Feasibility discussion: Discworlds semaphores/'Clacks'.

Not sure, but i think that the Syllabary (even if somewhat larger) would be more suited for the logical transformation and doing it at a reasonable pace, than the Latin Alphabet.

China could be a good bet, but so could a Rome that somehow had PoDed away their alphabet and instead used a heavily simplified Linear B (in a similar way that Japanese 'Kana evolved out of the logographic Chinese script)
 
Unless you put a lantern behind them. Requires more complexity for a proper signal but would work.

Not that much complexity. Remember this doesn't have to be automated, just have someone switch out the mirrors for lanterns in the evening. You're going to have to have someone light and monitor them anyway.
 
I've had a bit of thinking, and it strikes me that if this network tries to send a message down a serious distance (London to Liverpool is about 180 miles, which could easily mean 20 or 30 towers in total), then there is a very large risk at human error distorting the message. And the more code you use, the more likely that a change to just one character can completely alter the message's meaning. What you'd want is three observers all reading the signals from the last tower and quickly comparing what they copied down (three because if you have two, and one makes an error, you might not be able to tell which one made a mistake. With three, if one person makes an error, you still have the other two getting it right), before giving the message to the team working the shutters.

Note that this system doesn't have to be fast by our standards, it just needs to be a good deal faster than a galloping horse, about 45 kilometers per hour, or just under 30 mph. A London to Liverpool message on horseback would take about six hours with good roads and assuming neither messenger nor steed stopped for refreshment. Assuming we had to use 30 towers to get that message across by clacks, then the message would need to be received, decoded, checked for errors, then sent off within 12 minutes to be as fast as a horse, probably closer to 5 minutes to warrant investment. This is obviously not a system for long, intricate messages, even 50 characters seriously stretches it, assuming the shutters are moved at 5 seconds a character. Plus you likely need a character limit for messages as a tower can only work on one message at a time. Also you might be looking at a good hundred people or more reading this message, so private information probably shouldn't be sent by clacks. But public data that needs to travel very quickly, like stock market information, shipping arrivals and departings, police alerts for running fugitives etc. could be skimmed down into very short messages indeed. Nobles with too much money and free time on their hands could probably play Chess by clacks with an effective enough system.

Thinking about China, how would they encode the messages? Doesn't Written Chinese contain thousands of characters? With Roman numerals at least you only have to worry about 26 letters and 10 numbers, six shutters could represent 63 different characters, since 63 is 111111 in binary.
 
Last edited:
Thinking about China, how would they encode the messages? Doesn't Written Chinese contain thousands of characters? With Roman numerals at least you only have to worry about 26 letters and 10 numbers, six shutters could represent 63 different characters, since 63 is 111111 in binary.

I think you'd have to adopt a code per character with the limit being the combination capabilities of the machine. If you did nine shutters you wouldn't be able to transmit a book or anything who's code hasn't been prearranged but with several hundred combinations possible you could certainly transmit fairly complex instructions. In fact, for simple messages it might actually be faster than an alphabet code since something like "send help quick" would be three characters rather than thirteen in English.

For government use that probably improves value but for commercial use I think it severely hinders its adoption since you wouldn't be able to do something as simple as give names without it being prearranged with every tower.
 
Last edited:
I think you'd have to adopt a code per character with the limit being the combination capabilities of the machine. If you did nine shutters you wouldn't be able to transmit a book or anything who's code hasn't been prearranged but with several hundred combinations possible you could certainly transmit fairly complex instructions. In fact, for simple messages it might actually be faster than an alphabet code since something like "send help quick" would be three characters rather than thirteen in English.

For government use that probably improves value but for commercial use I think it severely hinders its adoption.

But then we get back to the human error problem. How long will it take to teach an observer every character and then the code for that character, to the point where they could correctly identify every character almost every time? It's 40 characters against hundreds or thousands. And if every character is a word, then errors have a much bigger effect. 'sand hilp wuick' is still legible enough to notice errors and correct them. And if a preindustrial society is trying to build this system, there probably isn't a deluge of literate people willing to spend all day in a cold tower in the sky for what is likely very low wages. It only gets more difficult with a larger alphabet.
 
But then we get back to the human error problem. How long will it take to teach an observer every character and then the code for that character, to the point where they could correctly identify every character almost every time? It's 40 characters against hundreds or thousands. And if every character is a word, then errors have a much bigger effect. 'sand hilp wuick' is still legible enough to notice errors and correct them. And if a preindustrial society is trying to build this system, there probably isn't a deluge of literate people willing to spend all day in a cold tower in the sky for what is likely very low wages. It only gets more difficult with a larger alphabet.

While I agree to an extent, that's already an issue in written Chinese and they seem to do fine. Besides contest is everything. If you got the message "The army should advance on fish" you'd know something was wrong with the message and ask the tower to resend. The only difference between written Chinese and a code is the way it's presented, as lanterns instead of pen strokes. You still have to memorize the code. Literacy was fairly widespread. Finding a few thousand literate people in China willing to become government flunkies and control a signal tower all day would probably be the easiest part.
 
Note that this system doesn't have to be fast by our standards, it just needs to be a good deal faster than a galloping horse, about 45 kilometers per hour, or just under 30 mph.

Alternatively, postal train.

A single train could whip the arse of the entire semaphore network every day for ever.
 

Puzzle

Donor
I've had a bit of thinking, and it strikes me that if this network tries to send a message down a serious distance (London to Liverpool is about 180 miles, which could easily mean 20 or 30 towers in total), then there is a very large risk at human error distorting the message.
Something like parity bits could be implemented which could improve reliability.
 
In bandwidth, yes.

In lag-time, no.

The clacks may be ten to twenty kilometers apart, and it may take as little as five minutes (ass-pull) to receive and transmit a short, priority message.

That's true enough. But that's the only advantage semaphore has - sending a single message over a very long distance, very quickly.

For all your other needs, postal trains can carry tens, if not hundreds of thousands of messages all at once. Not to mention parcels - you can't send that by semaphore... ;)
 
That's true enough. But that's the only advantage semaphore has - sending a single message over a very long distance, very quickly.

For all your other needs, postal trains can carry tens, if not hundreds of thousands of messages all at once. Not to mention parcels - you can't send that by semaphore... ;)

Well, in use,it has the opportunity of providing more 'real time' information. This could mean the difference between victory and defeat in an invasion, having a rapid communications system that the enemy cannot make use of is a hell of a force multiplier.

And a steam locomotive could be hundreds of years away from any potential early invention of the clacks. We know that nuclear fusion would be an amazing power source, but nobody has the needed technology for it, or even knows it is possible economically, so its superiority over other form of energy are academic right now.
 
I'll point out that networkers don't necessarily need to understand the code being sent. You could potentially have the human form of a data packet, where the only things they really need to understand are the source, address, etc. and the remaining patterns are simply copied and carried over to the other tower, until it reaches the destination tower and the code is written down by a transcriber, who hands it to a decipherer, and the message is carried from there.
 
I'll point out that networkers don't necessarily need to understand the code being sent. You could potentially have the human form of a data packet, where the only things they really need to understand are the source, address, etc. and the remaining patterns are simply copied and carried over to the other tower, until it reaches the destination tower and the code is written down by a transcriber, who hands it to a decipherer, and the message is carried from there.

Whilst this would absolutely work for coded messages - allowing you to send very classified information with safety, as long as only the necessary people have the (and it's not like that would ever fail, obviously :p ). However, this does raise the problem of Chinese Whispers again. If it's in code, there's no way for an individual tower to know that the previous tower just made a mistake and ask for a resend. To go back to the London-Liverpool example, all it would take is 1 tower in 30 to muck it up, and your message could arrive as nonsense.


Whilst I like the idea of Babylon using it, I doubt it, same as Egypt - the rivers offer too good a communication network. Forget the steam locomotive, a simple trireme will offer you high 'bandwidth' and speed.

For maximum effect , you need an ancient civilization with a relatively high level of technology, the need for communication towers, a good space to deploy them, a decent amount of 'spare' wealth and time, and probably a reasonable literacy (or ability to raise literacy). Which is a pretty tall order, to be honest. China seems the best bet to me - using the Clacks along the Great Wall for example.
 
Something like this in say the Mongols or other larger empires would allow greater unity and stability and in a place like south america it may allow a western size empire to pop up do it bypassing the lack of horses for comunication.
So if any if these create something like this they would place a great effort in it, has it is nearly has essencial has military power.
 
Not necessarily that complicated. The books make them into sort of proto-computers in their level of complication but doing an array of shuttered lights wouldn't require that much. You could do a series of weighted shutters with lights behind them at for night. It would be hell on the operator's arms but it's certainly doable with levers and a simple pulley system. Say a three man team in the tower at all times. One to watch left and send right, one to watch right and send left, and one to assist and perform maintenance like lubrication or light lamps.

Certainly buildable, but this is not a trivial challenge in terms of technology. You need strong lights, and the structure needs to be big enough so that you can tell their exact configuration even at a distance. The shutters have to work reliably every time, regardless of weather conditions. And you have to build it not once, but many, possible hundreds of times. Most preindustrial civilisations will find it difficult to maintain a system like this. No more difficult than, say, a navy, an artillery park or a major canal network, but probably not less so, either. You have to have a real and quantifiable incentive to build that.
 
Top