Over at On the Edge, Keith Campbell makes the case for packaging machines to become All Remembering:
Consider that $200 will presently buy you enough memory to store everything that you would ever read, everything that you would ever hear, and 10 photos per day for an entire lifetime. But, if we are on the cusp of such a breakthrough for people, consider how much easier it would be for machines. Without spending too much extra, we could know every position, bump, temperature, humidity, voltage, speed, reject, operator, maintenance procedure, … your get the idea ….., ever seen by the machine. If our machines had total recall, could we design them better? Could we maintain them better? Could we operate them better? I think we could.

What would it be like it machines had Total Recall?
On the consumer side, those cheap 1TB+ drives are great, but you’ll need more storage than that. 10 photos a day is great at your current camera’s resolution, but not tomorrow’s. HD Video takes up a tremendous amount of space, but the drives are keeping up.
But that’s a side note on his overall excellent point that there have been great advances in data collection and data search. Here are some examples of that:
I used to sort my inbox by project or event, but now I dump everything into a giant folder and search for what I need. It searches incredibly fast through thousands and thousands of emails and I find things much easier than if I “organized” them.
I have thousands and thousands of songs on my iPod and iTunes profile. I used to rate them and build playlists from the ratings, but I couldn’t keep up. It took too long to rate every song I added, so now I let Apple run through that data against other peoples’ playlists and it builds playlists for me based on a selection of a single song I like. (This is called Genius in iTunes).
So where does that leave machines? I think the first part (data collection) is already being done in many ways. Companies have mountains of data for their overall production, line downtimes, average machine speeds, amperage readings, etc., but it’s almost like knowing your blood pressure number, but not knowing whether that’s good or bad. Packagers can say with absolute certainty how they are performing on the current shift, but they can’t predict next week’s and that’s where this data will be useful: Finding and correcting problems before they occur. Ways of searching and making sense of that data need to get better before that can happen.
Part of the reason why this hasn’t happened yet is that as far as we’ve come, the computer revolution hasn’t come to packaging machinery yet. It’s better, but I have more computing power in my phone than most of the standard PLCs on the market… and guess which one costs more.