jump to last post 1-3 of 3 discussions (4 posts)

Suggestions for Changes to Hubscore and Hubberscore

  1. janderson99 profile image87
    janderson99posted 3 years ago

    Both of these are dysfunctional and have degraded value at the moment. I would suggest that they be replaced by transparent metrics.

    I suggest that the Hubscore be replaced with the following
    QAP (70%) + Traffic (20%) + Popularity (10%)
    =>QAP contribution is the score 100% rescaled to a value out of 70
    =>Traffic - an index related to HP traffic target - when reaches zero-> idled
    =>Popularity - an index based on internal traffic, likes etc.

    I suggest that on the 'stats' page the following be displayed perhaps by hovering over the 'score'. These are some examples:

    81% Q60 (6:7:6) T10 P6

    65% Q51 (7:6:4) T5 P9

    Hubscore (%)
    Q score out of 70
    Q score components 1-10 for (substance : organization : grammar and mechanics)
    Traffic out of 20
    Popularity out of 10

    I suggest that the hubberscore becomes the simple average of the hubscores for featured articles

    It has been suggested that revealing the QAP data and other information would lead to 'gaming' the system. I don't agree with that. If authors knew their scores they would know exactly what to fix, and they could see the outcomes of their edits. Surely this is what HP wants? HP provides advice on improving hubs - why not be transparent and provide useful feed-back on what to improve. There are various checks in place to stop clicking on own hubs etc.

    It has been suggested that revealing the scores would trigger an avalanche of enquires about the validity of the ratings. HP already has a policy of not providing detailed justification. It also provides the hubscore - why not make it more explicit. If people knew that the issue related to 'organisation' or 'traffic' this feedback would be very helpful and would answer many of their queries.

    Having the traffic score linked to idling - indicated by a zero value - would mean that authors could see what caused a hub the be idled
    0 for Traffic => traffic too low
    >0 for Traffic => idled for other reasons, not traffic
    Q value (out of 70) less than some threshold - 20 say - idled due to poor quality, with detail available as described above.

    I suggest that this would greatly improve the two key objectives: Quality + Traffic.
    I suggest that the revised Hubscore be used as the metric for ranking pages to display on the topic pages - this is a mystery at the moment and needs to be transparent.

    1. 0
      Geovanni Redidoposted 3 years ago in reply to this

      Hi there,

      I see you are making good views and I guess very much applicable. Now Mr. Smith is looking into your idea. I hope this will be successful if implemented.

      I am excited of better results from changes.

  2. Simone Smith profile image91
    Simone Smithposted 3 years ago

    Your suggestion is most appreciated, janderson99! We've been discussing changes to HubScore and HubberScore a lot (internally) over the past couple of days.

    It looks like we're going to hold off on refining HubScore until we're done tweaking the QAP. Makes sense, no? But your insights and tips are great.

  3. moonlake profile image89
    moonlakeposted 3 years ago

    Please no more changes.