There is an oft-cited claim that if you let a bug get into production code, the cost of fixing it increases a hundredfold, compared to if you had flushed it out during testing. This report <https://www.theregister.com/2021/07/22/bugs_expense_bs/> looks into the basis of that claim, and it seems to originate from a book published in 1987, which gave it a reference “[IBM81]”, which was “course notes at the IBM System Sciences Institute”. Only there was no such “Institute”, it seems it was really just the name for an internal training program for IBM employees. It also seems a related claim, that “defects cost more to fix the later you fix them”, also has weak, if any, empirical support. Nowhere in there is the issue of “technical debt” addressed, though. That is real, and can be very costly, and its costs do increase over time.