Dies to Removal

Eye-catching title, innit?

There an ongoing debate in the better parts of the internet about interesting advances in the field of "stylometrics." What this is is a field of sociology that concentrates on the variations in an individual's behaviors; for the purposes of our discussion here, it is basically the field of research devoted to being able to figure out whether you're lying or not.

The root technologies of this latest siege on the internet's war against fun can be traced back at least to World War II, when both Axis and Allied forces used advanced (for the time) psychological trauma techniques to determine whether or not a captive could and would tell them privileged information without resorting to torture. Back then, it was an arduous process of cross referencing vast tomes of data; today's average computer can quantify that same amount of raw data a million times
over every second without even having to turn on the cooling fan. With this advance comes another boon as well: thanks to the internet itself, the number of permutations has been focused to those people are likely to type out, instead of having to look for quirks written to formalize a note or letter to a friend or family member. This confluence of technological hokum means that programs written for the express purpose of determining whether or not you're hiding something have become eerily accurate in the past few years.
It should be noted that it isn't necessarily the technology itself that is threatening our ability to lie; instead, it's the ease of use, accuracy, and speed with which we receive the information that has grown by leaps and bounds in the past five years. What used to take an intense amount of time to feed into a computer is now as simple as a few other posts, probably from the same subject.
For instance, if you have any experience with speech-to-text (also known as talk-to-text)
software such as Dragon's "Naturally Speaking" line of software, you may be aware that in the past four or five years the technology has gotten dramatically better, and this is because they have integrated stylometric software more directly into the root product; the more the software learns about how you speak, the more prescient it becomes in knowing what you might mean or what you might say next. If you refine that process down to the simpler goal of "does this post read like this other post?" instead, the accuracy likewise goes up, and the need for more sense data drops, until you need only a few other posts by the same author for the computer to tell whether or not the post in question is similar or dissimilar (apparently, as a percentage, from what I have seen). This in turn determines whether or not the person is hiding something. That seems bad for Mafia.

So what we currently have is:
- a system that is free (or at least probably torrent-able) to the public
- that can
within nanoseconds determine key reference points in a poster's text
- that can be compared, again within nanoseconds, against a current post and determine with fair accuracy whether or not that person is posting in a fashion similar or dissimilar to how they usually post... posting in a dissimilar fashion can be determined to be a pretty good scum tell, especially if the question they're answering is "Are you scum?"

[And to give you an idea of the reality of this, I've downloaded a program that runs in a sidebar that stores key source data for users and can give this fair assessment in real time for each individual post, with a snazzy shading system that turns the post more red or blue depending on the similarity/ dissimilarity of the post... if I can get it to work :sweatdrop: ]

And what we will probably have in the next few years, if the optimists are to be believed:
- a system that continues to be free
- that will get more and more accurate the more you use
it
- that will require less input to be accurate
- that will tell you where the key source data inputs differ, allowing you not only to determine if the person is lying but how

So what do you think? Even if this type of program only gives a fairly minor edge to a Mafia player, it seems like technology is slowly making it harder to lie on the internet, or even keep information hidden. Is this going to be the death-knell of forum-based gaming, or is it all blown out of proportion?

[Note- while usually I'd be providing you a glut of links here, I'm not doing it because A) stylometrics is a very, very broad topic with ramifications ranging from law enforcement to internet games, and a simple link isn't going to do the material justice; and B) because I don't want to help people cheat at Mafia]