I recently read the article, Facebook’s Gen Y Nightmare. It is worth reading (or at least skimming). Here are the highlights of the article.
In 8 to 10 years, a fictitious company named Narrative Data will be able to perform character and personality analysis. Narrative Data will analyze peoples social media accounts and other online activity to determine things such as: productivity, effectiveness, personality type, and various other traits. Then the article goes on to mention Narrative Data is able to identify a particular person suffers from acute migraine headaches. The migraines occur a couple times a month. As a result of this finding, that particular person is not considered for a job opening.
Here are my thoughts. First, is something like this even possible? I would think yes. More importantly, how useful is this analysis? My assumption is that there are no perfect people. If you look long enough you can find flaws with anyone. So, in order to avoid the above situation, a person would have to drastically censor his/her online activity or just withdraw completely from online activities. Then Narrative Data would not be able to find any problems. This situation will lead to the same problem employers have today (not enough information). The above situation in the article is an example of too much information. Given the choice between hiring a slightly flawed person with lots of available information or hiring a person who’s flaws have not yet been revealed because of a lack of information, I would choose the slightly flawed person.
If a company like Narrative Data ever does exist, it is almost certain that the opposite type of companies will start to exist. By that, I mean companies will be created with the intent of helping people hide character traits. That messes up all the analysis.
What are your thoughts? Will something like this exist? Would it really be beneficial?
Having worked in transactional due diligence, where the majority of the work consists of online investigation, I have to tell you that it is very *very* difficult for even human analysis to accurately pinpoint specific information about a person online, unless they have a very unique name or a large media profile. I just can’t see algorithms being able to make these kinds of judgment calls without so many false positives that it makes the whole endeavor pointless. Social and personal information aggregation sites are notoriously bad at accurately compiling information about a specific individual, without numerous results that actually pertain to other people, so much so that we basically ignored those sites for anything but the most easily gathered identifying information. Now, with that said, machine learning algorithms could conceivably conduct accurate research; however, they would have to be extensively trained beforehand with identifiers and the results would have to be checked by humans for false information before any decisions are made. Also, speaking from experience, I daresay that a lot of information will fall through the cracks without some sort of human intervention in the research process.
However, should the problem of identification be solved, I can see the situation proposed by the article as being likely. Which is why I’m moving away from qualitative analysis to analytics. 🙂
Wow, thanks for the extensive information.