Ask HN: A solution for "How to recruit?"
LinkedIn fails due to the eBay problem. You can't publish the connection graph or text-based recommendations and maintain any sense of reporting balance. The results are always biased and too many people are attempting to game the system.
My proposal relies on a system similar to page rank as calculated over a set of hierarchically arranged skills. The more highly regarded the reviewer's related skills, the more certain the system is that the reviewer is a real person, and the more similar their opinions to yours, the more weight their opinion carries. Because it's based on skills, searching for contacts is easy; you simply select the requisite skills in order of importance and can determine a minimal value. For instance, you can look for a software engineer that is skilled in PHP, PostgreSQL, and French whose French is at least fluent.
There are a number of analytics approaches I'm considering to limit the ability of individuals to game the system. Further, the system seeks to incorporate a "humanness" factor. That is, each user's identity is verified as much as possible. For instance, by validating a credit card (and checking it's name against the user's name); the charge on the card would be used to send a certified, restricted-delivery letter via USPS with a confirmation code.
The data would use the ZKDB (zero-knowledge database) design to ensure user privacy and security and no data associated with a user and their opinion of another would be stored on our system. This should allow the user to be confident in their assessment of an individual as being private while allowing the system to rate the opinions of "guest" accounts very weakly. Businesses would be similarly identified.
No reporting would be available for an individual until a certain criteria has been met (number of reviews, et cetera). Thus a user can't easily track changes or attempt to deduce who provided what review to the system. This would help alleviate the eBay problem wherein I'm above average, you're above average, and we're all above average.
Sophisticated analytics would be employed to identify mathematical oddities which indicate gaming attempts like disjoint sets and orbits. To be responsive, the analysis would be done in a distributed fashion on a cloud and then stored in an easily-referenced fashion. The incentive for people to add information is basic quid-pro-quo. You all want an easier way to hire people; the cost is simple, you must evaluate coworkers and employees. Because the system hinges upon aggregate opinions of people and since no one person or group can significantly impact the reputation of an individual, there should be no legal concerns with libel. We're merely rating an individual's skills on a relative scale according to the analytic system of our software. It's the same for everyone and thus cannot target an individual.