Hacker News Reader

Comment by credo | Hacker News Reader

Ironically, this submission is at #8 even though it has 290 points and was submitted just 4 hours ago

Meanwhile the #2 item was submitted 5 hours and it only has 90 points.

IMO anyone who observes HN reaction to any Microsoft topic (that isn't a virulently anti-Microsoft submission) knows that these topics will get pushed down to the back pages. In some cases, the reason may be flagging, in other cases, it may be the flamewar reason (that pg cites above)

Regardless of the reasoning, it is unfortunate that pg is unwilling to admit that this is a problem (and unfortunately, he does also reflect the thinking of most HNers). Fwiw in terms of disclosure, I was an SDE, then dev-lead at Microsoft in the 90s/00s and dev-manager for a few years later on, but I left the company in 2007 and since then, I've been running the company I founded.

View on HN

Ironically, this submission is at #8 even though it has 290 points and was submitted just 4 hours ago

This is a selfpost, and selfposts are automatically penalized.

I.e. it's an invalid comparison, because the #2 item is a link to a URL, whereas this post isn't. Selfposts require many more votes to keep pace with URL submissions.

Are you speculating that this is an invalid comparison or do you actually know that the only reason (or even ... primary reason) for the push-down was the fact that it was a "Ask HN" post ?

I've been on HN for much longer than your 2-month old account and (based on the ranking of other "Ask HN" and "Show HN" posts I've seen), I suspect that you're speculating and that your assertion is incorrect.

The huge inconsistency between points and ranking happen on a large number of Microsoft-related posts and almost all of them point to external websites.

[update:Reply to LukeShu] Thanks LukeShu, it is good to know how ranking is implemented.

Btw I didn't question whether the absence of URLs reduced the ranking or not. My question was whether the absence of URLs was the only reason (or at least the primary reason) for the push-down of this specific post.

It seems clear that the .4 multiplication rule could not have been the only reason.

As per the .4 multiplication rule, the post with 290 points should have been downgraded to 116 points. Yet, 4 hours after submission, it was at #8. The #2 post only had 90 points and had been submitted 5 hours earlier. So it seems like the push-down was partly caused both by the 0.4 rule and partly by other factors.

Anyway, I appreciate your investigation. In spite of being on HN for the past few years, I never knew that posts without a URL, got hit by this 0.4 rule.

It is an invalid comparison. That isn't to say that the community doesn't punish Microsoft-related posts, but self-posts are penalized by the algorithm. Specifically, the score is multiplied by .4 if it doesn't contain a URL.

HN's ranking algorithm[1] (as of 2010-10-12):

    (= gravity* 1.8 timebase* 120 front-threshold* 1
       nourl-factor* .4 lightweight-factor* .17 gag-factor* .1)

    (def frontpage-rank (s (o scorefn realscore) (o gravity gravity*))
      (* (/ (let base (- (scorefn s) 1)
              (if (> base 0) (expt base .8) base))
            (expt (/ (+ (item-age s) timebase*) 60) gravity))
         (if (no (in s!type 'story 'poll))  .8
             (blank s!url)                  nourl-factor*
             (mem 'bury s!keys)             .001
                                            (* (contro-factor s)
                                               (if (mem 'gag s!keys)
                                                    gag-factor*
                                                   (lightweight s)
                                                    lightweight-factor*
                                                   1)))))

[1]: https://news.ycombinator.com/item?id=1781417

> it is unfortunate that pg is unwilling to admit that this is a problem

I think pg thinks there is a problem, and is almost surely willing to admit it.

I think he just doesn't see a way of solving it that doesn't involve the discussions here on HN becoming flamewars... Which is also a serious problem...

This looks to me very much like a doctor prescribing a drug with unfortunate side effects. You still do it, but you're also keeping an eye out for new drugs that don't cause as many headaches.

Good point. It's an error tradeoff between detecting too many false flamewars and not detecting enough flamewars. It might be kind of interesting to study the effects in terms of bias on the actual reported articles vs the submitted articles.

Reply on news.ycombinator.com