While wandering the floor at tradeshows, I’m always on the lookout for interesting swag. I’ve been in the business long enough to become jaded – I don’t care about t-shirts, pens, or the normal run-of-the-mill stuff. I like the quirky, funny, and unusual items that you don’t normally see. In that vein, I picked up a magic 8-ball at the SGI booth and, while keeping my expectations low, decided to test it to see if it actually was magic and if it could help me with everyday decision making. In just a few short hours, the 8-ball helped me select a restaurant and lunch – with great results. It also helped me set up my WAN connection when I couldn’t get through to Verizon.  After just a few hours of consulting it, I started to depend upon its advice and counsel. But there are a few concerns:  1) Sometimes the answer isn’t clear, and I have to retry my question; and  2) I’m not sure I’m maximizing the 8-ball to its fullest extent.

I think there are two clear choices of action:

1.  Stop relying on pieces of plastic floating in colored goo to make every important decision in my life.

2.  Take the necessary steps to transform the magic 8-ball into a true decision support platform.

I’m leaning towards #2, but think that I’m probably going to need many more magic 8-balls in order to build up enough scale to get reliable decisions. Not sure if I want to use them in parallel – i.e., shake them all at once so that they can all ‘vote’ on a particular question…or…use them serially to run down Monte Carlo-like decision trees. It would probably be best to have both infrastructures in place, as some questions may naturally lend themselves to one method or the other.

With this in mind, I asked some pals on the HPC side of the industry for some feedback….will post as they come in…