Hacker News Reader

Comment by dionidium | Hacker News Reader

It's not just about discussion. It's a general stance positioning the perfect against the pretty good. Here's just one example: questions often get closed as duplicate after good answers start rolling in. Sometimes, the new answers are better than the old ones. The mods argue that the users should move their answers over to the old question. Or that they should improve the old answers. It's a lot of should, should, should.

But why are we talking about shoulds that will never get done when a really good answer is already here? You've got users actively providing good content and your response is to reprimand them and tell them that content should be over here instead? You know, we have these things called links on the world wide web. The internet isn't running out of them.

Sometimes it feels like the mods are building a really tidy card catalog at the expense of the books in the collection.

View on HN

While I'm ranting, I'll just note that the position that there's a canonical way to ask every question and a that there exists a single acceptable answer is a fairly absurd proposition that, once accepted, leads to all kinds of equally absurd behavior.

This is by far the worst problem with Stack Overflow. The mods seem to leap at the chance to mark questions that are similar, but not identical, as duplicates.

There's one easy fix for this, and several other problems with SO: Don't close duplicate questions. Just cross-reference them.

Reasons:

  - Gets rid of the "This was closed because of trigger-happy
    moderators failing to distinguish between 'similar' and
    'identical'" problem by making that situation impossible.

  - Improves answer discovery for users: If Google gets you
    to almost, but not quite, the right question, 
    cross-referencing gets you the rest of the way.

  - Improves answer discovery for users: Maybe the best answer
    today isn't the same as the best answer 5 years ago.

  - Improves gamification: Rewards are the lifeblood of
    a site like Stack Overflow.  A way of dealing with
    duplicate questions that rewards users for 
    cross-referencing them is infinitely more in line with
    SE's basic idea than one that punishes them with
    public shaming. 

  - Reduces opportunity for petty bureaucrats to take root.

  - Represents a more thoughtful, measured way of thinking
    about software engineering.  Many questions really can
    have more than one correct answer, or the correct answer
    can depend on subtle details.  Which answer gets the
    green check mark may have as much to do with the situation
    of the asker than the quality of the answer.

Absolutely right. And this is also the way every other bug tracker on the planet works.

(Yes, I think SO looks like a bug tracker used as a discussion forum. They're certainly very close, including hostile project owners, WONTFIX's, etc.)

Interesting observation. Recall that Joel Spolsky got his start with FogBugz, a bug tracker. Think that might have informed the design?

That's exactly what closing duplicates does: cross references it to the original. If someone wants to post new information, they can do it on the original.

Yes, they can. But they won't. (For various reasons, but primarily because it won't get any attention. Nobody likes yelling into the void.)

An alternative to shaming is that you could just accept this great new content that a volunteer is giving you.

There is a badge for exactly that - Necromancer ( http://stackoverflow.com/help/badges/17/necromancer ): Answered a question more than 60 days later with score of 5 or more.

Whenever a question gets a new post it gets bumped up to the top of the activity again - the same amount of possible attention as when an hour old question gets a new answer too.

Yes. I shouldn't have said nobody. I actually like answering old questions, for example. I've earned that badge 10 times :)

What it really comes down to, though, is that when somebody asks a new question, I want to answer their new question, with all of its subtleties, in response to its particular phrasing, in its context.

It depends on the question. If it previously didn't have an answer or only had one, it goes to the top. If it had 5 OK answers and you add a 6th really awesome one, the ranking doesn't change much - expect it to be totally ignored.

That is not always the case. Consider http://programmers.stackexchange.com/a/263542/ in which an answer was posted three years after the question was asked and two years since the last visible answer was posted ( http://programmers.stackexchange.com/posts/65281/timeline ) and is now the second highest answer in the question. That doesn't sound like 'totally ignored'.

I challenge you to find a similar question on StackOverflow. On some of the other, slower StackExchange sites I can see that it might be different.

It does happen. In fact, I think that a couple of my best (and highest-voted) answers are necromancers. However, I should note that the question you linked has 51464 views. And a lot of the questions I've earned necromancer for were similarly well-viewed. That's not typical. Most old questions get a fraction of those views.

Edit: In fact, I should admit that I usually only put the effort into answering old questions when it's something that has a lot of views. It's an indication that the question is linked somewhere prominent or is ranking well for some search terms. Those questions are worth putting the time in for. Most aren't.

It does cross-reference them. But read the rest of my bullet points. It isn't just about cross-referencing. It's about cross-referencing in a way that encourages a vital community that resonates with SE's gamified nature by keeping the site fun for all players.

Closing with a link to the original is cross-referencing in a way that does just the opposite: It decreases the vitality of the community by taking away opportunities for newcomers to take part in the gamification feature.

Let's frame the issue another way: What's Stack Overflow really for? Is it for people to ask questions and get answers? Or is it just a gamified Wikipedia?

It doesn't take a mod to mark as dupe. Five normal users or one with a gold badge in the tag the question is in suffice. Took me a bit by surprise the first time I closed a question with just my own vote.

The number of times I've come across a question that was marked as a dupe, and the linked to question it was "duping" had nothing to do with my issue pretty much means anytime I see an SO link, I know I'm in for a world of hurt, bad answers, and links to no where.

SO has hurt the industry more than helped, I believe.

It reminds me of the Wikipedia "notability" fight from early last decade. The cost of adding an extra page to Stackoverflow/Wikipedia is beyond trivial, there is no reason to restrict additions except to please these mods who want to build a tidy card catalog.

It's almost like there should be a mechanism when marking a question as a duplicate to automatically move the answers from the new question to the old question... along with a 301 redirect on the removed-as-duplicate URL.

Not sure if that would work since most "duplicate" questions are similar but not identical. Therefore an answer responding to the duplicates' specific use-case would be confusing in the context of the original question. Perhaps include the duplicate question in the answer when it's merged. Hidden behind a link e.g "This answer was posted in response to a duplicate question" and a link to view the content of the duplicate question.

This, and a thousand times this.

If a question gets asked more, maybe having more indexed landing pages for it for search engines will make it easier to find. It seems to be a popular question after all! Why limit yourself to just one answer?

A closed question isn't automatically deleted, so the search engines can still find it. You'll just need to click another link to find the answers.

Theoretically it's better to have all the answers in one place instead of scattered about, so that you can compare and contrast them. In practice it's far from ideal. Often a new question will generate a fresh take on an answer, and the diversity would be welcome.

If there's a subtle difference between the "duplicate" question and the older one, you can edit it to highlight the difference. Unfortunately that's hard to do, and you're unlikely to get the reopen votes even if you do a good job.

Reply on news.ycombinator.com