Tuesday, January 5, 2010

Thresholds of Truth (a.k.a. Futurism, Episode 3)

I humbly propose that in the future we will reach the Photographic Truth Threshold.

We are already closing in on it, thanks to Photoshop. The PTT is the point in time at which you can no longer believe that a photograph, no matter how realistic looking, wasn’t faked, either in part or in its entirety. Specifically, it may be the point at which photographs are no longer admissible evidence in courts of law, at least not without strict standards to their origins and handling (e.g., having official chain of custody on the camera from the moment the photo was taken), because even an expert could not identify a fake any more.

Take a look at this woman. She’s not a particularly attractive woman, is she? Am I cruel for saying that? Not really, because she doesn’t exist. She’s a face created via averaging many faces to create a prototype “unattractive female face.” (The site also has prototypical male unattractive, and both gender attractive, faces, for the curious; I just picked that one because it was first on the page.) Surely it’s not much of a stretch from images like that to simply creating a photorealistic human being out of thin air.

Photoshop itself is becoming more and more accessible. Once, you would need complex tools and understanding, or at least a lot of patience and skill, to de-blemish a model’s image a magazine. Now, there’s a one-click tool that will do that for you. It’s not long before there’s a tool that identifies face structure and allows you to realistically wrinkle the skin in a photo, or move the cheekbones, or change the apparent ethnicity, or gender. Eventually, it may be commonplace for not just highly skilled graphics designers, but random shmoes like you and me, to have a program on our computers that automatically de-blemishes any photo we post of ourselves online, click-free. (The main obstacle to this right now is that, eventually, you have to meet people face-to-face...)

But the PTT is only the beginning.

Next after the Photographic Truth Threshold will be the Videographic Truth Threshold. Video is, after all, just photos in rapid sequence; it’s really just a matter of more work, realistically faking a video, plus a little understanding of make videorealistic movement. Certainly Avatar had moments of near-realism, but we’re not just talking about blue people.

To understand the technology that is going into this, watch this video. It’s kind of techno-babbly, but believe me when I say it makes sense by the end.

The implications, once this starts to become just a little bit a reality, are dangerous. Of course, if someone fakes a video of a UFO, we’ll all be incredulous (what happens if someone captures a real UFO? We’d never believe them), but if someone does something more innocuous and believable, like make it appear a person was in a location they weren’t, we could get into trouble. Presumably, if you could disguise the underlying coding or remove the traces of changes, no one would be able to tell through technological means; or at the very least, the ability to detect a fake would be inaccessible and highly advanced, so if some prankster made it look like you pissed your pants in a video, you’d have a hard time disproving the visual evidence.

Not to mention, if, say, a news network with a political agenda wants to make a point, they can fake the video. If the experts say its fake, they can produce counter-experts who say it isn’t, because, you know, they’re a news network. As of right now news networks and would-be propagandists are pretty much limited to carefully cutting and selectively reinterpreting existing documents of verifiable reality; but if they can insert a false smile at the right point here, or add someone in the background who wasn’t present there, all bets are off. Or, to take an alternative and less massive-conspiracy-theory scenario, imagine a grainy store surveillance video being manipulated.

Quick, without looking at the telltale video descriptions or comments – guess which one of these is CGI, and which one is not: Exhibit A - Exhibit B.

Of course, this only applies to visuals. Faking audio seems like another matter, but consider the advances we’ve had with ProTools and Auto-Tune, that can make a rank amateur vocalist on an detuned ukulele sound at least pleasant, if not quite like a pop star.

If the technology improves, however, and we can learn to manipulate the recorded sound waves so minutely that we can change the timbre of someone’s voice – then we’ll be in trouble.

Lastly, and most distant-future-fantastically, will be the Reality Truth Threshold. Assuming advances as mentioned in my post on brain-computer interfaces, there could eventually be a time when nothing you experience is necessarily real; someone could be screwing with your sensory input. Of course, on some level, that’s not technically different than right now – any introductory philosophy student will tell you that for all we truly know we could be brains in a vat, or inside the Matrix – and additionally the possibility of hallucination and schizophrenia cast our subjective experiences into doubt – but when we’ve observed the actualization of technology that allows the manipulation of sensory experience, it’ll be a whole different box of cats we’re dealing with. After all, the Matrix is only a theoretical possibility as far as we know at the moment, and schizophrenia can’t be artificially and intentionally induced (though a little LSD can help…), but if the technology is produced and verified, you’ll no longer know for sure when you’ve woken up in the morning that someone hasn’t implanted something in you and tweaked things just a little bit, so that whenever you step into the shower, the water feels cold, even though it’s actually scalding. You wouldn’t know the difference until your skin started peeling off. (Or the stall filled with steam.)

(I readily admit this is an idea oft-noted in sci-fi; but very rarely, as far as I am aware, does the writer of sci-fi have his protagonist in a world which is *actually* reality, but no one can be sure if it is. Always - because it's more interesting plot-wise - it turns out that it was Not Real All Along. But, if this comes to pass for you and me, it Will Be Real, but We Won't Know For Sure.)

So there, I’ve gone from near-future actual possibilities to sci-fi doom scenarios. But isn’t that always the slippery slope? Till then, have fun enjoying each other’s Facebook photos… while you can trust them.


David Pratt said...

Photoshop an image of my brains flying out, because my mind is blown.

Ozkirbas said...

What lengths, do you think, would be necessary to ensure that technology capable of creating perfect renderings of either real or fictitious persons could be identified as fakes?

Brett said...

Perhaps real-live cameras could require some sort of program that automatically inserts a micro-code or identifier into any real photograph, that can be traced to the camera itself - and its hard drive, which could contain an undeletable code-record of every picture taken, if not the picture itself. In other words, if you have an image which might be faked, use a program to find the code; get the camera that the code goes to; check the hard drive of the camera, and there will be a code that essentially says, "Yes, this picture was taken with this camera."

That's all theoretical, though, cause I really have no knowledge about if such a technology is possible. But it does seem that, if images are infinitely manipulateable, only the original photographing device itself can be trusted.

B.Graham said...

I trust technology will evolve with.. itself. At least eventually. There's always an equal and opposite reaction.

Brett said...

If there's always an equal and opposite reaction, what's the equal and opposite reaction to the nuclear bomb? Or wiretapping, identity theft, or Rohypnol?

Dialectric said...

Most PD's now use RAW data photos, which have a shocking amount of information about the camera and the photo embedded in them, some of which cannot be decoded except on the original camera. But as the author notes - this is all academic. Everything that has ever been accepted as the paradigm of evidentiary truth has always been both a lie and a euphemism - doubt is ALways reasonable, and nothing can be proven beyond a reasonable doubt, with the exception of certain needless solipsistic axioms. (Thank you, Rene.)