Work Underway To Finally Build Babbage's Analytical Engine
mikejuk writes "Last year John Graham-Cumming launched a project to create a fully-functional implementation of Babbage's original design for a computer — the Analytical Engine. Now it looks as if the project is going ahead. The first phase is to digitize all of Babbage's papers and designs. These will be available to the general public in 2012. The machine to be built is no simple calculator: it is a full computer with a store for between 100 and 1000 values, each of 40 digits, and it was programmed using punched cards in a modern 'operator/address' format. There was even a plan to send the output to a printer. When this device is built it will make it clear that the computer age nearly began in the 18th century."
The entire design for the anaytical engine was extremely impressive. The main thing to realize is that the Analytical Engine was Turing Complete http://en.wikipedia.org/wiki/Turing_completeness. This means essentially that given enough time and memory it could emulate any program you want to. There's an idea called the Church-Turing thesis which says roughly that the set of things which a Turing machine can do are precisely the things which humans can algorithmically simulate. To appreciate how highly this speaks of the actual design of the Engine one should realize that many early computers like the Harvard Mark I were not Turing complete (although all the early Zuse machines were.)
The 1800s are the 19th century, how did this not get edited?
Ok, I know that the original design required making it out of brass and steel and whatnot making it big and expensive. But if I had one of those new (cheap) 3D printers, could I make a (smaller?) scale version out of whatever plastics or resins those printers use? Or are the tolerances too demanding? Would the job be made a lot easier if I "cheated" by using electric motors judiciously placed instead of the (possibly) steam powered original?
Now THAT would be one heck of a weekend project!
(Failing that, I heard they were going to make a computer simulation of it first to "test" it. It would be great if they could use some commonly used engineering program like Pro-E or Solid Works and build the model in that. Then we could all play with it!*)
*assuming you have a license for one of these programs lying around.
P.S. Then again I guess a mathematical translation of the Analytical Engine to a Turing Machine would also be sufficient.
P.P.S. I guess some day some nano-technologist will make this thing out with each individual component being just a few ATOMS.
It should provide good hard evidence of prior art.... Now if only they could find an brass iPhone design in his papers.
I'd love to see the designs for Zuse's work digitized as well, even though his real work did get reconstructed. The man independently (re?)invented binary floating point, made the first real programmable computer, all apparently without study or knowledge of Boole or Babbage, simply because he was a civil engineer sick of doing math by hand. That's just awesome and needs to be commemorated.
Excellent video of his "Difference Engine" working http://www.youtube.com/watch?v=BlbQsKpq3Ak. Seeing the Analytical Engine working also would be amazing. It's mentioned here http://www.youtube.com/watch?v=BlbQsKpq3Ak&feature=player_detailpage#t=471s in the video.
Wannabe nerd.
You would need a physical simulation - i.e.something that does real material, friction, gravity etc. - to be certain that the thing would actually work.
Modern CAD environments like Autodesk Inventor and Pro Engineer support that. Generally, you'd model subassemblies with the physical simulator (with friction, torque, stress analysis) to make sure they'd work, then switch debugged subassemblies to kinematic mode, where gears work in an idealized way and big systems can be simulated.
From the article:
"In the first instance the digitized documentation will be restricted to John Graham-Cumming and Doron Swade for the purposes of Plan 28 and in 2012 will be made available for research purposes and hopefully will have full public availability in due course."
That's a bit much for century-old documents. Fortunately, Plan 25 is open source and on line, along with a simulator in Java.