Subject:
|
Re: Lego Purity
|
Newsgroups:
|
lugnet.general
|
Date:
|
Wed, 26 Sep 2007 15:26:31 GMT
|
Viewed:
|
4852 times
|
| |
|
|
In lugnet.general, Tommy Armstrong wrote:
> yea, I understand that, but I cannot be a pure non-purist which is what a
> 0.00 score would indicate if I thought 3 were blasphemous. I mean, by picking
> three as blasphemous, I have contaiminated the purity of my "non-purity" and
> therefore I should have gotten at least a 1.0. Or at least some score above a
> 0.00. The scoring system is obviously contaminated.Because I firmly do not
> believe in clones or modifying clones. I should get some deductions from
> "pure non-purity" for that.
Ahh, ok, I see what you mean. Yeah, this part is difficult. In my scoring
mechanism, I was trying to predict what the most likely deviation from the mean
would be, but I didn't have any data at the time.
Under the covers, the way it scores you is it checks how far away you are from
the average. In theory, if the average vote was a perfect "Blasphemous!" across
the board (1/5), a vote of "Of course!" (5/5) would earn you a difference of 4.
So you could in theory be 128 points away from the average. More likely, the
maximum difference from the average would be 64 (because the average is more
likely to be closer to 3/5 rather than 1/5 or 5/5).
The problem I had was that if I kept the maximum deviation at 64 (or whatever it
is in the specific case), all the scores sort of stayed within a very small
range. Everyone would get roughly in the range of 4-9, and the 0-3 range and
9-10 range was effectively unheard of. So I set up a cap at 30 from the average,
and made that the effective maximum, so that it was more likely for people to
receive other values.
Looks like what I *ought* to be doing (cuz I agree with you that you shouldn't
be a flat-out 0) is making it more of an asymptotic curve towards the extremes,
rather than linear progression, hence keeping the wider range for the normal
person, but making the extremes of 10 and 0 be far less likely. Hm. I'll see
what I can do...
DaveE
|
|
Message has 1 Reply:
 | | Re: Lego Purity
|
| --snip-- (...) Youy could try something like 1/2[1-tanh(k(std-30))] so that it smoothly approaches zero. (...) Tim (17 years ago, 28-Sep-07, to lugnet.general)
|
Message is in Reply To:
| | Re: Lego Purity
|
| (...) yea, I understand that, but I cannot be a pure non-purist which is what a 0.00 score would indicate if I thought 3 were blasphemous. I mean, by picking three as blasphemous, I have contaiminated the purity of my "non-purity" and therefore I (...) (17 years ago, 26-Sep-07, to lugnet.general, FTX)
|
43 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
|
|
|
|
