Glad to see confirmation of using a rubber band to fix an IBM mainframe. Been there done that, but the story is so loony I wondered if anyone would believe me. http://laughtonelectronics.com/oldsite/comm_mfg/commercial_i...
In the 2016 HN thread "Strange bug workarounds" I posted a much gnarlier problem (and oddball solution): https://news.ycombinator.com/item?id=12485921
(I often check the posting history when someone posts something interesting.)