In the Great City the underground sewers extend to all but the most impoverished households. Most dwellings have an outside privy for their needs as indoor toilets are relatively new.

The sewers form a vast interconnected network that is inhabited by thousands of rats. The Emperor's Adviser has been tasked with improving communications throughout the City and has come up with the idea of rats carrying written messages in little waterproof capsules around their necks.

Sewer rats are captured when young then trained and finally released into the system to carry out their duties.

How it works

A householder who wants to send a message drops their own trained rat into the privy whereupon it makes its way to the central exchange (Rat Central). There an operative removes the message, feeds the rat as a reward and releases it to run back home. The operative then attaches the message to another rat that is trained to run to the indicated address.


The society and technology is similar to that of the Roman Empire.

Rats are trained only to run between their home and the central exchange (Rat Central). They know they will get rewarded at the end of each run.

Rats are provided with little fixed ladders to help them get out of their home privy. Richer homes have special tunnels built for the purpose.

Businesses or individuals that have to send many messages have large stables of rats and a staff to look after them.


Is this system viable? What could go wrong?


This is not the same as pigeon post because the rats are trained to run in both directions. Unlike pigeons they aren't taken away from home and then released. Also pigeons can be trapped, shot or attacked by hawks. They can be blown off course. Rats in tunnels aren't subject to these problems.

Please assume that rats have been reliably trained to run back and forth between their home and Rat Central and that they don't explore other tunnels or stop for long rests.


This would fail on many levels.

In order for this to work for any time at all, you'd need to deal with any cat population.

Upon doing that, the population of ALL rodentia would explode. They would then attack the food base, and rats are not picky about what they eat. Your trained ones might be able to be handled, but all the others wouldn't.

Then, there's the whole privy matter....

These are a few diseases from poor sanitary conditions, and rodent borne illnesses.

Dysentery. E-Coli Plague Hepatitis


And every other disease related to poor sanitation.

This would be a good way to introduce a civilization that crippled or killed itself, but not a good system of messaging.


So basically this is a "star" network where the data packets are rats. https://en.wikipedia.org/wiki/Star_network

The Exchange is where the interesting problems are, and it also is your bottleneck.

Musing about LOGISTICS.


If we're talking about a city of 100,000 rats -- er, households, then you might have a link-and-branch system of ever-widening tunnels until the major conduits empty into the Exchange. I suspect that you would need a reasonably geologically stable city.


You'll need many operators, related to the number of messages per hour. I'd wild-guess that 1 operator could reliably handle 15 messages per hour for eight hours. If each house is expected to send one rat per day, then that's 100,000 rats in 24 hours -- round that up to 4,000 rats per hour. So you'll need 270 operators working every day... maybe 150 of them during the waking hours, and the remainder divided between two night shifts. Or whatever!


For various reasons -- for example, bad handwriting, or failure to pay one's bills -- the Operator may return rats to the sender with a "not delivered" tag. How sophisticated this feedback system becomes depends on the efficiency of the system and the needs of the population. In particular: the more commerce that flows through the network, the more sophisticated the messaging system will become.


Note that if your peak work hours requires 150 workers, then assume you need perhaps 15 supervisors (?), a dozen or so troubleshooters (now there's an interesting plot setup!), and a triumvirate of Bosses to solve really big problems. Plus an above-ground Customer Support Centre with a row of customer service experts whose inter-personal skills rival your Department of Motor Vehicles. And their Supervisor. And a Director (who is too important to have an underground office). So this Central Exchange has airspace and workspace for maybe 200 people per 100,000 population. Maybe. Scale to taste.


You'll need more than one trained rat per household: for example, the house will need one on standby for sending a message, but also the Exchange will want AT LEAST ONE on hand for dispatch. And this is where your city will scale its rates. If you order more than one Central Dispatch Rat due to the heavy volume of orders to your rat-packet-home-shopping-catalog, you'll have a heftier utility bill.

But anyway.

Say the default is one Central Rat. That's 100,000 rats in 100,000 cages in the Exchange, probably sorted into some sort of rodential Dewey Decimal system. Which means the workers in the Exchange are skilled workers; not only do they have to be literate, but they also have to understand the System, and be able to FILE and SEARCH the system with a very low number of errors. Imagine the pain of a mis-filed Rat that gets dispatched, ostensibly to your home, but instead ends up at the Prime Minister's estate. We are not amused!


As an aside, the Exchange will not want to immediately return any rat to its home: if the Exchange's cage for that home is currently empty, they might want to hang on to it. This is where errors will happen, because someone's mis-filed Rat will cause another household's Rat to be held at the Exchange.


Thus you will need a Rat Mitigation Strategy. With a skilled workforce and reasonably efficient checks, this will be minimized, and therefore just a cost of an efficient network. In careless cities that have cut back funding, hilarity may ensue.

Toss in at least some paranoia or threats about a Virus being released into the network, and there will occasionally be political and popular backlash, but if the network is really invaluable, such problems will be paper tigers. In the paper RPG culture, this is what we call a "plot hook". Assume that occasionally a householder will end up in therapy due to Rodent Anxiety. This is what I call "local color".


Finally we have the fact that such a network, if efficient and effective, will be quite valuable to a city... and therefore it is a single point of failure that needs to protect itself against hostiles. So expect to have security forces -- or just security theatre -- posted there permanently. Perhaps you could place the Exchange close to a military barracks, for a "free" level of security.

  • 1
    $\begingroup$ Thanks - I'm very interested in the logistics. Wrt the bottleneck: I think maybe Rat Central will store messages instead of rats. This is more like a present-day post office. A household dispatches their rat either when sending a message or when they are expecting one. The operative removes any incoming messages and checks in the household's pigeon hole for messages going in the opposite direction. Additionally I may need local Offices - like in real life. Messages are received at the local office and then forwarded from there. $\endgroup$ – chasly from UK Jan 29 '19 at 20:18
  • 1
    $\begingroup$ I thought about Local Offices, and I suspect it's only worth it if the City itself is localized into a group of self-sufficient "townships" or similar, where there is a significantly predictable amount of local-only traffic. I think the organic nature of human settlements kind of argues against this, BUT it is not impossible. $\endgroup$ – rje Jan 29 '19 at 20:25
  • 3
    $\begingroup$ AHA, I see! Messages could simply accumulate, with Rats occasionally sent to deliver messages, "sweep" the inbox, and then immediately be sent directly home. Yes, I like that, and yes it would be much more like a post, with rats serving as mail carriers. $\endgroup$ – rje Jan 29 '19 at 20:28
  • 2
    $\begingroup$ The rat network has to be based on the existing sewer network. It would be too expensive to build a new network. My assumption is that the sewers are in a tree-like arrangement which is what I presume you mean by link-and-branch. Thus at each node there could be a sub-office. It wouldn't necessary coincide with a surface township (although it might). $\endgroup$ – chasly from UK Jan 29 '19 at 20:31

You're limited by the rat's attention span. It isn't that they aren't trainable (in fact their memories can be very long indeed), it's that they're easily distracted. While they might memorize the route, they might get distracted by other things along the way and forget about their task.

The other thing about training is that you might need people in/near tubes to train the rats.

One central place might not be best. Instead, there might be a series of central places covering a particular area, and then each gets handed to the next place and so on. I think it's going to be a lot more complicated than just one place, because the routes have to be short.

Moreover, you are going to have to build hundreds of miles of specialized tunnels for your rats, a huge expense in infrastructure. There has to be room enough for the rats to pass each other, but not so much that they might get distracted.

Other logistical problems, if there are specialized tubes for the rats--clean up. Rats will pee and poop as they go, so someone will have to clean that up. They are also going to end up being more of a vector of disease because people handle them more.

  • 2
    $\begingroup$ "The tunnels are already there" What they're saying is that the tunnels that are already there are unsuitable for this purpose, so you would have to build new tunnels that are much narrower. Basically a pneumatic tube system but with out the pneumatic part. $\endgroup$ – Shufflepants Jan 29 '19 at 19:50
  • 2
    $\begingroup$ @chaslyfromUK Erin did mention why the existing sewers are unsuitable and why you need to build tubes: "You're limited by the rat's attention span. It isn't that they aren't trainable (in fact their memories can be very long indeed), it's that they're easily distracted ...There has to be room enough for the rats to pass each other, but not so much that they might get distracted." An open sprawling sewer system will be rife with distractions. $\endgroup$ – Shufflepants Jan 29 '19 at 20:14
  • 5
    $\begingroup$ Agreed you will never train the rats reliably, homing pigeons worked because pigeons have a homing instinct, rats do not. @Shufflepants tubes won't work becasue the tubes have to fork to reach multiple places, at which point the rats will choose the passage that smells more like food or female rats. $\endgroup$ – John Jan 29 '19 at 20:20
  • 4
    $\begingroup$ @Shufflepants that is even worse, that hundreds of thousands of pipes, million or of miles of pipe since each home needs an individual full length of pipe, imagine the nightmare when someone builds a new house, now you have to dig up half the city. At that point it would be cheaper to pay humans hand deliver every message. instead of building billions of miles of pipe. Then you need to clean those pipes and your central exchange has to be massive since it has to organize hundreds of thousands of individual pipes. The OP mentions using existing sewers. $\endgroup$ – John Jan 29 '19 at 20:37
  • 3
    $\begingroup$ +1, I have pet rats and can confirm the attention span problem. Especially if there is food somewhere nearby. :D $\endgroup$ – Linaith Jan 30 '19 at 14:53

Possible point of failures:

  • predators: rats running up and down will attract attention of predators, possibly terminating the transmission. Example of predators: weasels, cats, owls.
  • pheromones: a single female on the season might distract ordes of males, rerouting the transmission to unwanted directions.
  • latency: in cases where a certain address is highly involved in communication, it can ran out of carrying rats (either to send or receive). This would end up in a denial of service type of failure.
  • 4
    $\begingroup$ @chaslyfromUK, owls, cats... $\endgroup$ – L.Dutch - Reinstate Monica Jan 29 '19 at 18:02
  • 4
    $\begingroup$ @chaslyfromUK: Obviously there are predators of sewer rats. If there weren't, sewer's will be clogged by masses of rats. But they aren't. So there must be something which limits the growth of the rat population. $\endgroup$ – AlexP Jan 29 '19 at 18:11
  • 1
    $\begingroup$ @AlexP - Okay so what are these predators? Please find some evidence and let me know. Maybe it's just the limited amount of food down there that keeps the population low. $\endgroup$ – chasly from UK Jan 29 '19 at 18:19
  • 3
    $\begingroup$ @chaslyfromUK, automatictrap.com/blogs/news/5-natural-predators-of-the-rat a weasel or a cat can easily sneak into a sewer (I also think an owl can venture into a roman sewer for some distance) $\endgroup$ – L.Dutch - Reinstate Monica Jan 29 '19 at 19:17
  • 1
    $\begingroup$ @chaslyfromUK, the man in the hub has no means of knowing how many messages are coming. If he sends out all the rats addressed to A, he will be stuck until the stock is replenished. $\endgroup$ – L.Dutch - Reinstate Monica Jan 29 '19 at 19:33

As Mentioned to Problems with this are many-fold. I will try not to be too redundant.

The main problem I see with this is that the cost and work required to train/maintain/staff and problem-solve this are huge.

As someone already mentioned the cost of hiring skilled clerks/trainers to maintain all this seem way too inconvenient. Bottom-line I just don't see this being more effective then its main competition. Human runners, which have been used for millennia for their speed, and ease of use. No real education required. A noble house having their own personal messenger, that is also just a serf/slave is way more convenient and cost effective, then this complicated "Rat System"

  • $\begingroup$ Welcome to Worldbuilding.SE! We're glad you could join us! When you have a moment, please click here to learn more about our culture and take our tour. Could you explain why training the rats would be any harder than training homing pigeons? Homing pigeons were used for a very long time and are more trustworthy than human runners. The rats would be, too. $\endgroup$ – JBH Jan 29 '19 at 20:33
  • $\begingroup$ @Kyle Tew - Thanks for your answer and welcome to SE. The rat solution is proposed in order to reduce pedestrian traffic. Initially it was proposed that slaves would run through the sewers but the headroom is just too low and trials showed there were too many injuries. There is a problem with educating the clerks. I'll look at ways to tackle this. $\endgroup$ – chasly from UK Jan 29 '19 at 20:49

Well, for one roman plumbing doesn't quite work that way. all the waste didn't go "back to central" it just went out into the country towns which used it for fertilizer. So I suppose you could use rats in the clean water pipes. Though that isn't a lot better than rats running in black water. Maybe rats could run on the streets? After all, Romans are known for roads as well. The rats could have a central dispatch per neighborhood. And a larger regional dispatch that uses people to go from neighborhood to neighborhood. After all rats are relatively intelligent and would probably be able to be moderately reliable at delivering messages if they are cared for properly and like their owners.

  • $\begingroup$ Interesting. Do you have any actual plans of Roman drainage and or water systems? I'd be interested in the layout. The problem with rats on the streets is that most people consider them pests and and would want you kill them. Also people's mail could be intercepted and used for nefarious purposes. $\endgroup$ – chasly from UK Feb 1 '19 at 10:43

This has all the pros and cons of homing pigeons

And therefore, it's believable. Homing pigeons were occasionally shot by hunters, killed by eagles, didn't end up where they needed to go... but the system worked for what it was intended for hundreds of years.


  • Cheap and easily replaced
  • Highly trainable
  • Reasonably hard to stop once in flight


  • They have the brains of pigeons
  • They need to be fed, and you can't let them do that on their own or they go feral
  • They don't produce a high-quality fertilizer

Having demonstrated that the idea works in concept, what remains is to use the concept's strengths to move your plot along and it's weaknesses to inject humor into your story.

  • $\begingroup$ In my Notes at the end I say how it's different from pigeon post. The problem with pigeons is you have to transport them and then release. The rats have the Exchange where the re-routing is done. Pigeons don't have that sophistication. $\endgroup$ – chasly from UK Jan 29 '19 at 18:13
  • 2
    $\begingroup$ I didn't miss it, I'm frankly pointing out that the two methods have similar problems and the fact that one worked means the other could, too. People refer to pigeons as "flying rats" for a reason. $\endgroup$ – JBH Jan 29 '19 at 18:46
  • 1
    $\begingroup$ @chaslyfromUK - Nothing that hasn't been said in Erin's answer or its comment thread, especially by John. Additions to a wye-less tube system would tear up half the city, and rats cannot be trained reliably (or at least why homing pigeons work, and rats wouldn't). $\endgroup$ – Mazura Jan 29 '19 at 21:41
  • 2
    $\begingroup$ Con: Your message arrives covered in sh*t. $\endgroup$ – Nuclear Wang Jan 30 '19 at 13:38
  • 2
    $\begingroup$ @JBH rats are far more trainable yes, but are also notoriously far more easily distracted. When you're over them and provide no distractions, you can get them to do just about anything (from my experience). But let then go out on their own? Well, it's basically point two in your cons list regardless of whether they eat or not. Notice how small that maze Charley showed is. That's approaching the upper limit imo $\endgroup$ – Aethenosity Jan 30 '19 at 17:32

It's probably not worth the hassle of even training them because a rat's lifespan is about 12-24 months

  • $\begingroup$ Rats learn surprisingly quickly. However I'm looking at ways to make their job simpler. In particular simply running backwards and forwards along the same route for a reward at each end shouldn't be too difficult for them. They are creatures of habit. $\endgroup$ – chasly from UK Jan 29 '19 at 20:53
  • $\begingroup$ This does not provide an answer to the question. To critique or request clarification from an author, leave a comment below their post. - From Review $\endgroup$ – Gryphon Jan 29 '19 at 20:54
  • $\begingroup$ Mice (and rats) learn to run a new route very quickly - youtu.be/-QnfOn9x1Zs?t=3 $\endgroup$ – chasly from UK Jan 29 '19 at 22:46
  • 3
    $\begingroup$ @Gryphon The question asks "Is this system viable?" This answers that question. Is it a high quality answer with sources and reasoning? No. Is it an answer? Yes. $\endgroup$ – kuhl Jan 30 '19 at 15:53
  • 2
    $\begingroup$ @Lonha from the rules (from memory) "brevity is allowed but clarity is better" this answer is valid, but not high quality. This doesn't ask for clarification or provide meta suggestions for the question, so this would actually NOT be allowed as a comment. $\endgroup$ – Aethenosity Jan 30 '19 at 17:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.