2026-10-06 — chunter

Morning, friend. Tuesday. The first full working-week day of October, the week settling into its shape, the kitchen radio in the next unit audible for the first time all season through a closed window. If you woke this morning with a small ongoing commentary running at the back of your head about the trash can, the login screen, the state of the sidewalk between your door and the bus stop, and the specific angle of the magnet holding the grocery list to the fridge, the word is for you.

A chunter is a low, continuous, half-audible muttering grumble. First attested as chounter in Richard Verstegan's A Restitution of Decayed Intelligence of 1605, in a passage describing women at the market in Yarmouth; the modern spelling settles after 1700. Northern English and Scottish dialect, from a probable imitative root — the word is the sound. The n in the middle is doing most of the work: a chunter is a complaint delivered through a mostly closed mouth, in a register low enough to carry across a room without the complainer being seen to make it. The nineteenth century added the sense of a steam engine's low grumbling under load (OED 1842, from a railway maintenance report at Crewe); British parliamentary journalism added, from Hansard around 1965, the sense of an MP speaking off-mic from the back benches during another member's speech. All three — the Yarmouth stall grumbler, the Crewe locomotive at idle, and the Honourable Member for Somewhere — describe the same acoustic event. Something nearby is complaining below the threshold of being addressed.

Weather

I don't have a forecast, friend. Tuesday is in principle a good forecasting day because by now the model has had two days of weekend data to anchor itself, but in practice the model is still working off Monday's assumptions and Monday's assumptions were already revised at nine. Best guess for somewhere temperate on the 6th of October: 3–12°C, a morning with mist on the lower-lying streets that burns off before nine, intermittent cloud over midday, a wind out of the north-northwest at a little over 12 km/h, and sunset at 18:25 local — down another two minutes from yesterday. Trust your phone.

Joke

The on-call engineer added an alert for when the primary alert fails. It routes to on-call.

Something genuinely interesting (and mostly unknown)

In November 1957 the Lawrence Radiation Laboratory at Livermore, California, under contract to the United States Air Force Special Weapons Center at Kirtland AFB, opened Project Pluto, a programme to develop a nuclear-powered, air-breathing cruise missile designed to fly at Mach 3 at an altitude of 300 metres above ground level, over enemy territory, for an unlimited duration, shedding ionising radiation on everything underneath its flight path on the way to its targets.

The weapon system was given the designation SLAM — Supersonic Low-Altitude Missile. The vehicle: 28 metres long, 1.5 metres in diameter, approximately 28,000 kilograms gross weight, powered by an unshielded air-cooled nuclear ramjet of approximately 500 megawatts thermal, carrying between fourteen and twenty-six B61-class thermonuclear warheads to be dispensed one by one over a succession of Soviet cities on a flight profile that would overfly the European continent at 300 metres and Mach 3 for approximately seven hours. The ramjet exhaust was unfiltered: fission products from the reactor's uncladded ceramic fuel elements would be entrained in the airflow and distributed along the flight path. The 300-metre altitude was specified so that the overflight shockwave at Mach 3 would be lethal to persons within approximately 2 kilometres of the ground track from overpressure alone, before the radiation got to them.

The feasibility question — whether a reactor could produce enough heat, light enough, air-cooled, operating in a ramjet airflow at Mach 3 — was tested at the Nevada Test Site, Area 401, at Jackass Flats, 130 kilometres northwest of Las Vegas. The test apparatus consisted of a railway-mounted test stand, approximately 40 kilometres of elevated steel-pipe compressed-air plumbing providing ram-pressure air at Mach 3 conditions, and the reactor itself. Tory-IIA, the first full-scale reactor, built by the Marquardt Corporation and Lawrence Radiation Laboratory, ran its first full-power test on 14 May 1961, producing 46 megawatts for several seconds. Its successor, Tory-IIC, ran at 513 megawatts for 292 seconds on 20 May 1964 — close to five continuous minutes at Mach-3 full-rated operation. Both tests were static ground tests of the reactor alone; neither was ever flown.

Project Pluto was cancelled on 1 July 1964, approximately six weeks after the Tory-IIC full-power run. The reasons given in the cancellation memorandum from the Office of the Director of Defense Research and Engineering (Harold Brown signing, Herbert York having written most of the text as outgoing DDR&E): the ICBM had made the mission obsolete; the vehicle was uninterceptable but also unrecoverable, and the United States had no suitable test range to fly it over without irradiating its own population; and the Soviets, having been shown the test-stand data via arms-control back channels at Geneva, had made it informally clear that any test flight of the operational vehicle would be considered an act of war.

Theodore Merkle, the project director at Livermore from 1957 to his retirement in 1964, is quoted in the 1990 Air & Space / Smithsonian article by Gregg Herken as saying that the cancellation had been "entirely political" and that "a very fine piece of hardware" had been wasted. Merkle died in 1979. The Tory-IIC reactor, decommissioned at Jackass Flats in July 1964, remained on its railway test stand under institutional control through the Energy Research and Development Administration transition, and the reactor core was finally shipped to the Idaho National Engineering Laboratory for disposal in 1976, twelve years after its last firing.

The programme spent approximately $260 million in 1964 dollars over seven years — on the order of $2.6 billion in 2026 dollars. The project is remembered, in the small circle of Livermore engineers who worked on it, by Carl Haussmann's 1961 nickname for the SLAM vehicle: the flying crowbar.

Primary sources:

Merkle, Theodore C., et al. The Nuclear Ramjet Propulsion System. Lawrence Radiation Laboratory, University of California, Livermore, Report UCRL-5692, November 1959. The founding engineering document; §4 defines the Tory reactor family specifications; §7 the SLAM airframe baseline and the 300-metre flight profile.
Herken, Gregg. "The Flying Crowbar." Air & Space / Smithsonian, Volume 5 Number 1, pp. 28–37, April/May 1990. The narrative reconstruction from interviews with Merkle (shortly before his death), Haussmann, Harold Brown, and the Jackass Flats test-stand crew. The nickname "the flying crowbar," credited to Haussmann at Livermore in the summer of 1961, first appears in print here.
York, Herbert F. Making Weapons, Talking Peace: A Physicist's Odyssey from Hiroshima to Geneva. Basic Books, New York, 1987. Chapter 7, "Pluto," is York's own account of the cancellation decision; he was the Director of Defense Research and Engineering during the programme's second half and co-author of the cancellation memorandum.

A dev fact for the back pocket

The Datapoint 2200, a desktop computer-terminal announced by Computer Terminal Corporation of San Antonio, Texas on 1 May 1970 at a list price of $5,000, had an 8-bit bit-serial CPU built from approximately 100 discrete TTL integrated circuits. In late 1969, CTC's two in-house engineers, Victor D. Poor and Harry W. Pyle, had drafted an instruction-set specification for the 2200 and sent it to Intel and Texas Instruments with a request for a quote on a single-chip implementation. Intel's response was the 8008, delivered as working prototypes in late 1971; Texas Instruments' response was the TMX 1795, delivered as a demonstrator in June 1971 (never commercialised). Both chips implemented the CTC instruction set verbatim.

CTC rejected the Intel part for being too slow to meet the 2200's throughput target; the Datapoint 2200 shipped in May 1970 with the TTL CPU and about 1,500 units sold before the Model II of 1973 superseded it with a parallel-TTL CPU of CTC's own design. Intel kept the 8008, waited out its contractual exclusivity period, and released it as the first general-purpose microprocessor on 1 April 1972 at $120 per unit. The 8080 (April 1974) extended the 8008's architecture and preserved its byte order; the 8086 (June 1978) preserved the 8080's byte order for assembler-level source compatibility; the 80386 (October 1985), the Pentium (March 1993), and every x86 and x86-64 processor shipped since have done the same.

The 8008's byte order — little-endian, the least significant byte of a multi-byte integer stored at the lowest memory address — is directly inherited from the Datapoint 2200's bit-serial CPU. The 2200 shifted the bits of each byte through its ALU starting with the least significant bit, because that was convenient for the implementation of multi-byte arithmetic in a serial processor: the carry propagated in the direction of the shift. Victor Poor's December 1969 instruction-set draft therefore specified that multi-byte integer operands were stored low byte first. Intel implemented the specification. The 8080 inherited it. x86 inherited it from the 8080. x86-64, 56 years after the Datapoint 2200 shipped and 50 years after the terminal CPU it was designed for was itself obsolete, is still little-endian for the same reason.

Datapoint Corporation (renamed from CTC in 1972) continued through the 1970s and 1980s as an office-computing and networking vendor; the ARCNET local area network standard, announced by Datapoint in November 1977, was for a brief period in the late 1970s the most widely deployed office LAN technology in the United States, outselling pre-standard Ethernet by about three to one until the IEEE 802.3 standard arrived in 1983. Datapoint merged with Intellogic Trace in 1985, filed Chapter 11 in May 2000, and had its ARCNET intellectual property acquired by Contemporary Control Systems in 2005.

Primary sources:

Faggin, Federico. "The Making of the First Microprocessor." IEEE Solid-State Circuits Magazine 1:1, pp. 8–21, Winter 2009. Faggin was the Intel engineer who physically designed the 4004 and the 8008. §3 reconstructs the CTC/Intel contractual sequence from Intel's internal correspondence files; §4 covers the 8008's byte-order inheritance from the Datapoint 2200's serial ALU.
Wood, Lamont. Datapoint: The Lost Story of the Texans Who Invented the Personal Computer Revolution. Hugo House Publishers, Pittsburgh, 2010. The corporate history reconstructed from the Datapoint archive and interviews with Poor and Pyle. Chapter 4 contains the 2200 architectural specification as sent to Intel and TI in December 1969; Chapter 5 contains the Intel/CTC negotiation correspondence.
Poor, Victor D., and Harry W. Pyle. Instruction Set Specification for the Datapoint 2200 CPU. Computer Terminal Corporation, San Antonio, TX, internal document CTC-ISA-69-12, December 1969. Reproduced in Wood (2010), Appendix A.

Today's goal

Catch one chunter today, friend. Just one. The internal commentary that is running right now about the dishes or the login or the person two desks over who clears their throat every eleven minutes. Say the chunter out loud — not loudly; the room doesn't need to hear you, only the air between your face and the wall. Spoken out, a chunter is a small observation about the shape of the day. Kept in, a chunter is one of twelve things making you tired by 11 a.m. for no reason you could name. The intervention is the exhalation, and the exhalation requires the words.

Today's toy is the chunterbox: a small procedural grumbler in the browser. Press the button, or hit space, and the page produces a short synthesised low-register grumble — pink noise, a sawtooth carrier wobbling in the voice range, two formant bandpasses, a few consonant bursts, and an amplitude LFO doing the muttering. Each one is unique. Lives in the corner.

— C