Hacker News Reader

Comment by dang | Hacker News Reader

How would you guys feel if we tracked outbound clicks on HN? I've always assumed people would hate it, but on the other hand users frequently ask things like how many people vote for a story without clicking on it, and it would gratify curiosity (the name of the game here) to know things like that.

Edit: I suppose it's a dumb question because the answers can only be one-sided.

View on HN

HN is currently one of the few sites where Privacy Badger [0] and Ghostery [1] tell you that there are no trackers.

Don't change it, please.

[0] https://www.eff.org/privacybadger

[1] https://www.ghostery.com/

If you're already running those, why not?

You don't need a 3rd party or anything having remotely to do with advertising to implement this; there is no reason Ghostery would know.

HN is "tracking" you when it routes a comment or an upvote anywhere other than /dev/null. Following a link to a story would be a similar interaction with the application, which also sends you to the article.

Following a link is, by default, an interaction with the target of the link rather than the source. Hacker News is fundamentally an application that gives you a HTML document, which contains instructions for your browser to request further content from linked sources. Actually getting the content from the linked source goes to that source directly, unless HN implements some non-standard tracking system.

The non-standard tracking system is super simple. That's the point of the comment you responded to. It's not a proprietary ad-tracker. It's 30 minutes of hacking on HN's source-code. Instead of linking you to the story directly you would be linked to an HN endpoint and then redirected. That way they would know when/what you clicked. The HN server would be the only machine that knew this information. No 3rd parties. HN would still pass Privacy Badger and Ghostery.

The reddit outbound link rewrote something so that when you mouse-over the link the browser (at least FF) shows the expected link. But then clicking on it you instead are sent back through reddit.com. This is intentionally deceptive.

> This is intentionally deceptive.

I strongly disagree. I would be quite proud of myself for finding a hack to maintain expected functionality in that case. As a user of the web I constantly mouse over links to see where they will take me. It annoys me when that's broken.

"expected functionality", as the other poster said, is telling you accurately where the link goes. If the link goes to an internal tracker, then that's what it should say. What document your browser requests after that is a separate issue.

No it's not. You're being overly technical. A typical user expects to see where they're going to end up. Not the first link in a chain of transparent redirects. Even if you disagree it as at the very least controvertible. I can see how a web developer would have assumed it was expected. And in that case, it wouldn't have been intentionally deceptive. That's all I'm trying to say.

Well put. This is the internet I'd like.

From a tracking perspective, I'm not (particularly) fussed by it (although it's not my favourite if it's not anonymised); but when I can actively see the bounce address before it starts loading the destination page I find it very frustrating (because it's clear to me that the tracking is slowing down my browsing experience).

(Bounces seem like they should be quick, but often seem not to be -- and if I'm sitting there with a bounce page open rather than the article I want to read when I've just run out of reception, it's super lame (in my mind: infinitely lame).)

From a statistics perspective, I often find an article elsewhere and then come here to vote for it (or submit it) -- so it's inaccurate from that perspective.

Ultimately: indifferent.

Is there a reason to do something like this using a bounce address? I mean, that's what at least all the big players seem to be doing, but couldn't you just as well use something like A PING [1] or some sort of Ajax pingback? There's browser compatibility of course, but supporting old IEs shouldn't be critical for something like this.

[1] https://developer.mozilla.org/en-US/docs/Web/HTML/Element/a#...

Holy shit that's a thing

Hmm...

I know a few sites that do it (including Google and DDG); so I guess some people feel it's the best way to do it?

Edit: Ping doesn't look to be in the HTML5 spec anymore, maybe that's got something to do with it?

https://www.w3.org/TR/2014/REC-html5-20141028/single-page.ht...

> Ping doesn't look to be in the HTML5 spec anymore

Interesting, I didn't know that. And now that I look at it, ping is only supported in Chrome and Opera, which explains it even better. I would still expect there to be some other alternatives to bounce addresses, though.

As long as we can still right/control-click on a link and copy it unmodified for someone, I see no problem. That's one of the worst things I hate about Google.

That would be up to the browser's context menu functionality, wouldn't it? I really don't know, but if so, a sub-menu item (or whatever the links are called below the post title) would be sufficient for me.

I hate that too, with a passion. Is it still an issue? I think I installed a browser extension to stop it.

Indeed the behavior is still there. It's especially annoying when you have to share links to products on Steam/Amazon etc. with someone. There's no way (without extensions) to mouse-select the original link underneath the result title because they're usually long and get truncated. I end up going to the sites themselves and search from within them.

Why do they do this? To somehow extract something from the associates of a Google user? I think it borders on scummy.

how many people vote for a story without clicking on it

There is only one vote button, but there are (at least) two distinct things I use it for:

1) I've read the story. I'm upvoting the story because I think people would be interested.

2) I haven't read this particular story, but I've recently read about this topic elsewhere. I've looked thru the comments and the comments are interesting. I'm upvoting because I think people would be interested in the comments.

I often learn far more about a topic from the comments here than from the story itself.

Just chiming in to say that I use the upvote button in both of those ways as well. I don't see a problem with it.

As long as when I copy a link by right clicking the URL I get the actual link, and not some hn.hn/Q3dsa garbage I don't really care.

It matters to me less that you know what I clicked, it matters to me more that your site is useful.

(FWIW, I often read something really interesting and then go "I wonder if this is on HN" and go over here to find the comments, so that would count as voting for a story or reading the comments without actually clicking on it. IDK if that's normal)

To quote you The mandate of the site is "stories that gratify intellectual curiosity", and it seems pretty clear that both the curiosity and gratification here are more voyeuristic than intellectual

Would it gratify intellectual or voyeuristic curiosity?

When I wrote that I did not have website analytics in mind.

I'm glad someone noticed it, though!

It may gratify curiosity but will it make HN a better site?

Genuine question, in case the tone comes across snarky.

Well, sure, that'd be the only reason we'd do it. (There are no current plans to do this, btw, but the idea does come up occasionally.) Making HN better where possible is the thing we care about.

Not sure how big of a demographic I represent, but I actually left Reddit behind because of this and other privacy-unfriendly changes, and found Hacker News when looking for an alternative. So, yeah, I personally would stop using this site, too, if you did implement this kind of user tracking...

What privacy issue does reddit have that you can't turn off with the preferences?

No idea what you can or cannot turn off with preferences. I'm not one to be okay with privacy invasion, just because you can opt out from it.

As for things that drove me away, this Reddit-post sums it up pretty well: https://np.reddit.com/r/privacy/comments/4ll9tc/it_looks_lik...

Make it opt in, because most people wouldn't like it here. Even if only a certain percentage of people make up your sample size, you can still get valid data.

I suspect that the sort of person who would opt in to this would not be a random sample of the population of HN, and so the data would be biased by default.

> you can still get valid data.

That would be stretching the definition of "valid".

You would get statistics about a group of "I don't care I'm being tracked" people, which is surely statistically different than the group of "I do care about being tracked" people. At the very least, they would have significantly different click rates (and involvement) with the subject like "how to disable web site tracking", but I suspect it's deeper than that -- e.g., they will have statistically significant involvement with security related subjects.

Yep. I would count any 'opt-in' analytics as irretrievably tainted even if by some miracle the sample size made it useful.

I'd be OK with this if it were for a limited time - i.e. run it for a month, with a very clear banner at the top indicating what's happening. Make a blog post afterwards with the data (or better yet, have a separate page showing persistent data), shut down the feature after that.

I wouldn't like this to become a regular feature.

No, please and thank you.

Please don't. Let HN be, it works fine, doesn't it?

Third-party solutions : not a fan.

In-house solutions for analytics : makes sense.

If the implementation is lightweight, robust, and not leaky I'm not paranoid enough to care.

For only statistical purposes, it would be enough to track a small sample of the clicks, right?

Wouldn't even the most basic JS stats package do this by default?

Reply on news.ycombinator.com