In some previous entries I described my skepticism of Riya -- a supposedly extraordinary new image recognition technology that could amazingly discern between twins, among a plethora of other amazing features (such as discerning the specifics of highly obscured, stylized text). The "blogosphere" went nuts, with a lot of believers hoping they'd catch the leading wave of this amazing new innovation, leading the trend.

Lots of predictions were made about Google or Yahoo paying hundreds of millions for this revolution.

Looks like Riya is now a handbag finder.

Nonetheless, today I see a featured Slashdot story declaring the privacy invasion that an amazing new facial recognition will potentially bring. One again it's a "coming soon" technology that's going to revolutionize the image analysis market.

[EDIT: Remarkably, without even the slightest ounce of proof of any real world value, this "technology" has been featured on many of the largest news sites. Now that's P.R. that's worth its pay]

Of course this one is very cleverly marketed: If they simply said "We're going to introduce something amazing sometime next year!" most would rightly dismiss it as vapourware, waiting for actual proof. Instead it's introduced under the "what is this going to do to privacy?", a distracting question that lets lazy technology writers run with it without actually researching the validity of the underlying promises.

Will there ever come a time when people are just a little bit more skeptical of this sort of thing?

QuickNotes

Visual Studio 2005 SP1 was released last Thursday. The computationally demanding 442MB installer runs for hours, so set aside some time before you start.