This is a brief blog, spurred by reading on a sunny Sunday afternoon in my garden.
Take a read of this and tell me if it rings bells. I’ve obscured a couple of details.
…the philosophy of the company was that many defects could only be detected by test running of completed [deliverable systems]. Therefore a row of test cells went across the entire rear section of the [location]. Each [system] was run for 8 to 13 hours, then completely disassembled. The [components] were inspected, replaced as necessary, and reassembled. The [system] was run for 5 to 12 more hours and then, in the event no problems were found, it was shipped… this final safety net created an ‘assemble it, then tinker until we get it right’ mentality
Isn’t that pretty much how we do software testing today? And generally, we don’t have a better way to do it.
This is how Pratt & Whitney tested aero engines, up until 1946 (as described by Womack & Jones in Lean Thinking). While I don’t currently know what Pratt replaced this method with, nor do I have a software domain proposal, it’s yet another parallel between current Software practises and those of the early manufacturing era.
There are other examples from even earlier in Pratt’s history, but I’ll leave those for another day, when the weather isn’t so enticing.