Was it IBM or ICL that "upgraded" to a faster machine by removing a resistor board?
Pretty much all the mainframe manufacturers (IBM plus the BUNCH - Burroughs, Univac, NCR, CDC and Honeywell - ICL didn't appear because this was US jargon!) did this. It actually made economic sense for both manufacturer and customer, mostly because there was just a single type of system to be built rather than half-a-dozen, giving economies of scale (and making upgrades really, really simple).
Back in the day (c.1980) we ran a Honeywell mainframe, middle of a range of 5. Part of the maintenance contract (excruciatingly expensive, this was actually where most of the money was made - you could get a good deal on hardware, but there was rarely any negotiating on maintenance) was a visit every other week by an engineer to run diagnostic tests. To save himself time, he would reach inside the machine to the 'secret' microswitch that turned it into a top-of-the-range model. Of course, the operators soon sussed this out, which meant that work scheduled for an entire weekend could be accelerated to complete in under a day, leaving extra pub time ...