Suggestion: levels of warning based on statistic maths

Post Reply
Chanio
Posts: 1
Joined: Wed May 28, 2008 2:39 am

Suggestion: levels of warning based on statistic maths

Post by Chanio » Wed May 28, 2008 2:39 am

Hello,
I am very interested in seeing WOT improve as to become a standard for the www.

During the last months I have noticed an increase in the number of non dangerous sites that get the warning screen. This is surely going to downgrade WOT's strength of influence in their users. The warning was even, suggesting me to lower the level of warnings! This is bad.

The warning screen should be the last resource of warnings, that justifies the aggresive way of communicating it. There should be lighter levels of warning difficult to overcome before reaching to the warning screen level.
There should be an accurate number of trusted votes to reach to such a status. (let's name it DEFCON 2).

When I vote against a site, I should provide a number of reasons and data for voting negatively with a significant value. That might count high, as well as if my level of participation with WOT is high.
So, negative votes shouldn't be displayed with the warning screen (DEFCON2) until a certain status has been reached. Perhaps, by the number of votes or because the proportion between possitive and negative votes is very evident in a big number of votes (see the math applied on statistics or ratings).

My last negative experience was today when visiting www.cameroid.com . I don't know the site but is recommended by honest people at the web and shouldn't be considered as DEFCON 2 without showing a reason or an evident argument...

Please, post me if I can help defining the levels or if I should explain my point any further.
Best wishes!
Alberto

Sami
Posts: 6987
Joined: Sat Oct 07, 2006 11:43 am

Re: Suggestion: levels of warning based on statistic maths

Post by Sami » Wed May 28, 2008 7:51 am

Our system actually does rather complicated statistical analysis when computing the ratings. You seem to suggest we should also compute the reliability of the rating and use this information to reduce the number of false warnings? In fact, we do that already. The confidence indicator (the little guys next to the rating symbol) tells you how reliable our system considers the rating and the add-on only warns you if the confidence value goes above the preset threshold.

Anyway, have you set your warning level to "Very effective" by any chance? This causes warnings for an unsatisfactory rating already, which undoubtedly gets really annoying after a while. Lowering the warning level to "Normal" should solve the issue.

Post Reply

Who is online

Users browsing this forum: No registered users and 3 guests