pretty pictures! 
Not so pretty to analog signal engineers! In diagrams like these, you want thin, tightly defined lines. Without heroics, though, you end up with a picture like the upper-right one–not just fat lines, but a homogeneous fuzzy blob. It’s astonishing that one can massage the signal to get anything useful out of that. It’s black magic.
So what they are saying is that my TI-59 programmable calculator had more horsepower when I played Lunar Lander than the real thing had just twelve years earlier?
Actually, no. The real thing had a lot more memory, for one. Later TI calculators were faster but not this one.
In addition, the Apollo stack had analog computers that aren’t in that calculator, and they were effectively performing operations that would be way too slow in the calculator.
Well, yeah. That’s why those things were more than adequate. In the days of mainframes, I worked in a plant where the engineers had a process model that would tell you to the nearest pound how much of the millions-plus of material was in process at any given time. This, despite, for example, material input via a feed belt weightometer with a 5% accuracy and a hiccup in the reading every time the staple in the belt went by. (Of course the engineers knew this - it was simpler to run with 8 digits than to account for tolerances in the model; they just made a “correction” based on inventory and estimates every 6 months). In fact I learned this when one of my jobs was to stop the belt, scoop off and weight 20 feet of feed, and compare reading to reality to fine-tune the model.
But then a slide rule is a perfect example of your point. The world got by on 2.5 digits accuracy (or less) for a century or more. Also, a mechanical system like an old auto engine worked fairly well. OTOH, far more precise and adaptable computerized fuel injection and ignition has made a huge difference in performance, cleanliness, and efficiency to that same internal combustion engine… and this is the promise that computerized process control holds out for a huge number of processes - provided we can get past the 2-digit accuracy of many inputs.
I should also mention the early process control device I helped with - where the X-squared variable was approximated with a linear calculation. As long as the process stayed within certain bounds, the approximation was “good enough”.
You’re probably right; I remember the story but not the specifics. Probably conflating it with one of Voyager’s slide rule covers. Wonder why they called him “Slipstick” then if he didn’t need one.
Clarke’s Law in action.
They didn’t call him “slipstick” because he used one; they called him that because he was one. He was a human calculator.
It’s fun to make old mechanical calculators divide by zero.
[quote=“Enola_Straight, post:48, topic:810986”]
It’s fun to make old mechanical calculators divide by zero.
[/QUOTE]Ah, yes. In high school we had a Friden Comptometer (looking something like this - top of page in our computer room, and making it divide by 0 was a lot of fun. We also had an early electronic calculator, an Olivetti Programma 101, but that just gave a boring error message.