How wonderful to see something new! At SIGGRAPH last autumn Avidan and Shamir presented Seam Carving for Content-Aware Image Resizing, together with a striking video:



There’s now a GPL implementation, seam-carving-gui. It’s amazing. I grabbed an OS X Universal Binary, and was instantly goofing off. It does everything the video shows, except that you can’t stretch and shrink images interactively. Instead you type in the new size and it goes away and thinks.

I started with Trafalgar Square:



Then I crudely painted over Nelson’s Column and removed it to produce an alternate reality - Place De Le Emperor Napoleon perhaps:



Then I stretched it without modifying the appearance of the people to create another timeline with Albert Speer’s aesthetic in London:



I suppose it’s the graphical analogue of those weird boxes that beat-match tunes without distorting the pitch!

The last new thing like this that I saw was Dasher, an innovative approach to text input, back in 2002:



Why is something new such a refreshing delight? Because it’s rare. Innovation is not as common as we would like to believe. Sure, to some extent we are looking at an S curve, because a lot of traditional activities have now been automated and we haven’t thought up many new activities that require heavy data processing support yet. Facebook is hardly a paradigm shift!

A corollary of that relates to the way people just don’t worry about sizing so much any more. There came a point, maybe about five years ago, when the processing and storage requirements of a whole bunch of apps were suddenly met by standard domestic PCs. So far, humanity hasn’t generated many problems > 2GHz / 2GB RAM / 50GB disk. Even those lost UK government records were on an Excel database!

The sobering truth is that for many years, the big progress has involved delivering and incrementally improving features that I had on my Sun pizza box at BT in 1990, to the domestic market. Hats off to the materials scientists who have accomplished their repeated die reductions, but that’s the limit of the true innovation. In it’s day the KIM-1 was a solution looking for a problem, and in those days there was more innovation. Dasher and seam carving prove that innovation is still possible in the context of a single user with a PC.

I’m not the only person to see this dearth of things to be enthusiastic about. In his recent Yuletide greeting (and discussing the consumer market), 2007: The Miserable Year in Review, John C. Dvorak said, “I was not a fan of 2007. It was another crappy tech year—just the latest in a string of bad years dating back to 2000.”

As you might expect, I assign this reduction in true innovation to the effects of increasing background stress. If we can get that under control, we’ll be back on track for the Singularity. Hooray!