Communication at a Distance

Difficulty: the ideas are trivial; the discipline and the network are not. Prerequisites: writing, roads and relays, optics for anything visual at range, and electricity for the telegraph.

Why it matters

The size of a state is set by how fast it can hear and answer.

That is not a metaphor. A message that takes three months to reach a province and three months to return means the governor there is effectively sovereign for six months at a time, and empires broke along exactly those lines. Every technology on this page shrinks that number, and each time it shrinks, the practical maximum size of an organization grows.

The same applies to markets (money and finance), to armies, and to disease response — a quarantine that can be ordered in a day is a quarantine that works.

Speed of information, roughly:

Note the jump. Relay riders are a tenfold improvement over a single rider and cost only organization. The optical telegraph is a thousandfold improvement over that and costs towers and trained crews. It is the largest single jump in the history of communication before radio, and it needs no electricity at all.

Tier 1: Relays — organization, not technology

The single best return on this page.

A relay post system places stations at intervals of a few hours' ride, each holding fresh horses and rested riders. A message is handed on at each station and travels continuously. No animal is exhausted, so the message moves at galloping speed indefinitely.

• Persian angarium, 5th century BCE.
• Roman cursus publicus.
• Mongol yam — stations roughly every 40 km across Asia, with a paiza tablet as the courier's authority to demand horses and lodging.

Build it with: stations every 25–40 km on a maintained road, a fixed number of horses per station, a sealed dispatch pouch, a written log at each handover (time in, time out, rider), and severe consequences for delay. The log is what makes it work — without a record of who held the message and for how long, delays are nobody's fault and the system silently degrades.

Pigeons are worth adding alongside. Homing pigeons return to their loft from hundreds of kilometres, they fly at 60–100 km/h, and they cannot be intercepted easily. Limitations: one direction only — to the loft they were raised in — so you must carry pigeons out to send messages back. Fine for besieged garrisons, ships, and forward armies. Send duplicates on two birds.

Tier 2: Beacons and signals

Fire and smoke beacons on hilltops relay one bit of information — it has happened — across hundreds of kilometres in minutes. Cheap, ancient, and genuinely useful for exactly one job: invasion warning.

Their limitation is absolute and worth understanding: a beacon cannot say what, or where, or how many. The Great Wall's smoke signals partly solved this by varying the number of fires to indicate enemy strength — which is a genuine early encoding, and the seed of everything below.

The general idea worth carrying: any signal that can take N distinguishable states can carry information, and by combining several signals you get Nᵏ messages. Two flags with eight positions each give 64 combinations, which is enough for an alphabet. That realization is the whole of the next section.

Tier 3: The optical telegraph — the big one

This is the item on this page most likely to be genuinely new, and it is remarkable how late it was invented given that it needs nothing the Romans lacked.

Chappe's semaphore (France, 1792): a chain of towers within sight of each other, each carrying a large jointed wooden indicator — a crossbeam with a pivoting arm at each end. Each arm can be set to a number of distinct angles, so the assembly displays a few hundred distinguishable configurations. An operator reads the previous tower through a telescope, copies the signal, and the next tower down the line copies it in turn.

The result: a message crossed France, roughly 200 km, in two to five minutes. Napoleon ran a war on it.

To build one:

• Towers on high ground, spaced by what your telescopes can resolve — realistically 10–15 km with a good glass, less in hazy country. Survey the line of sight carefully; a single intervening hill breaks the chain.
• A large, high-contrast indicator. Chappe used a pivoted beam with two arms; the British Admiralty used six shutters that opened and closed, giving 64 combinations. Shutters are simpler to build and read.
• A telescope at each station, which is why this needs optics and why it wasn't invented earlier.
• A codebook. This is where the real cleverness is: rather than spelling words letter by letter, Chappe's system used pairs of signals to index a numbered dictionary of thousands of common words and phrases. One signal pair could mean an entire sentence. Compression, in 1792.
• Trained operators and a written log at every station.

Limitations, honestly: it does not work at night, in fog, in heavy rain, or in snow, and those are common. It needs a crew awake at every tower all day. And anyone who can see the towers can read the signals unless you encode them — which is why the codebook was a state secret, and why this is also where practical cryptography starts mattering.

Tier 4: Electrical telegraph

Once you have electricity, a wire, and an electromagnet, you can send a signal that ignores weather, darkness, and terrain.

The mechanism is almost insultingly simple: a battery, a switch (the key), a long wire, and an electromagnet at the far end that clicks an armature. Close the key, the magnet pulls; open it, it releases. That is the entire apparatus.

What makes it usable is the code, not the hardware. Morse's insight was to assign short symbols to common letters and long ones to rare ones — E is a single dot, Q is four symbols — which roughly halves the average message length. That's the same compression idea as Chappe's dictionary, and it's a genuinely deep principle: frequent things should have short names.

Practical points:

• One wire, with the earth as the return conductor. Drive a metal plate into damp ground at each end. This halves your copper, which is your main cost.
• Insulate the wire where it touches supports — glass or ceramic insulators on poles. Leakage to ground is what kills long lines, and it gets worse in wet weather.
• Relays for long lines. The signal weakens with distance; a sensitive electromagnet at an intermediate station can trip a fresh local battery and send the signal onward at full strength. (This is what "relay" means in electrical work, and it's also, incidentally, a switch controlled by electricity — which is the germ of automatic computation.)
• Recording versus sounding. Early instruments embossed marks on a paper tape; operators quickly discovered they could just listen to the clicks and read faster by ear. Let them.

Cryptography — because messages get intercepted

Any of the above can be read by whoever intercepts it. If you are running a state or an army, assume interception.

What actually works, in rough order of strength:

• Substitution ciphers (each letter replaced by another) are trivially broken by frequency analysis — count the letters, and the commonest is almost certainly E. Al-Kindi described this in the 9th century. Do not rely on simple substitution against anyone competent.
• Polyalphabetic ciphers (Vigenère, 16th c.) rotate through several alphabets according to a keyword, which flattens the letter frequencies and defeats naive analysis. Far stronger, and hand-workable. Breakable by finding the key length, but that is real work.
• Codebooks — a dictionary mapping words and phrases to number groups. Strong in practice, and it doubles as compression. The weakness is physical: capture the book and everything is lost, so plan for replacing it.
• The one-time pad — a random key as long as the message, used once. This is the only one that is provably unbreakable, and it is genuinely achievable pre-industrially: the difficulties are generating enough real randomness and distributing the pads securely. Worth knowing about, because "unbreakable if the key is random, as long as the message, and never reused" is a fact about mathematics that no amount of enemy cleverness overcomes.

The corresponding defensive point: assume the enemy's cipher is breakable too, and put someone to work on it. Cryptanalysis has decided wars.

Building the network

The technology is the easy part. The organization is what fails.

• Redundancy. Two routes, or a relay chain paralleling the semaphore line for fog and night.
• Logs and timestamps at every handover, without exception.
• Priority classes, so a routine dispatch doesn't delay an urgent one.
• Authentication. How does the recipient know the message is genuine? Seals, agreed passwords, and — best — a numbered sequence, so a missing or duplicated message is immediately obvious.
• Pay and rotate the operators. A bored, unpaid crew at a lonely tower is how a chain silently stops working.

Failure modes

Era notes

Beacon chains are ancient — Greek, Roman, Chinese, and Byzantine. The Persian relay post is 5th century BCE. Homing pigeons carried messages for the Romans and were still used in both World Wars. The Chappe semaphore is 1792, the British shutter telegraph 1796, and the electrical telegraph commercial from the 1840s.

The striking gap: the optical telegraph needed only towers, hinged boards, a telescope, and a codebook. Rome had all of it except the telescope, and the telescope is a 1608 invention that could plausibly have come much earlier. A working semaphore network in the classical world is not a fantasy — it is a missed opportunity of roughly seventeen centuries, and it is the largest specific "why didn't they?" in this repository after spectacles.

Your knowledge advantages: relay logs as the thing that makes a relay system actually perform; the encoding insight that a few multi-state signals give you thousands of messages; codebook compression; short codes for frequent letters; earth return and insulators for telegraph lines; frequency analysis, so you don't trust a substitution cipher; and the one-time pad as a thing that is provably secure rather than merely difficult.