Newsgroup Discussions: Building the Clacks

Building the Clacks


Subject: Re: How old is Binky?
Date: 18 Jun 2002
From: Orin Thomas


Although SG could happen both at the time of RM and 100 years before - it wasn't explicitly stated when the two centuries had been merged.

Also, I've been looking closely through ToT for evidence that the events in ToT happen around the time of the events in the majority of the chronicles. There might be a case for setting ToT 100 years after Reaper Man ... Susan is considering the burdens of being eternal (which is kinda odd for someone around 21 years of age) - the trips to Nanny Ogg are clearly happening outside the normal timeline.

I'm reasonably sure that there are no direct date references (other than Death's comment to the Angel) about when the book occurs ( unlike others that say "Carrot had been in the city 2 years" which ties it pretty directly to the events of G!G!).

I'm not sure that Vetinari is in power, that Ridcully is Chancellor, that Vimes is Watch Commander, that Dibbler is about, or that the Librarian is an Orang at the time of ToT as I don't think any of these characters turn up. Of course if anyone can point me to where they do ... I'd be most thankful.

At the moment I have a placeholder for ToT after The Truth and before The Last Hero ... but I'm still not 100% sure that is when it happens.

Date: 19 Jun 2002
From: Nigel Waite

How about the part where Igor is delivered to Jeremy ? Isn't there some comment that Jeremy had become interested in the new Clacks systems because they used some of the same technology as clocks ? This would put ToT at about the same time as T5E and The Truth rather than a hundred years afterwards, when Clacks would hardly be described as a new technology.

Subject: [R] Building the Clacks, was Re: How old is ....
Date: 20 Jun 2002
From: Orin Thomas

I do agree with that about the Clacks - though here is my reservation - Jeremy notes that the cross continent semaphore system is increasingly using clockwork - which does seem to be an advance from the Clacks system described in T5E. This is probably a bit of a straw clutch though. To make my grip tighter on the straw ...

Speaking of the Clacks ...

Since their introduction, I've wondered exactly how long it would take to build a series of Clacks towers right across to Genua from A-M. If you look at the DW Mappe - it is a bloody long way. I'm not sure of the distance in miles - but eyeballing it looks like about 3/4 of the radius of the Disc itself.

According to the DWC, the Disc is 10,000 miles across - which means that a distance of 3750 miles between AM and Genua is in the ballpark. (that's about 6000 KM). The distance between Clacks towers is going to vary of course - but if we assume an average distance between towers of 20 miles (32 KM) [which is probably overly generous] there are about 188 towers between AM and Genua. (and this assumes also a straight line - diversions in Uberwald could make the route well over 4000 miles).

How long does it take to build a tower? I think that there was a comment in T5E that one could be put up in a few weeks. If we assume two months (remember we are talking about a tower that can be seen 20 miles away) average building time (longer for the most remote towers where materials have to be bought in, shorter for the towers near AM) - it would take about 32 years to construct the Clacks route from AM to Genua. Again, this is pretty generous but if you think about how you'd solve the problem today - and then try to do it with AM level technology - it certainly isn't a non-trivial exercise.

It is also fair to assume that the AM-Genua route wouldn't have been the first built. It is a project of immense magnitude. Other cities around the Circle Sea would have been linked up to the Clacks network first (say 10 years to network the Circle Sea cities completely).

So, given all of this, the Clacks aren't probably all that new to citizens of AM. The first Clacks route to Quirm would be novel and exciting - but by the time an AM-Genua route was established "The Clacks" might have been around a few decades.

Which, bringing this all back to my straw clutching hypothesis that ToT could take place 100 years into the future, it could be that Jeremy was commenting on the clockwork aspect of the clacks - a relatively new development - that might have occurred at any time. For the next 500 years they might be referred to as "those new fangled clacks".

Someone also mentioned that AM seemed relatively the same as it did in the rest of the chronicles. Although there has been some "technological progress" in AM, AM will probably always sit on the edge of the industrial revolution. That's up to the author of course - but, aside from the Philosopher's Boat in SG, is it all to likely that we will see Steam Engines and Railroads on the Discworld? Of course I don't know - but my guess is that in 300 years time the citizens of AM will not be sitting down around the television and driving up to the cabin in Lancre for the summer. AM in 300 years time will probably not look all that different to AM "today" (unless of course the Wizards manage to blow it up, or the Alchemists get inquisitive about the possibilities of banging atoms together and find their guild house landing somewhere in XXXX).

Anyway - always fun to think about.

From: Dan Shane

Of course this all assumes that you build the towers in series - if you build them all in parallel (as hard as that may be), it only takes 2 months... It seems more likely if you're building some kind of trunk route that you build as much as you can at the same time precisely to avoid taking 30-odd years to do it - there are always people around willing to be paid to knock down a few trees and make a tower...

From: Peter Ellis

How long does it take to build a tower?
If we assume two months..

Why on Earth? A clacks tower is semaphore. No fancy cabling or infrastructure running between towers, just line-of-sight. All you need is stone to make it stand up, and wood for the flappy bits. If you can build a hut, you can build a clacks tower. In fact, if you've got a disused tall building or convenient rocky outcrop in the right location, you can stick the flappy bits on and have it complete in less than a day.

it would take about 32 years to construct the Clacks route from AM to Genua.

Why are you only building one at once? It would take me about half a day to assemble a computer, test it and get it connected to the Internet. At that rate, expect the introduction of the World Wide Web in ooh, a few zillion years...

Given a sufficiently large workforce and enough funding, you could build the clacks route in... "a few weeks" - however long it took to build the single tower that took longest to build. More likely, you're looking at not more than a few years.

the Clacks aren't probably all that new to citizens of AM.

There was a route all the way to Uberwald in T5E. Clacks had been introduced at most a couple of years previously. Extrapolate from that to how long it *actually* takes them to network far-flung places. You've not got anywhere near 100 years, probably not even near your 32 years. Under a decade, I'd say. Sorry, and all that.

From: Glenn Andrews

This is for doing the system from one city to the next. You could either run city-to-city (most likely) or direct to Genua.

You'd start at both ends. You'd have a dozen teams working from each end. Each clacks construction team would operate as a forwarding post for materials to the next, a place where the cart-horses could be changed and the drivers could get a good nights sleep.

The surveyors would go on ahead. They'd take a bunch of guards on horseback with them, and send messages back to the closest clacks tower construction site by carrier pigeon. You only need to tell those people where to send the next construction crew on the way through.

You'd probably pay the wizards every month or so to do some scrying and make sure the lines were going in the right direction. The surveyors are only a couple of days ride from the last construction site, so you can catch them fairly quickly. The clacks would be operational from.

Depending on the complexity of the clacks transmission gear, you'd either build it on-site, or ship it out from factories in Ankh-Morpork and Genua. The first team, and company employees would go on ahead by coach to the next city, where they'd found an office, and hire labourers to start the run back.

Non-trivial? Sure. Neither was driving roads across Roman Europe, or the canal networks of the industrial revolution. But doable in much less than 20 years. Depends ho much money you had to throw at the project, and according to the books, Once the system took off in Ankh-Morpork, they HAD the money.

From: Terry Pratchett


You start with local traffic within AM and to the nearby towns and cities. When the cash cow is giving milk, you build towers. Down on the plains they'd be mostly wood, and that's just a matter of carpentry. Where there's likely to be serious danger of attack, you build a masonry base.

This is not dragon wizardry - it can be done cheaply by local labour. You use lots of work gangs, and out in the country - it's mostly country - you're playing cheap local rates, with a fast-completion bonus.

You'll need about 300 towers to Genua, max. You don't begin at both ends, you begin everywhere. Every three or five towers or so you have a repair/ maintenance tower with supporting buildings, and you get these done as soon as you can to serve as depots for the others.

Where've you've got even two adjacent towers up and running, you'll be making a bob or two off local traffic (at a nice cheap rate to encourage usage.) Even before all the towers are finished you have guys up there running a minimal service; the key thing is to be earning. That means not a lot of machinery at the start, and you haul that in from one of the cities and anywhere there's a craftsman who can read a blueprint.

When the line is running all the way across the continent many of the towers still aren't complete and the operators are working surrounded by sawdust. Doesn't matter - you're up and running and paying off your debts.

I reckon you could do it in two years, including surveying and land purchases. The sites which could well be dirt cheap - after all, you could be the first king on your block to have a clacks tower and special low rates for royalty. Indeed, nearby towns might bid for towers.

It's amazing how fast gold works.

From: Sherilyn

...and a few hundred dragoons...

You can use gold to turn the clacks dark for a few days, send some bad signals. Gold flows down the clacks, Captain Crunch likes the gold stuff and wields a mean whistle.

From: Terry Pratchett

I know where this is going...

Let's just say that I know how clacks crime will work on DW. And it's quite clever.

From: Speaker-to-Customers

Nope; the length of time it would take to construct the Clacks route from AM to Genua is not 188 x 2 months; it is 2 months plus the time taken to transport the 188 construction crews.

You build them all simultaneously, not one after the other. Or rather you start on the first one at the same time as sending out 187 other construction crews, the first of which start constructing Station 2 as soon as they arrive on site, the other 186 crews move on, and so on.

If you send 94 crews out by ship to Genua, to start there and work backwards, you cut the time down even more. Less than 94 really, as you could allow for the progress on the line out of A-M while the ship was en route to Genua.

Although you don't really need 188 crews, as Crew #1 would move on after finishing Station 1, and start on Station 23 or whatever. Crew #2 would start on Station 24, running a day or two behind, and so on.

I would estimate 2 - 3 years for surveying the route and planning, then another 2 - 3 years for the actual construction.

From: MP

This is a sort of general outline of the clacks from what I can find in the books that mention them.

In T5E, the clacks are obviously very new - the watch is running them in parallel with carrier pigeons, and more importantly, Vetinari hesitates when saying:

"Good. I will have a . . . a clacks sent immediately."
(talking to Sam about his assignment)

This indicates to me that he's not used to the technology yet.

Later in the book, Lady Margalotta is thinking:

The clacks tower had only been up for a month, and it was being roundly denounced throughout Bonk as an intrusion. But it seemed to be doing a good if stealthy local traffic.

Therefore, clacks reached Uberwald whilst it was still new in Ankh - within a year, I'd say. Bear in mind that the route Ankh-Uberwald is heavily frequented with ox carts - materials are easy to move around. Now, later in the same book, Carrot says:

"In a few months they say we'll be able to send messages all the way from Ankh-Morpork to Genua in less than a day!"

Hence Uberwald was connected before Genua - which makes sense since it is en-route between Ankh and Genua. We also know that the towers either side of the Bonk one are up and running. Therefore a tower takes less than a month to build, yet is within firework range of two or three other towers, for security's sake. Fireworks are visible over quite a long distance, especially at night, say 10 miles, on a clear night[1]. Therefore towers are needed every 5 miles or so.

Assume a working tower could be as simple as a pole with a cart next to it (so the travelling teams could keep in contact - you could even have temporary cart based towers!), so takes half a day to put up (you can then build the proper tower after, once you've got a working pole - after all, time is money!)

Then you could build the basic line

"The clacks do a skeleton service as far as Uberwald now, don't they?" - Vimes

very quickly, and upgrade as necessary (possibly prioritising those towers in dangerous zones - like Uberwald). Also this would fit with the notion that

The whole thing would likely be matchsticks after the next hurricane

cheap and fast to build.

By TT, the clacks have reached Lancre - in the opposite direction to Genua, almost. And the fact that they can be

opening offices in Sto Lat and Pseudopolis

indicates they have good reliable connections to these places (expected - they are in the Sto Plains). The only thing we learn from TOT is that they use clockwork to some extent.

[1] I've seen some in Coventry from Rugby - about 10 miles edge to edge.

From: Alec Cawley

Little detail - you don't have to build the towers one-by-one. In fact, you normally wouldn't. A line of towers which ends in open country is useless, so you would tend to send several working parties to build all the towers necessary to reach the next town of any size at once.

During the Napoleonic wars, the Admiralty had a line of semaphores between London and their main base at Portsmouth. I believe that, once given the go-ahead, the line was build over a single summer. For a longer project, how long did it take to build the Trans-Siberian Railway - a much bigger project than a line of clacks towers, because it is a continuous track though uncharted, and pretty hostile, territory instead of a line of points alongside a well-travelled road (i.e. no bridges to build, swamps to cross, forests to fell). I think that only took about 10 years to build. So I could see a line of clacks being put up within about 5 years.

Date: 21 Jun 2002
From: Daibhid Chiennedelh

Let's just say that I know how clacks crime will work on DW.

We've already seen a bit of it, haven't we? The "New Crimes" section of the Thieves Guild Diary, and an interesting bit of cracking in SODII.

From: Orin Thomas

Some further ponderings on The Clacks ...

I'm probably being overly cautious - but perhaps these are issues that might be addressed in later books.

Would there be a "Guild of Clacksmen" - do the Clack's operators need to be able to read the message - or just relay it?

Do they record messages as they go through - or just record that such and such a message passed and it was of such and such a size.

Would there be "error correction" built in? (probably - but it might have to be on a station to station basis as re transmitting a message across the entire continent would be cumbersome).

Are the Clacks owned by one company / individual - or are there a myriad of companies and standards in use across the Disc. One individual or company controlling the transmission of information across the disc definitely brings up interesting possibilities (especially on the Disc where power (except perhaps with Vetinari) tends to corrupt pretty quickly).

From: MP

I'd imagine they'd just need to relay the message on. Think of the little old lady who relayed the Apollo mission when the automated station broke down... On the other hand, there is a guild:

That's why the Guild was driving hell-bent across the mountains on to Genua, four thousand miles away.

Do they record messages as they go through...

It would take more time, unless they had an automated system (you press the buttons and then it both moves the flaps and punches holes, for example.

Would there be "error correction" built in?

Again, more time, but possibly. Semaphore doesn't have any (well, you wave your arms in the international signal for "What was that supposed to be? That flipping great dragon just got in the way") but I seem to remember the Victorians blanked the shutters to indicate repeat.

Are the Clacks owned by one company / individual...

Lots of companies.

It seemed as though everybody who could put together a pole, a couple of gargoyles and some second-hand windmill machinery was in on the business.

They are in a Guild though, so there is still the possibility of abuse of power. Of course, this raises the issue of Guild status outside Ankh-Morpork: what is it? Assassins and Thieves work outside the walls, but have different rules. Virtually all the other guilds are internal only...

From: Robert Shaw

It would take more time, unless they had an automated system

Possible, but we know some of the messages are encrypted, which raises other issues. Until now Discworld cyphers have been able to assume that most messages won't be intercepted. When a code is used over the clacks, every message can be intercepted which makes it much more vulnerable to cracking. Basically, all codes (except the one time pad) which can be encoded by people with just pen and paper would be insecure

The governments will soon be driven to stronger codes, which means coding/decoding machines like Enigma, powered by clockwork. Implementing a public key cipher, such as Leonardo seems to have invented in TFE, needs much more sophisticated machinery. A programmable computer would be best (and Leonardo probably could build a Babbage computer). Hook any clockwork capable of actual computation to the clacks and Hex is likely to become involved, which would probably inconvenience the clacks.

Even now, Hex could use the clacks for extra memory by sending a signal round a circular route. If that makes the clacks part of Hex, and Hex is partly magical, does that make the clacks semi-magical?

Error correcting codes are easy enough to design, and don't require retransmission, though they do slow the message down. If the only legitimate signals are 000 and 111 you can reasonably assume that 010 is an error, and automatically correct it.

From: Vivek Dasmohapatra

Misses the point of ciphering at all, eh? After all, why not just use clear text if you can discount interception? IIRC Vetinari has explicitly assumed interception at least once.

every message can be intercepted which makes it much more vulnerable to cracking.

Not if it's decent crypto, except in the sense of you can't crack it if you don't have it at all. Er. YKWIM.

I refer you to "solitaire" in Cryptonomicon (N Stephenson). Not to mention that codes (as opposed to ciphers) aren't vulnerable to cracking, as such: You must either obtain the codebook, or build up a database of message -> resulting action.

coding/decoding machines like Enigma, powered by clockwork.

Steampunk, woohoo! Also, don't discount magic, imps and hybrid beasties like Hex.

Actually, information is a weird "thing" even on roundworld - presumably it gets to be even stranger on DW.

From: Robert Shaw

Not discount interception totally, just assume it's rare. The more cipher text there is available, the easier it is to crack. A mono alphabetic substitution is secure enough for a five letter message, because you can't do a frequency analysis on five letters.

The Enigma codes are secure on a 100 word message, there isn't enough information there to crack the code, but send a few million words in the Enigma code and cracking becomes trivial, in a technical sense.

The more messages you see in a particular code, the easier it gets to crack. The higher the message volume the more secure the code needs to be and the clacks increase the volume of coded messages.

Not if it's decent crypto...

Even with the best crypto. To take another example, a perfectly random one time pad isn't humanly possible (short of using magic). If the chances of the characters differ by no more than 1/1000, then after 50,000 coded characters there's enough information in the crypto to do a frequency analysis as in 'Cryptonomicon'. If the pad is random to one part in 100 trillion sending a quadrillion or so characters using that pad will allow frequency analysis to crack it.


If you intercept enough of the code you can do a frequency analysis on it. The most common words will be the, a, I ... However codes are more secure than ciphers, but harder to automate. Using obscure or invented languages is more secure still but only Hex has a hope of doing translation. If the message volume gets large enough cypher machines become the only option.

Merchants have a limited vocabulary anyway, which makes codes easier to crack, e.g.

Saffron $3 per ounce
Cinnamon $4 per ounce
Ginger $2.50 per ounce
and so on for the same fifty spices twice a day.

Also, don't discount magic

A mirror like the one in The Last Hero would make reaching Klatch much easier, and wouldn't be stopped by fog.

imps and hybrid beasties like Hex.

Each of which could well have interesting drawbacks

Date: 22 Jun 2002
From: nyra

However, I think on the Discworld cracking an Enigma-type encryption could only be done by Leonard of Quirm and Hex, so they'd be safe enough...

Don't forget that historically, cyphers were rather trivial: spelling everything backwards, simple substitutions (it took me less'n half an hour to "crack" rot-13; but I already had a good idea of how the message began) and the "white noise" encryptions: only every fifth word, only every initial/final letter etc.

Can pure frequency analysis crack a cypher which simply hides its message in perhaps five times as much useless drivel? To make matters worse, you can generate a seemingly meaningful text to prevent interceptors from actually trying to crack your cypher.

It wastes a lot of "bandwidth", but I think such encryptions can already be made fairly safe, as long as the key doesn't get in the wrong hands. And that sort of stuff was old hat in the 17th century.

It could still be spoilt by excessive use, of course.

The most common words will be the, a, I ...

But codes will likely be keywords for key events; if you're mentioning that someone has caught a cold, that means that prices for pepper are down in Genua; different colours mentioned mean differing degrees of success in the secret negotiations over a military alliance with Lancre etc.

Merchants have a limited vocabulary anyway

Encrypting mere price lists wouldn't be very useful, I think. Larger orders to buy or sell based on confidential knowledge of events dictating higher or lower prices would be important, and they'd likely be transmitted with key words. If these keywords and the related events/business activities aren't very obvious, they could remain unnoticed for a long time.

E.g., the code could be exactly five occurrences of the word "and" between the first and the last occurrence of the word "me". Frequency-analyse that.

On the Disc, I suspect that most codes would eventually fail to be safe due to inertia and security holes: for a cypher /code to work, both sides must possess the "keys", and

  1. changing the cypher regularly is a lot of effort

  2. code books/ cypher sheets can get lost or stolen, and until the other side knows that the code/cypher is no longer safe, their messages will be readable for whoever got hold of the keys.

Date: 23 Jun 2002
From: Robert Shaw

There might be a vampire with enough patience, and Vetinari can't be completely certain that no one else has an eccentric genius in their dungeon.

If Leonard designs a decryption machine the plans will leak, perhaps not in Vetinari's lifetime but eventually.

Can pure frequency analysis crack a cypher...

Yes. Mary, Queen of Scots used such a cypher.

It [steganography] could still be spoilt by excessive use

Which is why the Discworld will be shifting to more secure codes than pre-clacks, because the cyphers are getting more use.

But codes will likely be keywords for key events;

True. Codes work well, until you have to send an unanticipated message. Most code books won't have a line for "Klatch has been invaded by talking ducks from the century of the Cobra."

Frequency-analyse that.

It's possible, if you have enough plain text private letters from the same writer. It's not difficult to recognise a writer's unique style. With rather more study they can recognise when the writer has picked a unlikely turn of phrase so they can squeeze in an extra 'and'.

Most users won't be too worried about the risk of decryption because their rivals won't have the resources. Only a few spies and diplomats will need to send a lot of sensitive encrypted messages. They're the only ones who are likely to need better codes than have been customary. I'd expect some of them to start using clockwork coder-decoders similar to Enigma, but nothing more complex.

Anyone else who used one would just be posing.

From: Glenn Andrews

This is the same universe that contains the Lancastrian Peacetime Army Knife, yes? A device designed to do everything except, perhaps be used as a knife?

And a universe from the same author who devised a code book containing

Have found Lost City of Atlantis. High Priest has just won quoits contest?

I think the example you cited would be in any code-book devised by rulers on the Discworld, especially Verence or Vetinari.

From: Daibhid Chiennedelh

According to Nanny Ogg's Cookbook, Discworld naval signals have a code for:

ship invaded by strange creatures in a metal saucer" am abandoning lunch.

From: Terry Pratchett

It is less unreasonable that you think. The ABC Universal Commercial Telegraphic Code contained - and I assume still contains - five letter codes covering everything from 'aeroplanes collided in mid-air' through 'great dam burst, flooding countryside' to the ever popular 'crew are all drunk'.

From: Stuart Painting

I suppose the importance of that depends on (a) whether the code book has a line for "Klatch has been invaded" and (b) whether the recipient much cares who is doing the invading.

To take another example, if Djelibeybi had completely disappeared, it might be sufficient to send a message saying "Now is not a good time to enter into trade negotiations with Djelibeybi".

I can well understand that Vetinari and his ilk would not be satisfied with a fixed code book, and hence would be interested in ciphers that could send any message whatsoever.

With rather more study they can recognise when the writer has picked a unlikely turn of phrase

Except that he is just as likely (I'd be inclined to say more likely) to have achieved the necessary result by inserting or removing entire paragraphs. In other words, he'd be sticking exactly to his ordinary turn of phrase, making it decidedly more difficult to spot.

Mind you, if you're that suspicious it would be a whole lot easier to abduct the writer and apply the thumbscrews.

From: Joerg Ruedenauer

It wastes a lot of "bandwidth"...

Because of the bandwidth-waste, it won't be used for clacks, where every word costs (perhaps even every letter). It would be quite suspicious if the merchant in Genua suddenly sends five pages of text to A-M. Perhaps the secret message couldn't be extracted easily from the text, but it could from the receiver in A-M...

Date: 26 Jun 2002
From: Mr. I. G. Jackson

[referencing The Science of the DW II]

Hook any clockwork capable of actual computation to the clacks and Hex is likely to become involved

Hex has become involved because Ponder Stibbons has seen an opportunity, and it is causing considerable inconvenience, the more so because Mustrum Ridcully has become involved and true to his wizardly principles is at once making sure that a free lunch can be obtained.

does that make the clacks semi-magical?

No, just a large-scale extended input-output device for Hex, though it may be significant that the first message sent by Ridcully is to a lady with a pointy hat. Characteristically, to avoid expense, she sends him a full and sufficient answer in a single word. (Though this may be as an act of charity towards Shawn Ogg.)

What Hex may well be able to do is to extract meaning from both codes and ciphers. We may be in for the ultimate man-machine contest if HEX tries to break Leonard's ciphers.

Up to the Index Page

This section of L-Space is is maintained by esmi and Diane L, who always welcome feedback and suggestions for new additions.
Please send your comments to

The L-Space Web is a creation of The L-Space Librarians
This mirror site is maintained by Colm Buckley