Hacker News Reader

Comment by kenneth | Hacker News Reader

I think the biggest problem in reverting to original titles is that oftentimes, the original title is not bad, but it only makes sense in the context of the original blog in which it appears. In a social aggregator, it suddenly doesn't make sense anymore.

Consider this title:

    A New Beginning

In the context of the PHP blog, it might indicate a change of direction of the project, a change of leadership, etc. It's a decently sensible title. On a social aggregator like HN, it is much less useful, even if printed next to a small (php.net).

We'd be better off if we let the submitted change it to:

    PHP project changes direction, elects new leader

View on HN

I've come across so many examples where a perfectly valid (imho) title has been changed to something which is basically meaningless with out the context.

Here's a recent one from memory:

https://news.ycombinator.com/item?id=6492781

The originally supplied title was something similar to: "My afternoon with a serial killer". It was changed to the "Center of the Universe" title.

Clicking a link titled "Center of the Universe" one would rationally think they were about to read something by an astronomer regarding the latest hubble deep space image.

I would prefer simply prefixing such simple titles with the missing context, e.x.

    PHP Blog: A New Beginning

To play the devil's advocate, we already have the URL so the title (at least directly on https://news.ycombinator.com, experience on hn-android and other clients will vary):

    A New Beginning (us.php.net)

Is that not good enough?

No. Titles like that are often heavily biased towards regular readers of a blog/forum/interest group, and meaningless to people outside that small circle, even though the article itself may be well worth knowing about.

Github submissions are typically terrible with it. Especially because people use their personal repos as a blog of sorts, which means that it's hard to tell when a submission from Github is an official blog post, a code repo, or an article to read.

For example, https://news.ycombinator.com/item?id=6573101 is on the front page right now.

The title is:

Modeling your App's User Session (github.com)

It's a blog post by Github staff about how they're now tracking user sessions with Rails, cookies, and their DB. Unfortunately, due to the way that the Github site is structured, from the title, one might think that this is a link to a library repo for user session tracking on a mobile app. A small change to the title, like "Github's modeling of user sessions" would convey the meaning of the blog post much better than the title.

That's a good point. I've had a similar problem with submissions from google.com, which may be official google stuff or may be comment from Groups or G+ (though I think this might have been fixed by showing more of the URL, haven't checked lately).

In that case, yes. But what if the domain is somedudesname.com and that is the blog of some company's CEO? It'd be preferable to have a title like "XCompany changes direction, elects new leader".

But he responded to a specific example. If you want to come up with more examples, fine. But don't tell him he's wrong or do a "yes, but".

The example was only there to show that there are cases where editing the title helps. If you want to get to the best answer, don't argue against what someone said, argue against what they should have said.

Being right or wrong about dismissing a bad example doesn't matter. The important thing is to fix the example and then evaluate it.

Then say something like: "True, for that example. But a better example is...".

I guess I have a problem with jumping on someone who was correct.

Thank you for your help. I agree that the information in the URL is not always enough. I am glad it started a discussion and I hope pg reads the rebuttals to what I wrote. (github.com) is a good example. It shows why github was so eager to move gh-pages to github.io

Perhaps a way to get around the ambiguity (at least in case of github) would be to encourage people to use/link to gh-pages (so we see github.io as distinct from github.com) when they are available?

That's still not enough useful info. It tells us that someone has posted a webpage about A New Beginning on php.net, but it doesn't give us enough info to infer that it's a blog post about a new leader taking over PHP. Even worse if it's from some generic URL (blogspot.com, etc).

The biggest problem with this is the unavoidable editorializing. Its as likely to become

   PHP project takes awful turn, elects incompetent new leader.

I would argue that is a lesser problem than nobody who is interested in PHP clicking through to the HN discussion because they don't have any indication that the post is about something they are interested in.

Anyway, re-titling horribly editorialized posts is fine. The problem is that the person/people doing it seem to exercise little to no discretion when doing it.

That's not unavoidable, and when I see editorializing like that it's very obvious that it's axe-grindey editorializing which I can flag and/or comment upon.

Well, doesn't the domain affix (php.net) provide the context?

Not necessarily, for example medium blogs just show 'medium.com' with no relation to the actual blog.

php.net, sure, but what about more obscure domains?

How about this one from the typescript blog? The domain is not enough to provide the context here.

    *Announcing 0.9.1* (blogs.msdn.com)

I think that's very potentially worse. "Changing direction" is far too opinionated.

No it is not worse! Even if the title is totally biased and link-baity and whatever at least I can still guess the approximate topic the article is about. Based on this I can then decide if i am interested in it or not. But with the original title this is often impossible.

For example a few days ago there was a submission with the title "Beer". Since i am interested in beer I clicked on it. But instead of something interesting about beer I got a piece of alcohol fueled drama in the Ruby community which I am totally not interested in.

By leaving the original title I almost always have to click on all articles on the frontpage to find the 4 or 5 articles I am interested in.

Perhaps we should get rid of the complete title to make it as unbiased as possible? Just put the number 1 to 30 on a page and link them to the articles and hide the domain via javascript. This way you prevent a lot of problems... For me almost nothing would change because I already have to click on almost all the links just to figure out what's in the article and if i am interested in it.

Reply on news.ycombinator.com