I think that the fundamental problem is not the one technology. Election systems have this odd requirement (compared to most stuff we build) they are used both

I think that the fundamental problem is not the one technology. Election systems have this odd requirement (compared to most stuff we build) they are used both very intensively and very infrequently. These 15 year old systems have probably been used less than 30 times. From a procurement perspective it seems wasteful to need to replace something after so little use, but so many of the computer systems that we use have anticipated lifespans of only a handful of years. I suspect that an electromechanical system that records votes onto a medium that can be easily interfaced with technology du jour (and audited by humans when necessary) would have an expensive but long lived set of components that deal with the large numbers of voting actions, and then the ticker tape (e.g.) output can be read with a device that plugs into an off the shelf computer