Reverse-Engineering

[folti78]

Well, the Polish resistance nevertheless operated a network of clandestine workshops making ammunition, small arms and grenades. They had to employ qualified chemists, though, and obviously based their operations on modern knowledge and industrial base.

I sit corrected ...

[Surlethe]

If that happens what are the chances that radiation and nuclear power gets discovered few decades earlier?

Remember that it took fifty years, give or take, from Roentgen's discovery of the x-ray in 1895 to the development of nuclear reactors in the 1940s. There was a lot of relevant scientific innovation in between -- the birth and development of the entire field of quantum mechanics, for instance, special relativity, and even the start of quantum electrodynamics. Remember, for our hypothetical 1840s submarine examiners, the only hint that the field of classical mechanics is wrong is Mercury's perihelion drift. They have no idea that x-rays exist; the periodic table of the elements won't be invented for another quarter century (1869); the Bohr model of the atom won't be along until 1913 and the first viable theory of nuclear structure doesn't arrive until 1935 with the liquid-drop model, on the eve of nuclear fission (first controlled reaction: September 1941, in Chicago). The point? Victorian scientists have no theoretical machinery to make sense of these observations, and the development of this machinery is going to take decades. Could radiation and nuclear power be discovered early? Radiation, sure; nuclear power, I think unlikely.

[The Duchess of Zeon]

So here's a question for you, Stuart.

Let's say those Germans in, oh, January of 1917 get ahold of that jet fighter and realize that they are quite impossibly out of sorts in replicating it.

But let's say it's a Mig-15, just for example.

Would they be able to field tens of thousands of 23mm and 37mm autocannons along the Hindenburg Line by June and July of 1918 which can chop through tanks like butter?

I.E., what are the more exact limits of reverse engineering? Obviously 18th or 19th century scientists can't reverse engineer a nuclear submarine, but could 1917 German scientists reverse engineer the autocannons on a 1950-vintage Soviet jet fighter?

[Stuart]

So here's a question for you, Stuart. Let's say those Germans in, oh, January of 1917 get ahold of that jet fighter and realize that they are quite impossibly out of sorts in replicating it. But let's say it's a Mig-15, just for example. Could they be able to field tens of thousands of 23mm and 37mm autocannons along the Hindenburg Line by June and July of 1918 which can chop through tanks like butter? I.E., what are the more exact limits of reverse engineering? Obviously 18th or 19th century scientists can't reverse engineer a nuclear submarine, but could 1917 German scientists reverse engineer the autocannons on a 1950-vintage Soviet jet fighter?

Certainly they could reverse engineer the 23mm and 37mm cannons but would they need to? Remember Hotchkiss built an automatic 37mm cannon in the late 19th century - the Boers used it which is where the nick-name 'pom-pom' for tw-pounder automatic guns comes from (a 37mm shell weighs two pounds). So, the automatic guns are well within the state of the art. The reason why they hadn't been built in great numbers is because they were actually banned by treaty. (Technically almost every fighter pilot in the world today is committing a war crime by flying a cannon-armed aircraft). There were a scattering of similar guns defined as anti-balloon guns. So there would be no problem in duplicating the guns on the MiG-15.

The problem now becomes would it be worth it? There would be several stages involved in that duplication. The guns would have to be dismantled, each part measured in detail and the parts re-drawn from that measurement. Then, the engineers would have to duplicate those parts, assemble them into a working prototype and test it. Then, they'd have to interview the survivors of the first gun crew and work out what went wrong The first test gun would almost certainly explode because of material and production standard differences so the engineers would have to work out what bits failed and how to reinforce them. Once they got the gun working, they'd have to design a carriage for it and put it and the gun into mass production. Then they'd have to distribute it to the front. All of that would take at least two years.

Now, there's a short cut. They don't have to reverse engineer the whole weapon. They have cartridges that work (23mm and 37mm) and upon which all the ballistic work has been done - developing new cartridges is not easy, All they have to do is take the existing 37mm Hotchkiss or its Mauser, Madsen or Browning equivalents and re-chamber them for the new round. Best way would be to take an existing 37mm and rechamber it for 23mm hoping the extra metal will allow for the higher chamber pressures. That would be a lot quicker than reverse-engineering the whole gun.

So, reverse-engineering the 23mms and 37mms is certainly plausible but the problem is that there are better ways to go, That's the trouble with reverse-engineering as a principle. If the object being reverse-engineered is more than a very limited advance on the state of the art, the unknowns in its construction will kill the engineers. If it is more or less state of the art, its quicker and simpler to develop one's own stuff. The only exception to that is if one lacks the personnel needed to develop one's own. The Chinese had that problem in the '70s and '80s, the Great Leap Forward and Cultural Revolution had left them desperately short of skilled technical staff and two generations behind in military development so reverse-engineering, for all its problems, was the only option.

[K. A. Pital]

I wouldn't say it's always "quicker and simpler" to develop one's own stuff when it comes to a huge technological leap. The unknowns won't necessarily kill the engineers, not all technological objects are deadly. Yeah, a nuclear submarine or a plane chock-full with explosives would. But those aren't the only things for reverse-engineering.

[Stuart]

I wouldn't say it's always "quicker and simpler" to develop one's own stuff when it comes to a huge technological leap.

My point was, one can't do reverse engineering where a huge tehnological leap is concerned, the unknowns involved are too lethal. Reverse engineering is only possible where the level of technology is roughly equivalent (like the automatic 37mm gun in WW1) and then its usually quicker to do one's own thing. There are some exceptions but they're rare

The unknowns won't necessarily kill the engineers, not all technological objects are deadly. Yeah, a nuclear submarine or a plane chock-full with explosives would. But those aren't the only things for reverse-engineering.

Not necessarily but that's the way the trend goes and the greater the technology leap involved, the more likely it is to have a fatal complication. It's not necessarily the big things, as we've already been discussing such buried factors as metallurgy or fabrication methods will do for the victim just as well as radioactive materials or explosives.

[K. A. Pital]

Well, an airplane would have incredible lethality rate and the submarine would likely be irreversibly damaged by accidental radiation release, but outside of those examples, is testing really so deadly?

A powerful internal combustion engine from 200X shipped to early XX century would lead to a great multitude of ruined engines, right, until they get the materials right or at least close - and they'd probably never get the same performance as the future technology does, but they will get some advancement. Some chemical compounds could be analyzed by breaking down the material, so they have some shot at the simpler chemistry as well, if they get modern materials.

And imagine them getting a stock of several medium-sized missiles. Yeah, they would have a hard time making absolute analogues, but it would be a very major boost to rocket science. After all, first rockets already took off in the early XX century, so they would get clues to the shapes, aerodynamics of missiles quite ahead of time.

What is undeniable is that some things will be always out of reach at the time of impact, and second, that they will waste lots and lots of test examples for trials.

But the technology in question would not necessarily kill them in any given event.

[Starglider]

A powerful internal combustion engine from 200X shipped to early XX century would lead to a great multitude of ruined engines, right, until they get the materials right or at least close

One major problem there; essentially all modern engines use computerised engine management units as an integral part of the design. It would be very difficult to retrofit them with a simple electromechanical control system (maybe impossible, I'm not sure). So a direct copy would not work, you'd be looking for general principles. Even those are mostly reliant on either material strength/durability, precision, or simply a large number of parts that would be impractical to mass produce with the factories of the time.

Missiles have the same problem; the guidance section is a nearly impenetrable black box (technically they will be able to electron-microscope the chips, but that's a multi-decade reverse engineering effort just to work out the processor architectures). Maybe the propellant formulation will be useful (e.g. AFAIK non-smokey AAMs are a fairly recent development).

I wouldn't give up though; some areas of engineering are largely a long string of incremental advances, and some rely more on brilliant insights (though still with a lot of work to actually realise them). The best potential for reverse engineering comes when it is a brilliant-insight-dependent area, particularly when the 'brilliant insights' are only seperable from the 'plausible but ultimately bad ideas' in hindsight. AFAIK internal combustion engines are in the former class not the later, once you have the basic idea.

[K. A. Pital]

Missiles have the same problem; the guidance section is a nearly impenetrable black box

Quite obviously they would not be able to make sense of the modern guidance systems, but they might make sense of the missile's combustion system, propellants, tank shapes, stuff like fin control mechanics, etc. And we're not saying they would be able to directly copy the modern technology.

Instead, they might get something that is very inferior, especially with their materials - but something still superior to what is being developed at the time, shortening the development cycle - possibly by decades. If they get the general principles of rocket science, from engines to aerodynamics, from several actually working examples - there would be not only a far greater acceptance of rocket science as a necessary tool for military purposes, which is a good way to get lots of money, but also a sound basis from which to work.

No guidance, of course, but even without it, that's quite a lot of progress.

And I believed that rather powerful combustion engines were developed in aviation already in the 1930s-1940s, so they are not so behind to make sense of the machinery, even if they would lack computerized controls. Even if it propels them to experiment with 1930s-1940s comparable levels of power in the 1900s, that's already something.

[Starglider]

but they might make sense of the missile's combustion system

Rocket chamber shapes are something of a black art and we do benefit from a lot from computer assisted design, so this is potentially very useful as long as the operating conditions of the rockets they want to build are quite close to those of the missiles.

propellants

The chemistry of liquid propellants is generally so simple that 20th century engineers aren't going to learn anything from it. Solid propellants, maybe a little, but only for optimisations such as the smokeless AAMs I mentioned.

tank shapes

Is this really particularly sophisticated? Isn't a cylinder with spherical endcaps pretty much the standard practical shape?

stuff like fin control mechanics

Again, servos aren't all that complicated, they're just lighweight electric motors (or hydralics for big missiles). The sophistication is the drive electronics again.

Instead, they might get something that is very inferior, especially with their materials - but something still superior to what is being developed at the time, shortening the development cycle - possibly by decades.

Well unfortunately we can't get a good estimate on this, because AFAIK there is no real life example of it occuring. We can say how easy or difficult (usually very difficult) it would be for scientists of period X to reproduce modern device Y, or how likely they are to blow themselves up playing with it. There is no real way to know what the impact would be at the conceptual level, over decades or more.

If they get the general principles of rocket science, from engines to aerodynamics, from several actually working examples - there would be not only a far greater acceptance of rocket science as a necessary tool for military purposes, which is a good way to get lots of money, but also a sound basis from which to work.

Frankly rocketry got rushed into military use pretty much as soon as it became viable and appropriate (before, in some cases). Any side could have deployed Katyushas in WWI, but it would have been industrially and logistically inefficient in long-term static warfare with an ample supply of heavy guns.

And I believed that rather powerful combustion engines were developed in aviation already in the 1930s-1940s, so they are not so behind to make sense of the machinery

Ah but the primary drivers in engine design since the 1960s have been fuel efficiency and pollution control, with power-to-weight as a secondary concern. Furthermore aircraft engines were able to use technology such as turbocharging and fuel injection much earlier than car engines because the high unit cost imposed by these features was acceptable to the military. Having the engine from a contemporary civillian vehicle would pretty much confirm what the engineers of the 1930s would expect; these high-end semi-experimental features would eventually make it into cheap mass production engines. I doubt it would give them much head start on that process though.

Even if it propels them to experiment with 1930s-1940s comparable levels of power in the 1900s, that's already something.

The problem is that if you go far enough back for there to be real gains by skipping the 'highly experimental' stage, you are removing more of the reference points needed to understand the design. Futhermore the materials and tolerances would not be able to take the (relatively) high compression ratios of the modern engine. Without any experimental baseline, initial attempts to replicate fuel injection will not go well, though you might still accelerate the development of working FI engines by a decade or so.

For a contrasting example from my field, you could send a paper or two on SVMs (support vector machines; a very popular machine learning technique) back to 1975 and they'd be able to use it immediately (albeit on relatively small data sets). Indeed I wish someone would, because then we might've avoided some of the hype over neural networks (which usually perform worse than SVMs while allowing their proponents to pretend that they're not just doing simple function approximation).

[Stuart]

Quite obviously they would not be able to make sense of the modern guidance systems, but they might make sense of the missile's combustion system, propellants, tank shapes, stuff like fin control mechanics, etc. And we're not saying they would be able to directly copy the modern technology.

But none of that is precisely new state of the art. After all, the British Army was using the Congreve rocket as artillery back in the 18th and 19th centuries and quite effective it was when used properly. The Congreves became obsolete when improvements in artillery meant that the same job could be done by regular artillery. The next jump that comes is when materials science moves up to allow components capable of handling conditions in the new designs. This bounces back to teh question I asked earlier, what is the time-window in which a system thrown back in time could be useful?

Instead, they might get something that is very inferior, especially with their materials - but something still superior to what is being developed at the time, shortening the development cycle - possibly by decades. If they get the general principles of rocket science, from engines to aerodynamics, from several actually working examples - there would be not only a far greater acceptance of rocket science as a necessary tool for military purposes, which is a good way to get lots of money, but also a sound basis from which to work.

A lot of these things were already known. As I've pointed out, rockets and their virtues were well-appreciated. What was amiss was that the accuracy problem hadn't been solved and couldn't be until guidance systems were available and they're way out of court. Salvo rocket launchers like the BM-21 were already available (back to Congreves again) but they were inefficient and ineffective compared to tube artillery in the tactical environment that then existed. Again, we come back to the problem that inventions exist within a specific environment and taking them out of that environment deprives them of most of their value. For example, suppose we took anti-tank missiles back to 1917. They'd have a limited value in picking off bunkers and so on but overall, they'd be of very limited value indeed (that begs the question of whether they'd work at all in the conditions of the Western Front. Certainly, the wire-guided ones would have serious problems).

And I believed that rather powerful combustion engines were developed in aviation already in the 1930s-1940s, so they are not so behind to make sense of the machinery, even if they would lack computerized controls. Even if it propels them to experiment with 1930s-1940s comparable levels of power in the 1900s, that's already something.

The problem here is that the new and more powerful engines depended on advanced materials technology. They also depended on advances in lubrication, cooling etc. Now some of that could be translated into 1917 technology but other parts couldn't. Also, those engines depended on production techniques that simply weren't available back then. Just because people could see what had been done didn't tell them much about how it could be done. It's interesting to note, by the way, that power output between 1917 and 1934/35 really didn't change very much as far as production engines was concerned. The top-of-the-line engines in 1917 developed around 400 horsepower, in 1934, it had increased to around 650 horsepower. However, that isn't the whole story. Special-purpose engines were already around that could develop twice that power but they were one-off hand-built entities with lives that were measured in minutes rather than hours. What allowed that class of engine to be developed into production entities was changes in production technology and materials science, none of which will be obvious from simply looking at an engine.

This brings us back to a question I asked earlier. There is a window of opportunity which allows a time-travelled development to have a major effect by reverse engineering. The question is how large is that window? How quickly do we move from "Wow, let's do it" to "ee, we knew it was coming and this proves it but....." to "oh my, what the blue blazes is this?" to "Well, let's try pressing this button - AAARRGGGGGGGGGGGG!" I'd guess that window would vary with the precise technology in question but I doubt if any of them would take more than a few years. For example, take my Corvette back into the 1950s and give it to a GM dealer to run a routine service.

"Ah, a Corvette Sir, yes we can do this. Haven't seen this model before, must be the new 1955 version. Ummm, where's the hood release?"

Stuart presses button and pops the hood. "There you are. Here's the maintenance manual."

Mechanic opens the manual to Page One Step One. "Plug the lead from the computer car status monitoring system into the USB port fitted to the onboard diagnostic system and run engine monitoring program. ??????????????????????????????????????????????????????" Loud crash as mechanic faints.

(Stuart revives mechanic by administering shot of his 25 year old McAllan - which still dates from 25 years plus in the Mechanic's future - after making sure he hadn't damaged the car when his head hit it on the way down). "Look, I'll call Onstar and get you some help."

Mechanic (weakly) "What's Onstar?"

[MKSheppard]

Hey stuart, question from the crypt regarding Soviet Reverse engineering.

A late friend of mine in the CIA (died of cancer seven years ago) once said to me back in 2000:

Until the late 1960s the USSR had a thriving computer industry with innovative hardware designers, programmers, and users. Then the KGB stole essentially everything there was to know about the IBM 360 series of mainframes. The leadership assumed that IBM technology was necessarily superior to the homegrown kind and imposed it wholesale, killing off just about all indigenous innovation. This step more than any other assured the entire Communist bloc permanent cybernetic inferiority right through the Fall. They never really progressed beyond the IBM 360 and 370 series machines, and essentially missed the personal computer revolution until the West wisely began letting them import machines.

How much is that true?

[Darth Wong]

Here's a fun factoid for you: even if there is zero technology gap, it can be easier to design a new device from scratch than it is to reverse-engineer an existing one. I've been in a situation where I had to try to duplicate an existing piece of injection moulding equipment, and it actually took longer than it would have taken to design a new one from scratch.

It's hard to explain, but when you're designing something from scratch, you kind of have a "mental inventory" of what's happening, where everything is, what everything is for, etc. When you're trying to duplicate something that someone else made, you don't have that any more. You're constantly double-checking yourself, going back over previous work in case you missed something or made a false assumption, etc.

[MKSheppard]

I just remembered something. Victorian era! No doubt these people will extensively document every aspect of the new wonder vessel that appeared in their harbor. I wonder if they'd make the correlation between fogging of negatives and dangerous Aether (radiation).

[MKSheppard]

I'm afraid not. They're written in modern English (or Russian or whatever) and use all the jargon we take for granted..

What about the "Bluejacket's Manual"? I'd think that would be a nice place to start from. Also, aggregating all the paperwork and manuals on the boat, and putting them into boxes showing which compartment they were taken from and then studying each one would be a nice start.

*takes out blackboard and writes REACTOR on it.*

"Okay gentlemen, if you find anything referencing a 'reactor' in it's description write it's name under REACTOR."

*a few hours later*

"So, we know 'control rods' whatever those are, are connected with a reactor somehow....So is uh, Uranium -- isn't that used for coloring? Also, why are these people yammering about atoms AND fission? Don't they know that the atom is indivisible?

[Darth Wong]

Not only are manuals written in jargonated language, but blueprints often rely on arcane symbols rather than any comprehensible language at all. Of course, that's assuming you have blueprints.

[aerius]

Not only are manuals written in jargonated language, but blueprints often rely on arcane symbols rather than any comprehensible language at all.

For instance if I presented the board with this schematic, at least 98% of members would be left scratching their heads and have no idea at all of what it is. Only those with electrical engineering background would know what it is and what it does.

[MKSheppard]

Not only are manuals written in jargonated language, but blueprints often rely on arcane symbols rather than any comprehensible language at all. Of course, that's assuming you have blueprints.

Yes, but you can brute force your way through, and assemble context from multiple sources; until you've got a rough sketch of things.

Reading a book that mentiones "the VIRGINIA CLASS uses Mk 48 torpedoes and BGM-109 TLAMs to destroy targets." won't tell you much; you'll be asking "what's a torpedo? Whats a TLAM?" But if you remember what that book said; when you notice the marking on the side of that long cylinder that says "Mk 48 MOD 4", you will go "hey, Mk 48...I've heard that name before....AHA so that's what a torpedo is!"

[Darth Wong]

That depends. I've never seen a set of blueprints which came with a convenient welding symbols guide attached, for example. It is assumed that your welder simply knows what those symbols mean, because he spent years learning his trade.

[MKSheppard]

That depends. I've never seen a set of blueprints which came with a convenient welding symbols guide attached, for example. It is assumed that your welder simply knows what those symbols mean, because he spent years learning his trade.

True, but once you figure out the baseline and scale for a blueprint; you can start figuring out what it means. Remember, the scale/symbols on the blueprint have to be internally consistent with the actual object and within the blueprint itself, otherwise, it's no use.

[Darth Wong]

That depends. I've never seen a set of blueprints which came with a convenient welding symbols guide attached, for example. It is assumed that your welder simply knows what those symbols mean, because he spent years learning his trade.

True, but once you figure out the baseline and scale for a blueprint; you can start figuring out what it means. Remember, the scale/symbols on the blueprint have to be internally consistent with the actual object and within the blueprint itself, otherwise, it's no use.

Not necessarily. You can't necessarily tell everything you need to know about a weld by just looking at it, so you can't necessarily correlate to symbols on the drawing. And what if we go back far enough in time that they don't even have the necessary welding techniques yet?

Or what if there's a tolerance specified on the drawing and they don't have the machining equipment to make it to that tolerance, or the measuring equipment to verify that they're achieved that tolerance? What if the drawing says "case harden to 65 Rc" and they don't know how to do that? What if the drawing says "Heat treat" and fails to specify anything further, because the material in question has a recommended heat treat associated with it? What if it doesn't even bother mentioning the heat treat at all, for the same reason? What if they don't realize that this heat treat warps the piece, and so you have to hold off finish machining until after you heat treat, but you have to do rough machining before the heat treat? Drawings don't specify that; you're assumed to know already unless you're completely incompetent.

What if the blueprint just says "Class-A surface - see CAD data" because the shape is too complex to dimension on paper, and they don't have a computer or the means to translate the data into something they can understand, or the means to manufacture that complex shape anyway?

[MKSheppard]

What if the drawing says "Heat treat" and fails to specify anything further, because the material in question has a recommended heat treat associated with it?

That reminds me about a story involving the Merlin engine. The British only had about 60 or so blueprint sheets for the Merlin in production. When it was redrawn by Packard for mass production to meet US tolerances, the sheet count went up into the hundreds.

Why? The british basically only put the barest minimum of information on each sheet; figuring that the craftsmen at Rolls Royce would be able to interpret the blueprint and figure out the correct tolerances, fit, etc based on their experience.

Packard on the other hand, wanted blueprints that an idiot could read and turn out parts just as identical as that of a highly trained craftsman..

Anyway, you have good points all, I'm just approaching things from a different tack; if you want to reverse engineer or just operate something; you have to understand the vocabulary first.

What is a torpedo? What is a sonar? kind of things....

[Stuart]

That reminds me about a story involving the Merlin engine. The British only had about 60 or so blueprint sheets for the Merlin in production. When it was redrawn by Packard for mass production to meet US tolerances, the sheet count went up into the hundreds. Why? The british basically only put the barest minimum of information on each sheet; figuring that the craftsmen at Rolls Royce would be able to interpret the blueprint and figure out the correct tolerances, fit, etc based on their experience. Packard on the other hand, wanted blueprints that an idiot could read and turn out parts just as identical as that of a highly trained craftsman.

It wasn't just that, there was a much mroe significant factor involved. Mass production in Britain meant something different from mass production in America.

In Britain, mass production meant that the worker stood at a bench and the thing under construction came to him. He reached into a bucket of parts, took one and tried to fit it to the machine. He'd try parts until he got the one that fitted best then, he'd get to work with a file and file it down to fit. Once it did so, he'd install it, send the machine to the next worker and repeat the process. At the end of the day, there would be a few parts left in his bucket that were too far out to be fitted to anything. They'd go to specialized shops who built whatevers out of those out-size parts.

In the United States, mass production meant that the thing under construction would arrive in front of the worker who would reach into his parts bin and take out a part. If it didn't fit, he would throw it into a bucket marked "rejects", take another one and fit that. At the end of the day, the management would take the bucket marked "rejects" and count the number of parts in it. The company that made those parts wouldn't be paid for them. If it was more than a specified percentage of the total delivered, the contract to supply them would be awarded to somebody else. If the part was made internally, the QC staff at the division making them would get fired.

Under System A, if the whatever breaks, the engineer repairing it gets a handful of parts, finds the one that fits best and files to fit. Under system B, spare parts fit.

Classic example, most PPSH SMGs were made under system A. If you want a new magazine, you have to get several and find the one that works. M1 carbines were made under System B - you want a new magazine, you get one. It'll fit.

That's why Packard needed so many more drawings to make a Merlin. Their parts had to fit first time or the equipment couldn't be made on an American production line.

How does it work today? The worker stands at the line and the thing to be built comes to him. He takes a part out of the bin and installs it. If it doesn't fit, he hits a big red button and the entire line comes to a halt. The component supplier (nearly always outsourced now) is liable for the cost of all lost production until the QC staff find out what was wrong and fix it. Faced with those kind of costs, the component suppliers make darned sure that the parts they supply work first time every time - and that means we can delete the worker and replace him with a robot.

[Darth Wong]

What if the drawing says "Heat treat" and fails to specify anything further, because the material in question has a recommended heat treat associated with it?

That reminds me about a story involving the Merlin engine. The British only had about 60 or so blueprint sheets for the Merlin in production. When it was redrawn by Packard for mass production to meet US tolerances, the sheet count went up into the hundreds.

Why? The british basically only put the barest minimum of information on each sheet; figuring that the craftsmen at Rolls Royce would be able to interpret the blueprint and figure out the correct tolerances, fit, etc based on their experience.

Packard on the other hand, wanted blueprints that an idiot could read and turn out parts just as identical as that of a highly trained craftsman.

As Stuart points out in the post below yours, "highly trained craftsmen" are seriously overrated. Whenever I see a car advertisement that says "artisans" or "hand-crafted", I always think that I must be paying for some overpaid person to hand-correct bad manufacturing tolerances. Germans have been notorious for that in the past: turning a deficiency into a marketing campaign.

Anyway, you have good points all, I'm just approaching things from a different tack; if you want to reverse engineer or just operate something; you have to understand the vocabulary first.

What is a torpedo? What is a sonar? kind of things....

Indeed. Imagine trying to make sense of modern diagrams if you don't even know what electricity is. You'd basically have to develop an understanding of an entire new field of scientific study just so you could interpret the diagrams. Even if they came with extraordinarily detailed blueprints and operating manuals, you'd still be SOL. All modern diagrams and operating manuals are made with the implicit understanding that the reader knows what electricity is.

Without that understanding, even the most skilled engineers of the time would stare at even the most detailed diagrams in confusion and say "But ... what ... makes it move?" They could take apart an electric motor and still be asking themselves the same question. A simple $10 solenoid that you'd pick up from a spare parts bin today would have entire teams of scientists banging their heads trying to understand it.

[Darth Wong]

Mind you, you have to go quite a way back if you want to predate the discovery of electricity itself. But it wasn't until the 19th century that people even discovered the concept of batteries or realized that electricity and magnetism had anything to do with each other.