Lessons of the Chewbacca Incident | Hacker News Reader

Lessons of the Chewbacca Incident

binarybonsai.com

86 pointsjulian3730 comments

Wouldn't this have something to do with people opening HN items up in tabs and reading + commenting before moving on to the next one?

Yes, it would. "Time on Site" measurements do nothing to detect whether the site has been opened in the background, or whether visitors are actively engaged. General practice seems to be to measure the extent of this bias by comparing the duration of the first pageview to the duration of subsequent pageviews during multi-page visits.

Shouldn't "time on site" implement a pause timer function if blur() ?

It should. I was referring to Google Analytics in particular, which was the source of this data, and GA's calculation doesn't use blur tracking as far as I can tell.

I'm not sure that blur is fired if you switch tabs.

it is

It is indeed. I tested it out by opening this in multiple tabs and switching back and forth:

    <html>
      <head>
        <script type="text/javascript" src="http://ajax.googleapis.com/ajax/libs/jquery/1.3.2/jquery.min.js"></script>
      </head>
    <body>
      <textarea id="sd1"></textarea><br />
      <textarea id="sd2" rows="20"></textarea>
      <script>
        $("#sd1").focus();
        $(function() {
          $("#sd1").blur(function() {
            $("#sd2").val($("#sd2").val() + "\nblur!");
          });
          $("#sd1").focus(function() {
            $("#sd2").val($("#sd2").val() + "\nfocus!");
          });
        });
      </script>
    </body></html>

Many Reddit readers use the same reading technique and their time per page was still low.

I recently deleted my reddit account due to an addiction problem, but while I was on I would open lots of tabs to come back to later, and I don't do that on HN. I'm not sure how similar my experience is.

That's because they can't read... out of 20 submissions on front page there are always at least 10 posts of pictures without text.

this isn't behavior restricted to HN, though

So, with that graph floating in the middle of the columns, I'd expect to read the 4 chunks of text top-left, bottom-left, top-right, bottom-right. Did anyone else find it surprising that it actually reads top-left, top-right, bottom-left, bottom-right?

Also, if I offered you a javascript library that could lay out columns properly, on existing browsers, with two-column width floats, would you use it?

> Also, if I offered you a javascript library that could lay out columns properly, on existing browsers, with two-column width floats, would you use it?

This is relevant to my interests. I've been working on exactly that: http://www.wired.com/epicenter/2010/08/meet-treesaver-a-new-...

The left side is cut off and the right side overflows if the window isn't large enough to contain the full layout. Reducing the text size has no effect on the layout width, either. Columnized formats are annoying enough when they work but this one alienated me enough that I didn't even bother reading it.

I've been wanting to see proper support of combinations of column-height, column-width and column-span for a long time. I would definitely be a user of a library if it could give me these features.

Looking at http://www.w3.org/TR/css3-multicol/#spanning-columns it seems to me column-span is implemented exactly like the site in question, i.e. wrong from the point of view of existing behaviour (that behaviour being all the printed magazines and newspapers since we worked out how to mix text and images.)

Oh man, maybe that's why I didn't understand a damn thing about the article.

And yes, I totally would.

I don't think the average is an appropriate metric. The author assumes that everyone found the article interesting, but some people were more lazy than others. In reality many people could have just followed a link (perhaps because of a catchy title), found the article uninteresting and never even started reading it (myself included).

Given the length of the article I guess that it's not only the most engaged, but actually the only ones who read it. I don't believe that article can be read in less than 5-10 minutes.

This is the average though so you can assume people read the article from the other sources to, just the majority of them didn't where it would appear the majority of those from hacker news did.

It's not surprising that DF readers are so engaged; if you're a regular DF reader, you implicitly trust Gruber to promote stories that are going to end up interesting. There are vast amounts of new stuff put up on the Internet every day, and Gruber posts only a miniscule fraction of them to his site; it's heavily curated, more carefully groomed than any other referer on the list.

IIRC, Gruber specifically called out this article as an interesting read, which means that if you clicked through to the article, chances are you were doing it with the expectation of reading a good long-form article.

Link directly to the article: http://binarybonsai.com/2010/09/27/chewie-stats/

From the original, the darker bar is the average time spent on the site, while the lighter one is the average number of pages viewed.

Shame on boingboing for using a key-less graph.

Does anyone have a link to the article that people were visiting?

It's the author's first link, here is the HN page for it: http://news.ycombinator.com/item?id=1714989

This is the worst kind of pseudo-science, using a single example to make sweeping generalizations far beyond what the data even shows.

Can anyone give a quick overview of how viewing time on a page is estimated?

wonder if tools like instapaper are brought into the mix. I used to worry when folks would only visit my blog for 10 seconds. Now I just hope they're saving it for an enjoyable read later

i wonder if this implies that hn readers are looking to kill more time. procrastinating on the next big startup? :)

tl;dr

Reply on news.ycombinator.com