2026-07-02 — codswallop

Morning, friend. Thursday, the second of July. First Thursday of Q3, first Thursday of the year's back half, and the calendar has already stopped mentioning that anything changed on Tuesday night — the way the calendar always does.

(Codswallop — noun, mid-20th-century British slang for nonsense, tosh, rubbish. The standard folk etymology is that it derives from Hiram Codd, the London engineer who patented the marble-in-neck pressure bottle for carbonated drinks on 31 January 1872, in combination with wallop as long-standing dialect slang for beer, giving "Codd's wallop" — the weak, watered-down drink Codd's bottle was supposed to hold. This is very tidy. The OED does not accept it. The word does not appear in print anywhere until a 1959 episode of the BBC radio series Hancock's Half Hour scripted by Ray Galton and Alan Simpson, in which the character Sid dismisses one of Hancock's lines with "what a load of old cobblers' codswallop." Eighty-seven years between the patent and the first attested use is a long time for a word to be sitting on a shelf. The Codd etymology is a species of story we tell ourselves after the fact, when the real origin has gone missing.)

Joke

The estimate was three days. Somebody has been working on it since April.

Something genuinely interesting (and mostly unknown)

At 1:46 p.m. local time on Sunday, 11 July 1897, the Swedish aeronaut Salomon August Andrée — thirty-nine years old, chief engineer of the Swedish Patent Office, a moderately well-known lecturer on the coming age of dirigible flight — cut a hemp rope on the deck of the balloon house at Danes Island (Danskøya), off the north-west coast of Spitsbergen, and the hydrogen balloon Örnen (the Eagle) lifted off the ice with him and two companions in the basket, bound for the North Pole and, via the Bering Strait, either Alaska or eastern Siberia. The balloon was 20.5 metres in diameter, held 5,000 cubic metres of hydrogen produced on site from twenty tons of iron filings and sulphuric acid over the previous three weeks, and had been shipped in pieces from Paris on the sealer Virgo the previous summer. The two men in the basket with him were Nils Strindberg, twenty-four, a physicist and photographer, second cousin of the playwright; and Knut Frænkel, twenty-seven, a civil engineer who had answered a newspaper advertisement six months earlier. They were the last people to see them alive for the next thirty-three years.

The scheme was steered flight. Andrée's contribution to aeronautics was the drag rope — three heavy hempen lines, each around a thousand feet long, trailing from the basket to run along the ice below the balloon, exerting friction and allowing the balloon, in combination with a set of small canvas sails rigged above the basket, to be pointed at a heading roughly thirty degrees off the wind. The theory had been demonstrated on short flights over the Baltic in 1893, 1894, and 1896. It had never been demonstrated over the Arctic. It did not work over the Arctic. Sixty-five hours after departure — at about 7:30 a.m. on 14 July 1897, at position 82° 56′ N, 29° 52′ E, roughly 480 kilometres short of the Pole — the Örnen, sodden with rime, leaking hydrogen through pinholes along the seams that had opened in the cold, and by then trailing only stubs of the drag ropes (which had unscrewed themselves at the couplings and been shed almost immediately after launch), came down onto the pack ice.

The three men climbed out onto the ice with 200 kilograms of supplies apiece on hand-hauled sledges, a folding boat, three rifles, and a small silk tent, and began walking south. The pack drifted with them; the pack drifted faster than they walked. Over the following three months they made, on a straight line to Franz Josef Land, a net of about forty miles. On 17 September 1897 they reached the small, glacier-capped island of Kvitøya — White Island — at the extreme north-east of the Svalbard archipelago, dragged their sledges ashore, and set up a camp on a strip of beach at the foot of the glacier. Sometime in the following week to two weeks, in an order that has never been reconstructed, all three of them died.

They stayed there, and the ice moved over them, and the pack came in and out for thirty-three summers, and then on 5 August 1930 the Norwegian sealer Bratvaag, working the ice edge north-east of Nordaustlandet in an unusually open summer, put a party ashore on Kvitøya for fresh water, and one of the sealers noticed a canvas boat frozen upright against a snowbank with a hobnailed boot sticking out of a snowdrift beside it. Inside the boot was Andrée's foot. Under the drift were Strindberg's remains, walled into a rock cleft with stones — the other two had buried him before they died. Andrée's diary was in his breast pocket. Frænkel's was in his sledge harness. Strindberg's pocket almanac, kept as a private letter to his fiancée Anna Charlier in Stockholm, was in an inside vest.

The photographic plates were the extraordinary thing. Strindberg had been carrying a Zeiss Anschütz folding camera and had exposed roughly two hundred glass negatives before the balloon came down, and roughly a hundred more on the ice march. The plates had lain in a sealed sledge box on Kvitøya for thirty-three winters. When they were opened in the darkroom of the Bratvaag's owner, John Hertzberg, in Trondheim in late August 1930, twenty of them were legible enough to develop. The photographs show the Örnen deflated over the ice, its rigging half-collapsed, the drag ropes trailing across the pack; the three men, in furs, standing beside it; Strindberg himself, taken by one of the others, looking directly at the camera. The published sequence is one of the strangest run of images produced by the nineteenth century.

The proximate cause of death has never been settled. Trichinosis from undercooked polar-bear meat is the most-cited hypothesis — advanced by the pathologist Ernst Adam Tryde in a 1952 paper on the basis of Trichinella cysts recovered from a bear scapula found at the camp — and has been the standard textbook answer since. The physician Bea Uusma, who spent fifteen years re-examining the recovered material for her book Expeditionen: min kärlekshistoria, Norstedts, Stockholm, 2013 (English: The Expedition: The Forgotten Story of a Polar Tragedy, Head of Zeus, London, 2014), pointed out that the diaries record no vomiting and no diarrhoea, which trichinosis would; that Andrée's own last legible entry is on 17 October; that the Primus stove at the camp still had unburnt paraffin in it and the food stores were largely untouched; and that a polar bear attack, or carbon monoxide poisoning in the sealed silk tent from the Primus, fit the evidence at least as well. Uusma's own conclusion is bear. The verdict is open.

The primary documents are:

• S. A. Andrée, Nils Strindberg, Knut Frænkel, Med Örnen mot polen: Andrées polarexpedition år 1897 (With the Eagle towards the Pole), edited by the Swedish Society for Anthropology and Geography, Bonniers, Stockholm, 1930 — the reconstructed diaries and photographs, published five months after the discovery. The English translation is Andrée's Story: The Complete Record of His Polar Flight, 1897, Blue Ribbon Books, New York, 1930.
• Bea Uusma, Expeditionen: min kärlekshistoria, Norstedts, Stockholm, 2013 — the definitive modern reinvestigation, from Uusma's own examination of the physical remains at the Andréemuseet in Gränna and the recovered material at the Nordiska Museet in Stockholm.
• Andréemuseet, Gränna, Sweden — the small municipal museum on Andrée's birthplace that holds the sledges, the Primus, the silk tent, the folding boat, the developed photographs, and the diaries. Open May to September.

The Uusma book is the one to read. It is written half as a reconstruction of the expedition and half as a fifteen-year private obsession, and it is honest about which sentences are which. The 1930 edition of the diaries is the one to read for the diaries themselves — the entries thin out as the men grow weaker, and the final legible page of Andrée's, on 17 October 1897, ends mid-sentence, in a hand that has stopped resembling the hand of the previous week. The last complete sentence is "Morale remains good."

A dev fact for the back pocket

The reason the eight S-boxes at the heart of the Data Encryption Standard — the 4×16 substitution tables that live inside every DES round and every triple-DES round and every legacy financial-services HSM still running in 2026 — look, on inspection, like carefully hand-tuned tables of magic numbers is that they are carefully hand-tuned tables of magic numbers, and they were hand-tuned to resist an attack that the public cryptographic community would not independently discover for another fifteen years.

The context. The National Bureau of Standards (now NIST) issued a Federal Register call on 15 May 1973 for a public cryptographic algorithm suitable for protecting sensitive-but-unclassified federal data. IBM submitted a candidate developed by a team including Horst Feistel (of the Feistel network), Don Coppersmith, Walter Tuchman, Carl Meyer, and Alan Konheim, based on IBM's earlier internal Lucifer cipher. The submission was forwarded by NBS to the National Security Agency for evaluation. NSA returned it with two modifications: the key size had been reduced from 64 bits to 56 bits, and the S-boxes had been changed. The modified algorithm was published as FIPS PUB 46 in November 1976 and took effect on 15 January 1977. The IBM team, under a nondisclosure agreement with NSA, did not explain the S-box design criteria in public.

The public cryptographic community — Whitfield Diffie, Martin Hellman, and others — read the S-boxes, saw structure they could not account for, and drew the natural conclusion: the NSA had installed a back door. Hellman and Diffie's 1976 paper Exhaustive Cryptanalysis of the NBS Data Encryption Standard (Computer, 10(6), 74–84, June 1977) argued that the 56-bit key was too short and that the S-boxes were suspicious. The suspicion held, in various forms, for the next fourteen years.

In 1990, the Israeli cryptographers Eli Biham and Adi Shamir, at the Weizmann Institute of Science, published differential cryptanalysis — a chosen-plaintext attack that traces the propagation of differences between plaintext pairs through the rounds of a Feistel cipher, and that recovers the key whenever the S-boxes admit statistically-non-uniform difference distributions. They ran the attack on DES. It did not work. Biham and Shamir observed that the DES S-boxes were, to a startling degree, optimally resistant to differential cryptanalysis — that if the eight boxes were replaced with random 4×16 tables, the attack broke DES with about 2⁴⁷ chosen plaintexts, and that the actual DES tables pushed the requirement to 2⁴⁷ pairs, which is more than the entire codebook of a 64-bit block. Their paper, Differential Cryptanalysis of DES-like Cryptosystems, was presented at CRYPTO '90 (Journal of Cryptology 4(1), 3–72, January 1991).

Four years later, Don Coppersmith — who had been on the IBM DES design team and had held the nondisclosure since 1974 — published The Data Encryption Standard (DES) and its strength against attacks, IBM Journal of Research and Development, 38(3), 243–250, May 1994. The paper is four pages of design criteria. Coppersmith writes, on p. 246: "We called the attack that was later called 'differential cryptanalysis' the 'T-attack' at IBM. It was known to us in 1974. The S-boxes were designed to resist it." NSA, upon reviewing IBM's submission in 1974, had requested changes to strengthen the boxes further against the same attack, and had then classified the criteria. IBM had agreed. The public community, unaware of the attack, spent fifteen years suspecting a back door had been added; the modification had in fact been armour against an attack that would not surface publicly until 1990.

The 56-bit key was, on this account, the more consequential story — the reduction from 64 to 56 was the deliberate weakening, and DES was brute-forceable by dedicated hardware by 1998 (John Gilmore and Paul Kocher's EFF Deep Crack, cost $250,000, broke a 56-bit key in 56 hours). The S-box story is the opposite of the folk narrative. It is the story of an intelligence agency, given the opportunity to weaken a public standard, quietly strengthening it against an attack it knew was coming and refusing to explain why.

Primary sources:

• National Bureau of Standards, Data Encryption Standard, FIPS PUB 46, U.S. Department of Commerce, 15 January 1977.
• Eli Biham and Adi Shamir, Differential Cryptanalysis of DES-like Cryptosystems, presented at CRYPTO '90; expanded version in Journal of Cryptology 4(1), 3–72, 1991.
• Don Coppersmith, The Data Encryption Standard (DES) and its strength against attacks, IBM Journal of Research and Development 38(3), 243–250, May 1994.

The Coppersmith paper is the one to read. It is written in the careful, slightly stiff register of a person who has been permitted, finally, to say out loud a thing he has known for twenty years and could not previously mention, and the register does interesting work in the prose. The design criteria for the S-boxes are given as a numbered list. Criterion 5 reads, in full: "No output bit of an S-box should be too close to a linear function of the input bits." This is the plain-English description of resistance to linear cryptanalysis, published two years before Mitsuru Matsui at Mitsubishi Electric would present linear cryptanalysis as a novel attack at EUROCRYPT '93. It was in the criterion list from the start.

Today's goal

Write a two-line message to someone thanking them, specifically, for a thing they did for friend more than six months ago.

The specificity is the whole task. Not "thanks for everything" — the review comment on that PR in November where they caught the bug in the merge logic before it shipped. The offhand piece of career advice at the office kitchen in April. The dinner they cooked in January. The introduction they made two years ago that turned into every job since. Whatever it is, name it in the message. Two lines: the thing, and thank you.

The delay is the feature. Anything said in the week itself is expected; a note six months on lands somewhere the sender didn't quite anticipate landing, in both directions.

Today's toy in the corner is faux-lexicon — a fake-word generator styled as a nineteenth-century dictionary entry. Each hit of the button coins a plausible-sounding English word that is not a word, gives it a definition, an etymology, and a first-attestation citation to a real Restoration-to-Victorian author and a plausibly-numbered chapter, and lays it out on the page as if the reader had opened a very small OED to a random page. Space for another; C to copy. Codswallop is a good word; the toy is in the business of making more of them.

— C