| |
In lugnet.admin.general, Todd Lehman writes:
> You mean a positive-feedback loop producing progressively higher and higher
> values for popular articles? That's what the sqrt() function is for -- to
> dampen by applying a nonlinear but still monotonically increasing function
> to the divisor.
Ah - yep I understood the sqrt part. What I meant is that creating a list of
the most popular messages skews the vote - because the most votes will end up
being put on items on the list, only because people use the list to browse.
From a personal perspective, I would 'perceive' a vote to be more worthy, if it
had been scouted from a newsgroup, as opposed to one that someone had seen
already rated and decided to vote upon.
> > Maybe if votes that were referenced from the Top X page, were weighted
> > differently?
>
> Well, there's really no way to tell where an article is referenced from in
> the sense of where someone first read it. Someone might first read it on
> a "top x" page, or in their newsreader, or from somewhere else. All you
> can know from the referring page is where someone last viewed it -- a
> mechanism which is easily defeated by re-summoning up an article from a
> different click-path and then voting from there.
I grok that it can be easily defeated, but that's not the point :) Someone who
participates in the voting system regularly is more likely to value the system
and therefore not try to skew it. It's not such a bad thing if it can be
defeated - if you've got to go to the trouble of finding a different click-path
then someone obviously feels strongly about the issue in the post!
I don't think too much time should be spent on making it secure - I don't
envisage people who would skew the results by casting votes of equal weighting
would be trying to skew the results.. but would be doing so because that was
the way the system was set up.
Richard
|
|
Message has 1 Reply:
 | | Re: Article scoring
|
| (...) I almost get what you're getting at, but I'm stuck on one part of it. In the early stages of an article's appearance, let's say it's position 19 out of 20 on a top-20 page, and let's say only 7 people have voted +1 on it so far. Should it, in (...) (25 years ago, 4-Mar-00, to lugnet.admin.general)
|
Message is in Reply To:
| | Re: Article scoring
|
| (...) You mean a positive-feedback loop producing progressively higher and higher values for popular articles? That's what the sqrt() function is for -- to dampen by applying a nonlinear but still monotonically increasing function to the divisor. (...) (25 years ago, 4-Mar-00, to lugnet.admin.general)
|
52 Messages in This Thread:
 
  
- Entire Thread on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
This Message and its Replies on One Page:
- Nested:
All | Brief | Compact | Dots
Linear:
All | Brief | Compact
|
|
|
|
