A different take on Attention
I’ve panned the AttentionTrust quite a bit in my first few posts, but I must say it is really because they are so close to spot on. Powerfully close. Now if we can just jump from there to something we can actually use…
Squirrel Tao writes a bit about attention from a creativity standpoint. That’s attention with a little “a”. It’s nice, though, how the thinking applies equally well to Attention, with a big “A”.
The post finishes with:
Willian James wrote, “If we wish to keep our attention upon one and the same object, we must seek constantly to find out something new about it.â€
An absolute brilliant statement that subtly points out one of the key flaws of the GestureBank. What is the object of Attention in the GestureBank? Clickstream logs like the AttentionTrust Extension capture all activity and mish mosh it into a goulash, in the hope that after the fact, one can extract or identify the object of attention. The GestureBank then takes that data and makes an even bigger goulash. But as Chris Anderson writes in The Long Tail, top ten lists are useless without context. It is only in the niches that we can get value out of knowing the most common similar results. Clustering is one way to mathematically generate niches, but its use will prove limited to spaces where clean mathematical separation exists. So why does the GestureBank systematically strip the user context from the already context-free Attention log? That makes it pretty hard to discover the object of the user’s Attention.
Why not just let the user tell us?
Then we can avoid the whole post-activity reconstruction/clustering/meta-modelling thing.