My initial brainstorm would be that this idea is a perfect way to write a story that explains the effects of this technology on the world and the way it works: have a VOC supervisor in Batavia or an East India Company manager in Delhi opening up a machine shop.
Heck, someone could make up a company. That would show the effects of the technology on the rest of the world as well as give a logical reason to explain how it works.
A cheat around As You Know Bob!
Sounds like we're sticking a transmission between a lathe and the AE.
Precisely. As we all apparently aren't working from the same start point relative to the ability of the AE, speaking in such general terms are the safest. Regardless of what the machines themselves can do, they are mechanical devices that are powered by some means, and that power is turned into work through some manipulation of the machines. At the very least, the AE acts as a 'gear shift' to control the input of power and which tool head is being used.
Much like the player piano analogy I used earlier: each tool head can be used, but you'd have discord. In proper sequence it is harmonious production.
Granted, as someone mentioned below, this is not the first iteration of the device. This sort of thing is most definitely late 1800's TTL.
Well, then what is it bringing to the game, then? The NCs are unable to make movements unless the tools are manually adjusted or the cards are replaced. What is the AE improving?
I probably wasn't being clear in that. I meant that if you build a machine to do X number of tasks (Here, X includes all
possible tasks performable), but now you want X+Y number of tasks, in order to get the Y tasks you need to modify the machine. In the case of a mechanical computer, that would require going into it and changing things, which is rather complicated, while in an electronic computer that most likely requires just changing out a circuit board.
The monitor doesn't need to be there, but it is useful. When we talk about machines, these two are the most common and definitely the most useful: without the lathe and the Bridgeport (the design of vertical that the bottom machine is), no one would've been able to make any guns in WWI or II.
I wonder if a AE could simply have a rotating wheel with various data printed on it that would show up on a display. Seems simple enough in theory for it to be mechanically linked (After all, we have examples in antiquity of mechanically programed plays, and things like the Antikythera mechanism).
The analytical engine was much more powerful than what some people here have been implying. It was, in fact, a programmable, digital, Turing-complete computer, therefore (mathematically) equivalent to any modern computer. Provided you had the correct interface from the AE to the machine tool, it could theoretically do anything that a CNC tool could do today (though there would be issues with the slowness of the mechanical design relative to an electronic computer).
While insofar as the Wiki article on the AE agrees with the initial sentences of your post, I think you are jumping when you say "correct interface".
As it stands, utilizing the AE for a task like this is not simply calculating 20 digit decimals. It would require someone to calculate the basic if-then parameters to design the gears and other moving components, and then you could use the punch-cards to control which if-thens are engaged (although you could simply make a device sans punch-cards, but it would only do one task).
In engaging the engine, you would have it run through a series of operations (the gears whirring and bringing the task specific components into position) until which you would be in position for the conditions of the if-then statements (controlled via punch-card and tool presence) to be met and then the tools would be activated. The tools would then operate for a set time before being disengaged and another tool would be engaged.
The most straightforward way to employ it would be to use it to read and write punch cards for the NC-tools, rather than employing a human to do it, since it was designed from the start to read and write information to punch cards.. Then you need one Lovelace to figure out how to program the AE to develop instruction sets based on input designs, rather than one for each design you might want (in other words, a kind of CAD/CAM). Did anything like that actually happen where a computer was used, not directly, but to develop the program for the NC-tools, Mac?
Again here, I don't disagree with you, but I think you are jumping into the future too far. You are thinking of the AE as an actual computer, whereas I think the contemporaries of these machines TTL would treat them in this respect as a 'moderator' or an 'overseer'.
Now, where this gets really cool is when some enterprising AE-programmer figures out this can be folded back on the AE itself, that is the programmer can input a description of the program he or she wants to write, and the computer will write it for him/her! Et voila, we have a compiler and high-level language!
An awesome idea, but still in the future.
EDIT: Now, mathematically speaking (the original thing this post was supposed to be about!) having a Turing-complete machine (even better, lots of Turing-complete machines) floating about in the mid-1800s will make computational mathematics and physics much more valuable and interesting. While a lot of useful work was done in those areas IOTL, workers were hampered by the fact that there was no really good way to actually do the calculations required. It could take years to do a single computation. The AE might be able to do those in months, still slow but much, much faster than acting manually. I don't think it will have *much* effect on non-numerical mathematics--those fields (in my own experience) are difficult to "compuationize", so don't benefit as much. However, the likely attraction of workers to the computational field might have significant effects on some more abstract fields that were developed in the 19th century, like algebras or certain areas of theoretical physics, mainly by slowing them down (until and unless people figure out how those might be important elsewhere or just catch up).
Yes. This is very important. If the AE is used first as a sort of controlling means for machine tools, and eventually its basic principles become used in the typical idea of the AE, then these possibilities are opened up and we could easily see physics and mathematics getting a decade or more ahead of their time.
Heck, tool moves could be done, in reality: in this early version of machine shops, the tool moves would be the realm of quality control if I'm understanding you correctly.
What do you mean exactly by the text in bold?
A part could be checked by quality control once a day (like it is now) to double check hole sizes, thread measurements, and most importantly: machined locations. A machinist can't eyeball such things as if the centre of a hole on one side of a part is within .010 of an inch of the centre of a hole on the other side.
So let's say quality control checks the part and decides that the hole needs to be moved. They then consult a programmer with the move they want to make and show it to him on the blueprint. He goes out, finds the card (or cards, as the hole may be done by more than one tool) that would be machining that hole, and feeds the necessary information into an AE using the information they've got on the current dimensions of the hole.
Then the programmer comes back out, replaces the cards, and the machinist runs a new part to go to Quality Control for a double check. In this sense, the only limit to how accurately a machine shop could make it's parts starting in the mid-1830s would be it's tools.
This would be very interesting, as you could have two parts that except for the position of one feature are exactly the same, and you could then very accurately test each for superior quality. Firearms would benefit greatly from this, I think.
Wait...so you're talking about...what? Proto-CAD or something? Because that would blow my mind.
I don't think CAD could be done on purely mechanical systems, at least not ones like this. There's two issues that concern me, performance and input/output.
Running a CAD program is a hugely expensive in terms of computer time, especially when you think of image output for preview, without which it's no good. Imagine you're working, create a part, then change something that takes a second of computer time to process. If the AE has 1000 times less processing power than the processor on your machine, that half-second change now takes almost 20 minutes before the result show up on screen. So instead of taking a sip of your drink while it thinks, you could eat lunch while the same relatively minor task is being done. And I don't buy the AE being even 1/1000th the power. I'm not sure I'd buy 1/10,000th, or 1/100,000th.
And then once you've got data in CAD and processed, you have to display it, and be able to rapidly edit it. You'd need a way to process user input of complex data, either a mouse and/or a keyboard or something with similar function if not similar workings, but it's still a whole other additional job that your machine is doing in addition to number crunching. And then comes the killer...a display. I could maybe see mechanical pixels, but I don't think I'm seeing a high-resolution monitor, or a screen refresh rate of more than 2 or 3 times a second. And running that...hell, many modern PCs take a discrete computer chip, some with abilities nearing those of the main processor just to keep up with the graphics.
I agree here. CAD is, in my opinion, sometimes over-emphasized. In terms of design, any rendering I do in an AutoCAD/SolidWorks style program I can do by hand in about the same time. Drawing plans on a computer only came into widespread use when, in the 1970-80s? And then only with the intensely complicated machines of the time.
It is in the realm of calculations that an AE would be of great assistance. That which takes the most time when I am designing something on paper is calculations, and if I had a dedicated calculator instead of having to do somethings out by hand (which happens when I forget my graphing calculator) I would save a lot of time.
But again, that is a result of the AE being used as it was intended, an AE. In terms of moderating machine tools it's a different animal.
EDIT: Just for fun, I opened one of my CAD files from class this spring. You can see the output below. Simple blocks, each in three flat views starting from a perspective view off a worksheet as practice in turning 3D into 2D views. It's not fancy 3D rendering or anything, what you're seeing are lines created on the screen exactly as I made them with the line tool. The dimensioning CAD did, I'd pick two lines and it'd spit out the label for the dimensioning bar thing. We did this the first day of class.
The CAD file is 70 kB on disc. The AE Babbage designed was to be able to store the equivalent of 20.7 kB according to Wikipedia, so he'd need to triple the memory to even just store this basic file. To start the CAD program took about 25 seconds. To load the files took about a second. So, for an AE running the highly unrealistic 1/1000th power of my processor, loading a simple CAD file becomes a "take lunch while the job runs" situation, and booting the program (which I guess wouldn't need to be done, because CAD would probably be all this AE version would be able to do) would take almost a day.
In short, major power, graphics, and memory increases are all needed to even be able to consider handling the most basic CAD images, ones that are barely more than digital blueprints.
Exactly, best that you give me that task of designing it and I'll take only that lunch period to get it back.
Until the power of the AE or later computers becomes greater, they won't be able to take over the roll of proper engineers and draftsmen. They would, like other contemporary machines, take over simpler tasks that would take away jobs from less educated folk, and do it more efficiently and accurately.
After all, you still don't have computers designing our submarines, we have engineers doing that with computers as tools.
And can you imagine what a calculator would do back then?
Teaming up a device that can make precise machined surfaces and a mathematical machine that can consistently be just as precise with the numbers? Heck, that's enough right there.
You don't need a factory run by AE, a calculator the size of an air-conditioner could revolutionize the industry!
Exactly. One or two Calculation Engines combined with an engineering firm would make that firm very profitable.
I agree completely. CAD is completely out of the question, automated factories are out of the question, but the simple quality and repeatability or production with a little assist in programming could be as much of a revolution.
Mechanical computers I think preclude CAD, but using AE like machines to 'automate' factories is possible, if a much later development.
Your last clause, however, is the most important one of all. Repeatable quality production.
Sorry for jumping, I just was a little bugged by people getting way beyond what was actually possible, combined with just a little bit of wowing over what modern computers do in terms of performance[/URL] for what we think of as
simple tasks.
Yes, it is very important that images of what could be here are rendered down into the most basic level. It is best to err on the side of underestimation I think.