Curmudgeon said:
Bottom line is, it is far cheaper to buy stickers than to pay someone to actually inspect.
I think this is true, but the whole concept of 'inspection' is about 50 years obsolete as well. What rules is the schedule. You all probably realize that money tied up in capital investment and materials that aren't pounding out product is a very bad thing. You can argue all day that "Yes, but if turn out bad products you'll lose your business" which may be true enough, but the fact is that you MUST turn out products, and right now.
So, I’m on an internal IT project that I stupidly volunteered for. Having spent 25 years in IT and 5 years getting the hell out, I thought what the heck, let’s go see and learn from a large organization doing a big project. What an idiot.
Here’s how ‘inspection’ works. The developers have a schedule–and regardless of ANYTHING else, they WILL produce their products on time and get them in to test. Nevermind that they didn’t test their own work–they didn’t have time to. So, when test gets stuff that has known errors in it–stuff that everyone including management ACKNOWLEDGES has known errors in it, Kirk asks, “Why would you send something in to test that you already know is broken?” Well of course the BS starts to flow, but the real answer is, “So we can honestly state that testing began on schedule.”
So, while inspection should be checking to ensure process is in control, there is no process to control in the first place. All inspection does is highlight the endless stream of errors that always come out of a process-free system.
With literally hundreds of errors to deal with, someone has to prioritize them because they all can’t be fixed and still stay on schedule. Because there’s no process, the crap that gets fixed simply has a whole list of NEW problems that weren’t even there before (side effects), and within a few short weeks, both the development team (100 people) and the test team (20 people) are completely overwhelmed, up their ears in garbage.
The first thing you have to do if you’re going to survive at all is to stop doing re-work. You simply can’t fix the errors that inspection finds or you’ll kill yourself and production will stop. And hey, the guy who gets paid to ship on time DOESN’T get pinged for returns! So, what do you do? You ship [your busted crap] on time!
On my project, this problem is solved in a more sophisticated way. The errors are categorized into ‘fix’ and ‘enhancement’. Since we all know that we can’t let scope get out of hand, everything that goes into the ‘enhancement’ bucket is seen (by management, who has not Clue One about what any of this stuff is) as increased scope. Problem solved–20% of my broken parts can now be written off as enhancements and basically pushed off the deck never to be seen again.
But the project is on schedule and will be deemed a success.