There are various news articles around about during a presentation, a few seconds of the presentation was not of the CEO, Jensen Huang, but of a computer-generated fake of him instead. What I’d like to discuss is how misleading the initial articles reporting this were. The first one was from Tech Radar, and reported on a blog post from nVidia, and had the headline, “Jensen’s Kitchen Was a Lie.”

In fact, only a second or two of Jensen Huang’s kitchen was CGI; the CGI portion (which included a digitally generated Jensen) was only in the digital kitchen for a second, then it transitioned to a nearly black, obviously computer generated set. The computer generated set and CEO only lasted for fourteen seconds and the computer generated figure was actually very small in the frame. Here’s a screenshot from that section of the video:

In context, and if you’re familiar with the state of the art in this sort of thing and how much work it normally takes, this was still an impressive demonstration of computer technology. That said, the reports of it made it sound wildly more impressive than it actually was. Which brings me to why.

First, I’m 99.9% certain that this was an honest mistake. nVidia’s blog post was written from a very tech-centered point of view. It was very detail-oriented in terms of what nVidia technologies did what. Basically, it’s how engineers tend to write, because engineers can only do what they do because of tunnel vision. But that tunnel vision also tends to make them bad at communicating with non-engineers unless they conscious frame-shift.

Then we come to the tech reporters who took the nVidia post in the most sweeping way possible. Again, I think that they did this honestly. I think it highly likely that the writer believed every word he wrote.

So, what happened?

I strongly suspect it’s just selection bias at work. Tech reporters are tech reporters because they love technology. They want technology to be amazing. If tech reporters want technology to be amazing, tech readers want that tenfold. A hundredfold. This creates a selection bias; reporters who report on technology being amazing get more readers, because they provide the thrill that the readers seek. Ordinarily, this will mean that they report the same things as others, but do so in a more thrilling way. Tech reporting benefits tremendously from the world producing news on, approximately, a schedule. The ever-increasing performance of computers on roughly a yearly schedule means that there is a steady-state supply of genuine news. (If, granted, news that only tech-enthusiasts find interesting. But, we do find it interesting.) This is one massive advantage that tech news has over regular news, who only get newsworthy events rarely and haphazardly, and so have to make up most of what they report in order to fit their schedule (they make it up mostly in the sense of inflating the importance of insignificant events more than outright fabrication, but the spirit and effect are the same).

The issue comes in when the tech news to be reported is ambiguous. The enthusiastic, optimistic reporters who readers select for will tend to interpret the ambiguities in the most optimistic, impressive way, because that’s how they are and they’re the popular ones because readers like that.

Another advantage of tech news is that it doesn’t really matter. No one is going to do anything of any lasting effect because they believed for a few days that nVidia was able to fake their CEO for longer than they did, or more convincingly than they did. Tech news also tends to be fast to correct in part because real news will come along quickly to replace any mistakes. General news may go months or even years without anything that people need to pay attention to on a daily basis.

Beware of news.