I read that the recent market plunge of nearly 1,000 points was due to what we would call in the IT world, "user error;" in other words, an error that occurs someplace between the keyboard and the chair.

It’s been reported that a trader entered a "b" for billion instead of an "m" for million in a transaction that likely involved a Dow component, which in turn triggered the plunge.

That made me think about a project I’m currently working on and some of the discussions that we’ve been having. As I mentioned in a previous post about this project, we are upgrading an old warehouse management system (WMS) to a newer version. More accurately, we are replacing the old WMS with a new WMS since the two versions have so little in common; the systems are so different that it almost feels like they are from a completely different software vendor (which they’re not).

So during "future state" design discussions for projects such as this one, one of the team members – sometimes all of them – will say, "what if ‘Bubba’ presses Enter before he remembers to Tab over to the reason code field and plug in the right value? The old version wouldn’t let "Bubba" do that!"

This is the point where we imagine – and try to proactively avert– many of these Armageddon-type scenarios that could possibly cause the walls of the DC fall in and kill us all. (Okay, maybe that’s a small exaggeration.)

As this room is filled with a bunch of smart folks, we know that we wouldn’t do such a stupid thing, but Bubba would. So we are all paying very close attention to the options where we can limit the likely Bubba-caused disasters through proper change control and training, and those options that could really trigger some major systems’ problems and require extreme human intervention to reconcile. The latter really requires us to make the option Bubba-proof.

The bottom line is that you can never make a system idiot proof. As soon as you do, up will walk an even better idiot and prove you wrong. But what we are doing is walking the fine line in trying to Bubba-proof the system as much as possible while making sure that we capture all the key components to affect proper change control.

What has Bubba done to your system lately and what did you do about it?

--David

Tags: wms