Jinks!

I love XKCD, even when they steal my ideas. Compare this comic with this blog post from 2007. Of course, I embedded an XKCD comic in my last post, so I guess we're even.

Analyze, Acclaim, Break, Repeat

There seems to be a pattern that we follow when we are faced with a new technology. First we (meaning people who are interested in technology as technology) first want to know more about the thing. What is it? How does it work? What OS? What's driving it? and so forth.

Once we've established some basics, if the idea is any good, we are usually pretty quick to declare that something is a breakthrough. We like breakthroughs! This is especially true in the world of Apple and its products.

Once we've decided something is good, then it's time to see how far we can push it before it breaks. At this point we start mocking it for not being perfect, but if it's good enough we continue using it.

Two recent real life examples are iPhoto '09 and the latest Android update.

iPhoto has the much-talked about "Faces" feature, which goes throgh your photos, locates faces and does as much as it can to help you sort them by who they are. My photo library is about 10 GB, mostly people, so I figured it was a pretty good test site. I went in and started tagging people that would be repeated over and over. Me, my wife, my two kids, various siblings, and so forth. It picked up on who I am pretty quickly; boxes around my face went from "Unknown face" to "is this Nate?" after about 30 pictures. My kids are both under the age of five, so it's understandable that iPhoto can't quite tell them apart yet, but it actually does a pretty good job. My wife, however, seems to be a complete mystery to the iPhoto daemons. It confuses her with her sisters, my sisters, and occasionally asks if someone's hand or shoulder is my wife.

And thus we hit the "Break" stage. I decided to see how off it is compared to human recognition. Because a human would never do that. If you showed me 40 pictures of my two sisters-in-law, just their faces, I could probably tell you with 100% accuracy which was which unless you pulled out the traditional "one-day-old" pictures, because those all look the same, man. iPhoto gets about a 40% on telling them apart. And while we may tell ourselves that we see faces in trees, shadows, the clouds, etc., we don't for a moment think that those faces are our brothers. So we get to feel superior to technology for a while longer, and completely miss the point that Apple taught the computer to recognize and categorize faces at all.

The other one I just got my hands on is Google's voice search. My G1 auto-updated yesterday, and when it was finished my Google search bar had a cute little microphone on it. I spent a happy 45 minutes asking it for truly strange things like "football sushi" or "directions to the Los Angeles lockers" and it understood my voice (if not my questions) every time. So I tried to show it off to my brother-in-law on the commuter train home and it failed every time. Well, let's be fair, I was having a hard time "parsing" the things my brother-in-law was saying on that amazingly noisy train, so we have to cut it a little slack. But it still gave me a momentary "I'm still better than the machine" feeling.

I don't really have a point, by the way. This is mostly by way of obervation. But I do think it's healthy, and the "break" step is the one that keeps technology moving forward and interesting. So let's go see what shiny new breakthrough we can break today!