Users Online Now:
2,180
(
Who's On?
)
Visitors Today:
1,316,020
Pageviews Today:
1,800,829
Threads Today:
458
Posts Today:
7,667
02:19 PM
Directory
Adv. Search
Topics
Forum
Back to Forum
Back to Thread
REPLY TO THREAD
Subject
Something Just Went BEZERK in the Gulf of Mexico. The US Navy just sunk a French Submarine
User Name
Font color:
Default
Dark Red
Red
Orange
Brown
Yellow
Green
Olive
Cyan
Blue
Dark Blue
Indigo
Violet
Black
Font:
Default
Verdana
Tahoma
Ms Sans Serif
In accordance with industry accepted best practices we ask that users limit their copy / paste of copyrighted material to the relevant portions of the article you wish to discuss and no more than 50% of the source material, provide a link back to the original article and provide your original comments / criticism in your post with the article.
[quote:Anonymous Coward 1069153:MV8xMTEzNTg2XzIwNzQyNTE1X0NENjM0M0U0] [quote:Anonymous Coward 904889] I just made a very important connection between my academic training and what haz happened at GLP recently. My background is in risk assessment, including the development of classification schemes based on various psychological and behavioral dimensions. In the course of my work I have evaluated about two dozen risk assessment schemes for criminal behavior of multiple types, and conducted or participated in numerous qualitative and quantitative risk assessment research projects. What follows is what I think is happening at GLP. It is merely a theory based on my education, training and work experience. First the technical version, then I'll try to connect the dots. To develop a risk assessment "instrument" one must identify underlying dimensions of behavior and outcomes associated with them. Then you figure out how to measure the dimensions and what data are needed. Often one is unable to get enough information for a particular data point so they gather what is referred to as "proxies" - or data that reasonably represent the concept you want to measure. One that comes to mind is an instrument that wanted to get at the ability to develop relationships. So the question asked whether they had lived with someone in a committed relationship for at least two years. Well, for various reasons this was a failure as a proxy for the ability to connect with others and the item was dropped in revisions. Just want you to be aware of the potential for limitations by using proxy measures. So once you have your dimensions, you have to gather a data pool to validate your assessment scheme, the bigger the better. Every time you break your sample into smaller groups, you need to maintain a certain number of participants so that your statistical results are reliable, and robust under further testing. Each sample can be broken into an infinite number of subsamples based on whatever characteristics are most important to the researchers. Therefore, you need a really BIG sample if you want to end up with at least 500 in your smallest subsample (for example, women who become sex offenders - very rare so you need thousands to get enough to study that one small phenomenon). What if GLP has been using the last several year of posts to develop a data pool to be used in that type of research design? What if they developed the profile and applied it each time they've gone down recently? It would explain the erratic banning-unbanning where people find themselves caught up for whatever reasons we cannot identify from our perspective. What if they are letting some people back in on purpose to test their behavior? To see who they can bring back into the fold and who will follow them to other sites? This type of database is perfect for a merger of qualitative research software analyzing language to identify patterns, and an algorithm used in risk assessment, among other applications, that identifies risk based on how you score next to your "nearest neighbor". It was developed in nuclear engineering to identify various risks using feedback from multiple systems by looking for patterns that cluster around certain outcomes (usually a yes/no). I have had lengthy discussions with the guy that adapted it for risk assessment. I even read his dissertation. The algorithms are proprietary so I have not seen them. Even if I did, I'm not a theoretical statistician so it would make little sense to me, but the underlying theory does. It is perfect for use once you can translate the qualitative information (i.e. posts) into hard data points that can then be run in statistical equations predicting likely outcomes based on known behavior of others (e.g. they compare your profile with one who has the particular outcome they are looking for). � It is possible that they are trying out violence risk assessment by changing the site to keep the particular posters they did, and splinter the rest to other forums, where they are possibly keeping an eye on them (at least those with IPs that are constant and/or kept the same name). They want to see what type of attitude, either acting on or discussing in a forum, can predict who is going to have a particular outcome. They've bet the horse on one pool of potential outcomes by keeping a certain profile of posters (i.e. subsample), but other types of forums will have other types of outcomes because certain profiles have been drawn to them. They could the IP visits of all posters who have left and not come back through ISPs (rather than the site hosts, I do not imply that any sites are sharing data with GLP. I have no idea and this is just a theory anyway). Now that GLP has their research pool and have isolated their discussion to each other on the board and not directly, they have better control over the influences on their subjects. Better explanation of the outcomes by controlling exposure of the subjects to anything else that may influence the outcome, including other poster profiles by prohibiting unmonitored communication. That is why they do not want many posters back, we fit some profile other than what they are interested in. But someone else is likely interested in those profiles because if the theory holds, many were developed. More than you can imagine. Someone is likely tracking the other, non-GLP profiles. That's why they had so many threads on GLP asking about age, gun ownership, relationships, etc. They can directly learn or impute characteristics of each type of poster. Once they have enough data in the database, they identified a profile they wanted to keep. That's why they went down so much, they are working the filter/bans based on the profile. They probably also want to see the reactions of people who go back, and have been correlating IPs with ae to cross reference postings/personality types. Sorry, I got a bit ranty there. It goes much deeper than I can explain in something this short, yes short. They've profiled us, filtered us through bans, and are at the very least watching the outcomes of their own posters and manipulating the environment to either achieve an outcome or at the very least monitor what comes of their predictions. [/quote] :bump: finally someone with common SENSE [/quote]
Original Message
My girl friend has a D.E.D link on her laptop from the French Embassy. (She works at the embassy)
Crazy traffic on DED
Pictures (click to insert)
General
Politics
Bananas
People
Potentially Offensive
Emotions
Big Round Smilies
Aliens and Space
Friendship & Love
Textual
Doom
Misc Small Smilies
Religion
Love
Random
View All Categories
|
Next Page >>