Saturday, August 11, 2018

Can everything be faked? Digital editing and manipulated data


     Spoiler alert -- once upon a time (in 1993) a murder mystery movie was released called "Rising Sun". Part of the evidence was a digital video of an area which presumably was the time/space where a murder took place. It turned out that the digital video was edited (but not quite carefully enough) to fake the evidence.
     In 1993 (25 years ago as of the writing of this blog), this was relatively new to ponder. We knew that photos could be doctored/changed/altered but, in this movie, a central point was that digital video could also be changed. (Once changed digitally, film copies could be made if desired.) Twenty-five years ago, this took a lot of time and many computers. Doing such a thing still requires a lot of processing power but that power is now available on many home computers (or even phones or tablets).
     We are seeing this capability ad nauseum within the social media. Within less than 24 hours, a new set of memes, photos, and sometimes videos are released purporting to prove something or call something out. Some are real. Sometimes it is obvious (quotes from Lincoln about the Internet SHOULD be immediately recognized as fake although I have found that a lot of people tend to believe them) but often it is not. Sometime the source is a public satire, or irony, site and "should" be recognized as such by those that read them -- but they often aren't. This happens on all segments of the political spectrum.
     An awful lot (too much) of the time, a forgery is only discovered to have taken place by recreating how the fake was created (such as finding the original photos that have been merged together).
     The foundation problem is that if it CAN be faked, how does one determine the primary reality for this particular universe? For articles and "news" segments, it is possible to research multiple sources, find a general agreement from various sources and determine with a high probability (but not 100%) what the reality is. What does one do about forged supporting documents? It is not easy, but it is possible, to find component parts from various photos, videos, sound recordings, and such to be reasonably sure that something has been compiled together, and manipulated, into a different item -- a "photoshopped" photo, an edited sound recording, and so forth.
      However, we are reaching the point of being able to synthesize (or create) from scratch whatever we want. Computer-Generated Imagery (CGI) has gotten so good that, for films, they sometimes deliberately make it appear less real because people can be annoyed if they think it is "real" when it isn't; they like to know it is CGI. That objection doesn't apply when the person, or group, creating the artifact WANTS to have it be accepted as real.
     As per previous blogs, this doesn't just apply to distorting current things of interest -- it can also be applied to faking history and re-inventing the past. Nor (as per other previous blogs) can we rely on memory to be sure. There are those who do not remember what was said by a politician the previous day.
     So, to attempt to answer the question posed in the subject -- can everything be faked? Probably so. As is true that social change falls behind that of technological changes, our ability to create fake items is overwhelming our ability to prove that it has been done. What methods of verification do you use? What can be done to more easily separate the close-to-true from the false? Is this just a new reality that we need to accept and learn to deal with?

User Interfaces: When and Who should be designing them and why?

     I am striving to move over from blogs to subscription Substack newsletters. If you have interest in my meanderings please feel free to ...