eMercs - Electronic Mercenaries

People for hire that are paid to perform activites in the world wide web.

I see them most in regular comments who create comments with links to other websites. Today, these posters-for-hire use bots to electronically post these sites. Current counter-measures include simple tests like input words from images, spam filters, and community opinion. Each have their pros and cons, and none eliminates these types of comments.

With the growing industry in product reviews and other public opinions, there are now posters that post reviews that are not their opinions. Jobs for these posters include posting positive reviews to the clients' products, posting negative reviews to the clients' competitors. These are harder to track and easily misleads the general public who still believe most if not all information on the internet is true. Currently, the only countermeasure is to review the reviewers. There are patterns to these posters like uses the same wordings to describe products, uses very general descriptions that can be used for a wide range of products, higher volume of posts, etc. But like a friend's opinion that has been bought, certain opinions are nearly impossible to spot.

Besides reviews, many other social networking sites can be used to redirect traffic. Use of twitter, link farms, private/public sites, facebook, even less known social networks provide many sources where information may appear legitimate.

Anonymity plays a large obstacle in preventing these nuances. There are many ways to make it harder for these mercenaries.

One method is to create a network of trusted contacts, where you can rank their "expertise" (ie trustworthiness) in certain matters and their contacts. This would require a central server so that other sites, vendors, etc. can access to provide you the most relevant information. The premise is that your trusted friends are not going to include these unknown "mercenaries" at least not that role. In other words, although one of the friends might be a fraudulent poster, (s)he will most likely not post fraudulent information under known pretenses since those will typically be under different accounts. Thus using several degrees of separation, a value can be computed to the value of certain information.

For example, I will trust an engineering friend with technical opinions, a doctor with medical opinions, etc. I may rank a close friend with their medical opinion too. If I find a friend who is easily misled by unsubstantiated claims, then their opinion will be downgraded to a lower level. I also trust a friend of a second degree of separation but not as much as close friends, but if that is the only opinion and my friend trusts them then I am more likely to trust them too. Thus the lower the value, the less the information can be trusted. The value can also increase if many of my friends also trust the same person. Also if more friends says the same thing, the more trustworthy the information.

This can also be expanded to people's profiles. If they have worked in the industry, have personal interests or hobbies in related field, had the specific injury or close family/friend who had a similar illness, etc., these can also be used to automate some factors to trust.

Trust of Information = (some constantA) * (how much I trust personA) * (personA trust in information) + (some constantB) [ (how much I trust personA's friends) * (how much personA trusts their friends) ] + [[repeat for all friends]] + ... + (some constantZ) * (unknown source)

Of course, no information can be fully trusted. This only provides a scale to filter information by. Because many activities can be automated today, unknown sources can in theory override the other values. Thus, constantZ should be constantly be scaled appropriately... and similarly with all other constants.