Continuous partial attention
One of the challenges noted in this point is the overwhelming stream of notices that people get that they must check to see if a response is needed. This to me is often indicative of poorly thought out decision automation. If the systems were smart enough to take appropriate action 80,90,95% of the time people would only get the notifications they really needed to see. This is why I think Business Activity Monitoring needs to be matched with decision management - check out EDM and BAM.
Algorithms in the attic
The use of algorithms is what predictive analytics is all about and I could not agree more that the combination of data and processing speed makes these algorithms more attractive than ever.
When to sleep on it
I have blogged about this one before when I wrote don't sleep on it, automate it. I also think that decisioning has a role in helping employees do "blink" better.
If you take all three of these as a set I think you can see why I believe it is not about knowing first but about acting first and that what workers really want are systems smart enough to do a lot of things for them, smart enough to know when to "ask for help" and smart enough to help customers help themselves.