How long is a lifetime of practice in human-computer interaction? In my case, it has been 42 years, almost all of that in IBM Research (with a couple of sabbatical years in the then new Watson Group). Having queued up for access to DEC and Prime machines during my PhD work in the mid 70s, I was positively thrilled when, upon arriving at the T.J. Watson Research Center in 1978, I found a dedicated 3277 "green screen"terminal on my desk attached to a 370 mainframe. Hot-rodded with an onboard character buffer and double speed cursor circuitry, I could fly through writing and coding, personally experiencing the productivity benefits of sub-second response time . Beyond some great hardware (for the time), I also had extraordinary colleagues. I imprinted on the working style of John Gould and Stephen Boies as we built the second hardware iteration of what became the world's first commercial voice mail system and its final manifestation as the messaging system for the 1984 Summer Olympics. In this talk I'll share some of the still relevant behavioral principles of system design we learned in the course of that project . Beyond this, I'll review work in early (pre-Web!) Internet access for schools , home shopping , Web accessibility , and programmer productivity in peta-scale scientific computing . Each project will highlight a barrier to building useful and usable systems, barriers that still persist. More importantly, I'll tell you how we overcame those barriers, giving you something that will, perhaps, be useful in your own work.