Can't the fatal flaw always be duct-taped over? ;) The problem I often have is not with stuff that clearly just does not work. The problem is the stuff that "

Can't the fatal flaw always be duct-taped over? ;) The problem I often have is not with stuff that clearly just does not work. The problem is the stuff that "sorta maybe works" and then keeps working until it doesn't. You can't really prove it won't work, because hey, it does work. There's always some convoluted way of getting it to keep working. It's really tough to make the argument that further down the road when the software grows, the scale grows, or the requirements change, the proposed solution won't be easy to adapt or won't be as robust/reliable. This is the really important stuff. I would say that investing in things that just never work at all and can't be made to work is more rare, probably still happens, but more rare.