's wrong. See sker's comment above:
Right now there's a good example of this problem on the front page. The article about the Android/PRNG security flaw is aptly editorialized as "Google confirms Java and OpenSSL crypto PRNG on Android are broken," but the original article is titled "Some SecureRandom Thoughts," which is meaningless and I would never have clicked it under any circumstances. Thankfully the mods left this one alone.
And quadhome's response shows that he's the one who edited the title, so edits can be made that could be seen as putting words in the OP's mouth:
I revised the title for that link several times. My final edit lifted the style of the Ars Technica[1] headline, but specifically added the more technical "Java and OpenSSL crypto PRNG" that I worried would be incomprehensible to the layman...
I didn't until now.
Example:
Vulnerability in ICMPv6 could allow Denial of Service (was: Ping of Death Reloaded)
I agree there should be some marker to show the title has been edited, but I don't disagree with this particular edit.
Some can find it ok, many find it annoying, and some others, like me, find it good enough reason to hide, but no amount of moderation will make it disappear.
I don't really care if some modster change my submission title, but if my name is still there and someone googling for it can see it next to something I didn't wrote, then it is a big problem to me.
Like most internet discussion sites it's rather openly a benevolent dictatorship isn't it?
e.g.
| Vulnerability in ICMPv6 could allow Denial of Service (Ping of Death Reloaded) (microsoft.com) 1 point by vog 9 minutes ago | discuss | edit | delete
Shadow bans are so much different!
Source: I have a shadowbanned account.
A few people are unfairly caught out by it, but they are usually told about it.
And the fact you see problem posters continuing to post on their banned accounts shows it works well enough to prevent those people just changing IP and creating new accounts.
Ummm, no. If so, why would they persist in posting using their hellbanned accounts.
In any case, there are numerous threads where people have said that they had no idea they were hellbanned.
ii) You can't tell people if they have no contact information in their profile. You can leave a post in the thread and hope the hellbanned poster sees it.
Really? Because I had no idea what the original title was referencing, whereas the new title gave an accurate summary.
I just logged into an (old) account[0] with no comments or submissions. It has 1 karma and as far as I could tell it was able to upvote no problem.
[0]826 days old. Can't for the life of me remember why I created it, or what (if anything) I did with it but the karma count was at 1 and there were no listed comments or submissions.
http://ycombinator.com/newsguidelines.html
Secondly, I expect a certain quality and professionalism level from HN. The administrators have every right to improve those things through editing of titles, marking some threads as dead, banning trolls, and setting up some basic guidelines for online behavior.
The Internet is such a big place. There are so many news aggregators, sources for the original news, places to discuss the news, why would you frequent/support a site that has a basic policy of quality maintenance that you find oppressive?
Secondly, and more importantly, we are all about using algorithms to automate and take care of peskiness.
I applaud people who do that in nearly all circumstances.
Google has a strict policy of separating ads and organic, "do no evil". PG doesn't adhere to that. He even openly boasts about how he tricked his investors back in the day by having random people show up on meeting days, and encourages this type of behaviour in YCers - all the while expounding he doesn't like a-holes. Couple that with his decidedly evil hellbanning policy, he needs to do some serious soul-searching.
Do those posts get to go higher on less votes? I don't know but some proof would be nice before I agree.
And while its in context re: Google, as a personal bugbear, please reserve the word evil for events in Syria, and the like. Message boards don't come within spitting distance of evil.