Dots and dashes turned messages into electricity, but sending faster left a new question: what survives damage? Travel from codes and bandwidth in Washington, Paris, and New York to entropy and error correction in Cambridge and Bell Labs; polynomials and compression in Lexington, Moscow, and Haifa; and near-capacity codes and industrial standards in Brest, Kariya, and Ankara. Four experiments let information shift from ‘meaning’ to distinguishable choice and recoverability.
QUESTION FOR THE ROUTE
Without reading a message’s meaning, can we measure its surprise, add enough redundancy to reverse errors, remove repetition, and identify the maximum rate noise permits?
WHAT THIS RIVER DOES NOT CLAIM
The river is not geography. Distance downstream stands for time passing, and the light turns from dawn to dusk as the centuries go by. The objects by each stele are symbols of the kind of event and of how each century band wrote and calculated; they do not reconstruct any real artefact. The land around each stop sketches the natural geography of the scene’s real place, and an iconic building appears only if it already stood in that year. Each scene keeps its real place and evidence basis; open it on the map to read where it happened. This route does not make ‘the digital’ the invention of one person, year, or laboratory. Morse code is not renamed modern binary, Baudot is not ASCII, Shannon’s existence theorem is not a zero-error promise for every finite transmission, compression ratio is not semantic value, and error correction is not security. The line is an edited comparison among coding, information, repair, compression, and capacity—not a proven single chain of transmission; research, patents, papers, presentations, and industrial application sites remain distinct.
WHAT YOU SEE ON THIS RIVER
- Diagram in the sky
- A signal recovered from noise
- Emblem at the source
- An antenna sending a signal
- The real place around each stop
- Around each stele the land takes on the natural geography of that scene’s real place — sea or lake, plain, hills or mountains, the colour of the ground and its common trees — and, where one defines the place, its landform: a volcano, snow peaks, granite domes, a mesa, dunes, a fjord, islands, a rock hill, a gorge or loess terraces. The water near the stop takes the colour of the real river or sea, and the haze the place’s climate. A small globe on the stele marks where it is, with the route from the previous place. Where a city has an iconic building that already stood in the scene’s year, its schematic silhouette rises behind the stop and is named on the card. The land follows today’s terrain and climate as a sketch and the silhouettes are not measured reconstructions. Between stops the river itself stays symbolic.
- A figure board at every stop
- Each board draws the mathematics of that scene. When the boat arrives, the construction is drawn in and the key result rises in red. The drawings are schematic reconstructions, not historical manuscripts.
- Century bands along the banks
- 1750–1899 · Cast-iron plaque · gas lamps · iron truss bridge · factory chimneys, railway and steam train · steamboats
- 1900–1969 · Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft
- 1970 onward · Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites
Where the century band changes, the boat passes under a bridge of the new band. Villages, mills, factories, pylons and towers stand for the technology of each century, not for any real place or architectural style.
01·1844 CE·Washington DC(basis: Presentation)
Turning a Sentence into Timed Dots, Dashes, and Spaces — Morse and Vail’s Telegraph Line
On the congressionally funded Washington–Baltimore experimental line, Samuel Morse sent a message on 24 May 1844 and Alfred Vail received it in Baltimore. Encoding letters as short and long signals plus spaces let language travel through a physical wire. But code, apparatus, and operation were not Morse’s work alone, and a telegraph code that gave frequent letters short patterns is not the same thing as a modern fixed-length binary code or information theory.
- Pause and ask
- What must become electrical so a sentence on paper can be rebuilt at the far end of a wire?
- How thinking changed
- Instead of transporting letter shapes, encode letters as distinguishable patterns of short signals, long signals, and spaces in time.
- What we cannot claim
- Washington marks the sending demonstration; Vail received in Baltimore. The scene is neither a Morse-only invention nor the birth of modern fixed-length binary coding.
- This place
- Federal funding and the Washington–Baltimore railroad corridor turned an experimental apparatus into a long-distance public demonstration line. (Potomac banks · broadleaf trees · low hills · 38.9°N 77.0°W)
- Figure board
- Dots, dashes and spaces on a paper tape at the wire’s end carry letters as a timed on–off pattern.
- On the river
- A desk holding a written record · 1750–1899 (Cast-iron plaque · gas lamps · iron truss bridge · factory chimneys, railway and steam train · steamboats)
02·1874 CE·Paris(basis: Patent)
Giving Every Character the Same Five Beats So a Machine Can Read It — the Baudot Code
French telegraph engineer Émile Baudot represented characters by combinations of five equal time units in a multiplex system patented in 1874. Moving from a variable human rhythm to fixed-length mechanical selection allowed several senders to share timed slots. Five units provide only thirty-two states, so shifts between letters and figures were necessary; this was neither ASCII nor a completed modern theory of digital communication.
- Pause and ask
- Can machines share exact time slots instead of depending on each operator’s variable hand rhythm?
- How thinking changed
- Move from variable-length signals to fixed-length selection in which every character is represented by five equal units.
- What we cannot claim
- The 1874 first patent and later apparatus improvements are not collapsed into one instant. Baudot’s thirty-two states are neither ASCII nor a general computer encoding.
- This place
- French postal-telegraph administration and Paris patent networks supported apparatus that divided one line into timed slots for several communications. (Seine banks · broadleaf trees · flat basin · 48.9°N 2.4°E · landmark: Notre-Dame de Paris (1250), Dôme des Invalides (1706))
- Figure board
- Every character is a combination of the same five units (2⁵ = 32), and a distributor splits each turn into slots for four senders.
- On the river
- A sealed box · 1750–1899 (Cast-iron plaque · gas lamps · iron truss bridge · factory chimneys, railway and steam train · steamboats)
03·1928 CE·New York(basis: Presentation)
Asking How Many Distinguishable Pulses Fit through Finite Bandwidth — Nyquist
At AT&T, Harry Nyquist analyzed pulse rate, waveform, and bandwidth in a noiseless telegraph channel, making mathematical the reason signals cannot be packed infinitely fast. His 1928 paper extended earlier work from 1924 and became a shoulder for later sampling and communication theory. It did not by itself solve noisy-channel capacity or complete the theorem now called Nyquist–Shannon sampling.
- Pause and ask
- If a wire has finite bandwidth, how many pulses per second can remain distinguishable?
- How thinking changed
- Constrain telegraph speed through a mathematical relation among pulse shape, symbol rate, and frequency band rather than apparatus skill alone.
- What we cannot claim
- A noiseless telegraph result is not renamed noisy-channel capacity or the completed sampling theorem. The pin marks the presentation and New York work context.
- This place
- AT&T’s New York long-distance network and research organization made distortion and bandwidth recurring large-system design problems. (Harbor and Hudson · broadleaf trees · low land · 40.7°N 74.0°W · landmark: Manhattan skyline (1913), Brooklyn Bridge (1883))
- Figure board
- Pulses spread after a band-B line: spaced out they stay distinct, packed too tightly they merge — about 2B per second is the limit.
- On the river
- A desk holding a written record · 1900–1969 (Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft)
04·1928 CE·New York(basis: Presentation)
Taking the Logarithm of Possible Choices So Information Adds — Hartley
Bell System researcher Ralph Hartley proposed a quantity based on the logarithm of the number of distinguishable alternatives, so information from successive choices adds. His engineering model connected bandwidth, time, and available signal levels and became an explicit precursor cited by Shannon in 1948. Its emphasis on equiprobable choices and noiseless distinction did not yet provide average uncertainty for general probability distributions or a complete theory of noisy channels.
- Pause and ask
- Which function makes information add naturally when two sets of distinguishable choices occur in sequence?
- How thinking changed
- Measure transmission by the logarithm of distinguishable alternatives rather than physical signal magnitude, turning multiplication into addition.
- What we cannot claim
- Hartley’s equiprobable distinguishable-state model is an important shoulder for Shannon entropy, not a completed theory of general distributions, noise, or meaning.
- This place
- Bell System telephone and telegraph research created a reason to compare bandwidth, time, and signal levels within one infrastructure problem. (Harbor and Hudson · broadleaf trees · low land · 40.7°N 74.0°W · landmark: Manhattan skyline (1913), Brooklyn Bridge (1883))
- Figure board
- Choosing one of four distinguishable levels per slot gives 4 × 4 × 4 sequences; the logarithm turns this into 3 log 4.
- On the river
- A desk holding a written record · 1900–1969 (Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft)
05·1937 CE·Cambridge, MA(basis: Composition)
Translating the Algebra of True and False into Open and Closed Switches — Shannon’s Thesis
As an MIT master’s student, Claude Shannon showed that Boolean operations on true and false propositions could analyze and simplify relay and switching circuits. Abstract logic became a common language for practical circuit design. The 1937 thesis was not yet the theory of entropy and channel capacity; related independent work by Victor Shestakov and the contributions of device physics, memory, and timing remain outside a lone-inventor story.
- Pause and ask
- Can thousands of relays be designed and simplified as true–false expressions instead of traced wire by wire?
- How thinking changed
- Build a structural translation between propositional algebra and open–closed switch states, turning circuits into symbolic calculations.
- What we cannot claim
- Logic did not become identical to electricity; it became a language for modeling circuit states. Shestakov’s independent work and the physical-device layer remain visible.
- This place
- MIT’s differential-analyzer work, electrical-engineering graduate program, and library let nineteenth-century Boolean algebra meet twentieth-century relay design. (Charles River · broadleaf trees · flat land · 42.4°N 71.1°W · landmark: MIT Great Dome (1916))
- Figure board
- The switch x on both branches is factored out, so a four-switch circuit becomes an equivalent three-switch one.
- On the river
- A desk holding a written record · 1900–1969 (Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft)
06·1948 CE·Bell Labs(basis: Presentation)
Setting Meaning Aside to Define Entropy and Channel Capacity — Shannon
Shannon’s two-part A Mathematical Theory of Communication linked a probabilistic source’s average uncertainty to entropy and a noisy channel’s maximum reliable rate to capacity. Abstracting away meaning made telephony, telegraphy, and radio comparable in one language. The theorem says that suitable sufficiently long codes can make error probability arbitrarily small below capacity; it neither guarantees zero error for every finite message nor automatically supplies a practical code.
- Pause and ask
- Without knowing message meaning, can one mathematics describe source uncertainty and a noisy channel’s maximum reliable rate?
- How thinking changed
- Move beyond performance tables for particular devices by abstracting a probabilistic source, encoder, channel, and decoder and defining entropy and capacity.
- What we cannot claim
- Existence of long codes with arbitrarily small error below R<C does not mean zero error for every finite message, one capacity for every channel, or an immediately practical code.
- This place
- Bell Labs’ telephone, telegraph, radio, and wartime communication problems plus its mathematical-engineering community supplied a testbed for a general theory. (Woods · broadleaf trees · low ridges · 40.7°N 74.4°W)
- Figure board
- Source, encoder, noisy channel, decoder, destination: entropy H measures the source, and below capacity (R < C) errors can be made small.
- On the river
- A desk holding a written record · 1900–1969 (Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft)
07·1950 CE·Bell Labs(basis: Presentation)
Using Overlapping Parity Checks to Point at One Wrong Position — Hamming Codes
Richard Hamming began from punched-card errors that stopped computing jobs and published codes that add structured check bits, allowing a syndrome to locate an error. Hamming(7,4) sends four data bits in seven and can correct a single flipped bit. Redundancy became a coordinate system for repair, but the basic code does not correct an arbitrary number of errors or appear unchanged in every storage device.
- Pause and ask
- If one of seven received bits flips, can its position be found and repaired without retransmission?
- How thinking changed
- Turn redundancy from mere repetition into overlapping parity checks whose failure pattern acts as coordinates for the error position.
- What we cannot claim
- Basic Hamming(7,4) is a single-bit-error-correction model, not a general remedy for multiple errors, bursts, or implementation faults.
- This place
- Bell Labs relay computers and unattended weekend runs repeatedly exposed the cost of a single punched-card error stopping a computation. (Woods · broadleaf trees · low ridges · 40.7°N 74.4°W)
- Figure board
- Four data and three check bits sit in three overlapping circles; when one bit flips, the two failing circles point to its place.
- On the river
- A desk holding a written record · 1900–1969 (Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft)
08·1952 CE·Cambridge, MA(basis: Presentation)
Merging the Two Rarest Symbols to Shorten the Average Code — Huffman
In a term-paper problem for Robert Fano’s MIT course, graduate student David Huffman devised the algorithm that repeatedly merges the two least-probable symbols and published it in 1952. Frequent symbols receive short codewords and rare ones longer words, yielding an optimal prefix code for known individual-symbol probabilities. It is not an absolute optimum for every compression problem, context model, block code, or dictionary scheme.
- Pause and ask
- Can frequent symbols get short codewords while boundaries remain decodable without separators?
- How thinking changed
- Repeatedly merge the two least probable items into a tree, producing minimum average length with no codeword prefixing another.
- What we cannot claim
- It is optimal among prefix codes for known individual-symbol probabilities, not absolutely optimal over context, block, arithmetic, or dictionary coding.
- This place
- MIT’s new information-theory course and term-paper problem formed a compact research setting in which a student turned a limit question into a construction algorithm. (Charles River · broadleaf trees · flat land · 42.4°N 71.1°W · landmark: MIT Great Dome (1916))
- Figure board
- Repeatedly merging the two least likely symbols builds a tree: the common symbol gets 0, rare ones long prefix codes like 1111.
- On the river
- A desk holding a written record · 1900–1969 (Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft)
09·1960 CE·Lexington, MA(basis: Presentation)
Reading a Message as a Finite-Field Polynomial Evaluated at Many Points — Reed and Solomon
MIT Lincoln Laboratory researchers Irving Reed and Gustave Solomon published a code that treats message symbols as a polynomial over a finite field and transmits its values at several points. A low-degree polynomial can be recovered when some values are damaged, forming a strong foundation against burst errors. Their short 1960 paper proposed a new code and theoretical decoding; efficient later decoders and applications in space links, discs, and QR codes did not all arrive fully formed in it.
- Pause and ask
- When several bits fail in a burst, can a message be recovered from the surviving points of one polynomial?
- How thinking changed
- Move from individual-bit parity to finite-field symbols and low-degree polynomial evaluations, turning multiple-symbol errors into algebraic recovery.
- What we cannot claim
- The original code and theoretical decoding are not collapsed with every later efficient algorithm and product application. Military funding remains visible.
- This place
- MIT Lincoln Laboratory in Lexington assembled Cold War air-defense and communication contracts, large computing resources, and mathematical-engineering staff. (Woods · broadleaf and pine · gentle hills · 42.4°N 71.2°W)
- Figure board
- The message is a low-degree polynomial sent as values at seven points; when two are wiped out in a burst, the rest recover the curve.
- On the river
- A desk holding a written record · 1900–1969 (Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft)
10·1960 CE·Cambridge, MA(basis: Composition)
Drawing a Sparse Check Network that Could Approach the Limit — Gallager
Robert Gallager’s 1960 MIT doctoral thesis analyzed low-density parity-check codes built from sparse parity-check matrices and iterative inference. He sought protection whose storage and computation costs grew moderately with code length, but the method was too expensive for contemporary hardware and was largely forgotten for decades. The thesis predates the shortened 1962 paper and 1963 book, and not every LDPC construction automatically achieves capacity.
- Pause and ask
- If parity connections stay sparse as codes grow, can iterative decoding move close to Shannon’s limit at manageable cost?
- How thinking changed
- Replace dense algebra with repeated local messages on a graph where each bit participates in only a few checks.
- What we cannot claim
- The 1960 dissertation is the starting point and remains distinct from the 1962 paper. Contemporary impracticality and later rediscovery are not rewritten as immediate success.
- This place
- MIT’s information-theory research and degree system plus contemporary mainframe access enabled theoretical analysis and limited simulation of new codes. (Charles River · broadleaf trees · flat land · 42.4°N 71.1°W · landmark: MIT Great Dome (1916))
- Figure board
- Each row of a sparse parity-check matrix is one check tied to four bits in a graph; estimates pass back and forth along these few links.
- On the river
- A desk holding a written record · 1900–1969 (Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft)
11·1965 CE·Moscow(basis: Presentation)
Measuring One String by the Shortest Program That Produces It — Kolmogorov
Andrey Kolmogorov’s 1965 paper proposed an algorithmic approach to the information in an individual string: the length of its shortest generating program. It can distinguish a regular string from an incompressible one even when their lengths match. Ray Solomonoff’s earlier work and Gregory Chaitin’s independent development remain visible; universal-machine choices differ only by an additive constant, but the exact complexity is generally uncomputable.
- Pause and ask
- Can the regularity and randomness of one individual string be compared without a probability table?
- How thinking changed
- Move from average source entropy to the shortest program producing one object, linking describability with incompressibility.
- What we cannot claim
- The idea is not made Kolmogorov’s alone; Solomonoff and Chaitin’s independent lineages remain visible. Exact complexity is generally uncomputable.
- This place
- Moscow’s probability and recursive-function community plus a new information-transmission journal offered a venue for an algorithmic approach beside probabilistic ones. (Moskva banks · birch and pine · flat plain · 55.8°N 37.6°E · landmark: Saint Basil’s Cathedral (1561), Spasskaya Tower (1625))
- Figure board
- Of two strings with the same length, the regular one comes from a short program; the random one needs a description as long as itself.
- On the river
- A desk holding a written record · 1900–1969 (Concrete marker · electric streetlights · concrete bridge · pylons, apartment blocks and radio masts · barges · aircraft)
12·1977 CE·Haifa(basis: Main activity)
Growing a Dictionary from Repeated Phrases without a Source Probability Table — Lempel and Ziv
Technion researchers Jacob Ziv and Abraham Lempel published a universal sequential compression algorithm that refers back to previously seen strings, learning a dictionary from repetition without a known source distribution. The 1977 LZ77 method remains distinct from 1978’s LZ78 and later variants such as LZW and DEFLATE. Both authors were on sabbatical at US institutions, so Haifa marks their durable institutional and collaborative base rather than a known writing room.
- Pause and ask
- Without a known source distribution, can repeated phrases be learned while reading and compressed losslessly?
- How thinking changed
- Replace a code precomputed from fixed probabilities with a sequential universal code that points to prior phrases and grows a dictionary.
- What we cannot claim
- LZ77 (1977) and LZ78 (1978) remain distinct. Both authors were on US sabbaticals, so Haifa marks durable affiliation and collaboration, not a known writing room.
- This place
- Technion’s exchange between information theory and computer science sustained a collaboration joining Ziv’s statistical communication background with Lempel’s computing expertise. (Mediterranean coast · pines and olives · Carmel · 32.8°N 35.0°E · landmark: Shrine of the Báb (1953))
- Figure board
- While reading, a phrase already seen earlier is replaced by a (distance, length) pointer, growing a dictionary without a probability table.
- On the river
- A place of ongoing work, marked only by the route’s emblem · 1970 onward (Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites)
13·1993 CE·Brest(basis: Main activity)
Letting Two Decoders Exchange Soft Guesses and Move toward the Limit — Turbo Codes
Research at ENST Bretagne in Brest combined two recursive convolutional codes, an interleaver, and iterative exchange of bit-likelihood information. The performance reported by Berrou, Glavieux, and Thitimajshima at the 1993 ICC in Geneva startled the field by showing a practical code close to the Shannon limit. Research base and conference site remain distinct, and ‘close’ describes performance under particular channel, length, and decoding assumptions—not a proof of exact capacity achievement.
- Pause and ask
- Can two imperfect decoders repeatedly refine each other’s probabilities and move close to Shannon’s limit?
- How thinking changed
- Replace one hard decision with circulating soft information between two interleaved codes, gradually increasing confidence.
- What we cannot claim
- Brest is the research base and Geneva the public presentation site. Near-capacity performance is not exact capacity for every channel and length, nor proof that all three authors had identical roles.
- This place
- At Brest’s telecommunications school, algorithms, digital communication, and integrated-circuit research were close enough to test iterative decoding with hardware in mind. (Atlantic roadstead · broadleaf trees · low hills · 48.4°N 4.5°W)
- Figure board
- Two decoders swap soft estimates through an order-scrambling interleaver; each round the bits’ confidence grows.
- On the river
- A place of ongoing work, marked only by the route’s emblem · 1970 onward (Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites)
14·1994 CE·Kariya(basis: Main activity)
Making a Square That Factory Scanners Can Read Quickly Even When Damaged — QR Code
A DENSO development team responded to factory demand for fast tracking of varied automotive parts with a compact two-dimensional code carrying more data and Japanese characters. Three finder patterns aid scanning from any orientation, while Reed–Solomon error correction tolerates some damage. The result belongs to the team led by Masahiro Hara, manufacturing needs, and standardization; damage recovery does not guarantee a link’s safety, the content’s authenticity, or privacy.
- Pause and ask
- Can a factory mark be found quickly from any orientation and recovered when oil or dirt hides part of it?
- How thinking changed
- Expand a one-dimensional barcode into a surface combining finder patterns, two-dimensional data, and Reed–Solomon error correction.
- What we cannot claim
- The development team, factory, and standards network remain visible alongside Hara. Error correction repairs physical damage; it is not URL safety, content authenticity, or privacy.
- This place
- Kariya’s automotive factories supplied both the concrete demand and repeated testing environment for reading more part data during high-mix production. (Rice-paddy plain · broadleaf trees · flat land · 35.0°N 137.0°E)
- Figure board
- Three corner finder patterns locate the square from any angle, and Reed–Solomon correction restores modules hidden by dirt.
- On the river
- A place of ongoing work, marked only by the route’s emblem · 1970 onward (Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites)
15·2009 CE·Ankara(basis: Composition)
Polarizing Identical Channels into Nearly Perfect and Nearly Useless Paths — Polar Codes
At Bilkent University, Erdal Arıkan proved that recursively combining and splitting identical binary-input channels makes the synthesized channels approach either near-perfect or near-useless extremes at long block lengths. Sending information only through the good channels produced an explicit, efficiently encodable and decodable capacity-achieving family for a broad class of symmetric binary-input memoryless channels. This does not make it best for every channel and short length, nor mean that all of 5G uses only polar codes.
- Pause and ask
- Can many equally noisy channels be combined so some become nearly perfect and others nearly useless on purpose?
- How thinking changed
- Instead of improving each channel a little, recursively polarize reliability and place information bits only on the good synthesized channels.
- What we cannot claim
- The original theorem is asymptotic for symmetric binary-input memoryless channels. It is not best at every finite length and channel, and 5G does not use only polar codes.
- This place
- Bilkent’s information-theory research and an academic lineage through MIT reopened Shannon’s nonconstructive limit as an explicit recursive coding problem. (Dry plateau · steppe scrub · hills · 39.9°N 32.9°E)
- Figure board
- Recursively combining and splitting copies of W polarizes the synthesized channels toward 1 or 0; information rides only on the good ones.
- On the river
- A desk holding a written record · 1970 onward (Glass marker · LED lights · cable-stayed bridge · glass towers, wind turbines and data centres · ferries · satellites)