Disinformation: Live and Obvious in the Windows 11 Crazy Train

April 28, 2022

I noted that a number of OSINT experts sidestepped the issues of misinformation (making stuff up), disinformation (data which nuke other information), and reformation (moving the data walnut shells like a walnut shell wizard). The experts offered comments at a recent conference I attended, and I was fascinated by the avoidance of what seems to be as a showstopper for analysts.

Let me give you an example unrelated to the professional OSINT lecturers.

The first is the story in Ars Technica. The headline is “Businesses Are Adopting Windows 11 More Quickly Than Past Versions, Says Microsoft.” Straightforward and actual factual.

Now consider “Windows 10 Still Growing, But Win 11 Had Another Bad Month, Says AdDuplex.” This appears to report data slightly off course with the Ars Technica write up.

Okay, are both sort of true? Is one statement more accurate than another? Maybe one or both are baloney?

The problem is that in order to figure out which is disinformation, one has to do quite a bit of work.

Now imagine that a really smart machine learning system ingests the content and shoves it into a whiz bang smart software system. The smart software will do what? Identify the rightness or wrongness of each set of factoids? Will the smart software go with a simple voting method and the most likely rightness will emerge from the murky plumbing of the smart software? Will the system punt as some digitally learned systems do?

The answer is that manipulation of information can generate outputs that may be disconnected from what is shaking in the real world.

Is this a problem? Yep. Is there a fix? Nope. Are there downstream consequences? Does a calculating predator exist in the technology theme park?

Stephen E Arnold, April 28, 2022

Comments

Comments are closed.

  • Archives

  • Recent Posts

  • Meta