First, let me say that I heartily sympathize with your horror story.
Second, though, I do think that it is an apt illustration of not fully costing the product (and even "internal cad" is a product to the design team customer). When it is hardware, be it brick-and-mortar buildings or be it oscilliscopes and test equipment, we have no problem with the comcept of depreciation. "Hey", we say, "it is going to wear out (or become obsolete) and it is going to need to be replaced within X years, months, or weeks, so we better build the cost of replacement into the fee we are charging (to our customer or our P&L center)." Hence, we can judge if we are making a profit off of the entire life cycle of the tool (or building or whatever). Well, it is important to have built that same depreciation cost into any obsolete-able software product. Personally, I also believe that we, as both users of and/or developers of, such software, need to be mercy-less in our decision to slash and burn (that is replace) old software once the "cost depreciation designed-in obsolecent date" comes. That means, we need to be building and testing (or buying) next generation software before it is needed, so that it is ready when the previous generation is "tossed".