Email me when the next worthy source hits.
Hole one. December 1924, Geneva. Every large lamp maker on Earth signs one page, and for the next fifteen years a laboratory in Switzerland measures how long your light bulb is allowed to burn. Below is everything the desk holds on it, in the order you should take it, with the minutes each piece costs you, what it establishes, and what it does not. The dead ends are kept and labelled, because a dead end you were not told about is how you get lied to politely. We refuse to believe because we are told to. Here is all of it. You decide.
The episode, GNOME VS HISTORY 001, THE RECORD | Phoebus S.A. fined factories whose bulbs OUTLIVED 1,000 hours, is cut and plays here the day it goes public. Until then, start at WHAT IF I TOLD YOU and take the path. When the episode lands, the caption, the on-frame headline plate and every claim inside it trace back to the same sources listed below, which is the whole reason the video sits on top of the path and not in place of it. The voice and the pictures in it are machine-made; the words are checked by a person; claims are claims.
How the desk works: the method. Where it was wrong: the errors page.
He has held the gem up to the light on this one for two days straight. It still is not finished glowing.
The fine ran BOTH ways: every factory bound by the Geneva convention shipped sample bulbs to a single laboratory in Switzerland to be burned until they died, and a factory whose bulbs came back lasting TOO LONG owed money for it, in Swiss francs, scaled by the hour (item 01, CHECKED).
Five people have stood in front of a camera and told you about this cartel. Here is the exact minute each of them does it, playing from their own channel, in their own player, with their own ads and their own view count. The desk rips nothing and mirrors nothing. The embed IS the citation. Under each one: what that creator claims, as their claim, and what they cite for it, because the second question is the one nobody asks. Timestamps read from the caption track on 2026-08-28.
E1. Veritasium, "This is why we can't have nice things," 26 March 2021, at 3:21. WHAT IT SAYS: Veritasium says the December 1924 Geneva meeting became the Phoebus cartel, that its members agreed to cut bulb life to 1,000 hours, "cutting the existing average almost in half," and that factories shipped samples to test stands where a bulb lasting past 3,000 hours cost the company 200 Swiss francs for every 1,000 bulbs sold, a figure he gives at 4:50 and says there are records of. WHAT IT CITES: Krajewski's IEEE Spectrum article, Bernard London's 1932 pamphlet, Giles Slade's book "Made to Break," an NPR Planet Money episode and a third-party upload of the Dannoritzer film, all listed in the video's description. For the 200-franc figure itself he names nothing on the record.
E2. Technology Connections, "Longer-lasting light bulbs: it was complicated," 16 July 2023, at 16:17. WHAT IT SAYS: he does not dispute that the cartel existed, tested its members' bulbs or fined them; he says that by 1925 the lifespan-against-efficacy trade-off in tungsten lamps was understood, that 1,000 hours was "a decent sweet-spot in the middle," and that using Phoebus as evidence of planned obsolescence "rubs me the wrong way" because a 2,500 hour bulb was always dimmer for the same power. WHAT IT CITES: the Wikipedia entry on the cartel, read out loud at 17:58; an excerpt from the United Kingdom Monopolies and Restrictive Practices Commission report of 1951 at 18:33; and a 1981 Washington Post article at 21:24. The description carries no sources at all.
E3. Today I Found Out, "The Great Light Bulb Conspiracy," 4 September 2020, at 3:22. WHAT IT SAYS: the narration says the cartel fixed global light bulb prices, imposed production quotas with steep fines, and "more insidiously" that Phoebus "artificially limited the quality of its light bulbs, standardizing their longevity at only a thousand hours," sold to the public as a trade for more light per watt. WHAT IT CITES: nothing, for any of it. No source is named on the record for the cartel claims and the description lists none. It names Bernard London's 1932 to 1935 essays at 6:10, which is the only receipt anywhere in the video.
E4. Stuff They Don't Want You To Know (HowStuffWorks), "The Phoebus Cartel: Collusion, Corruption and Light Bulbs," 11 October 2014, at 1:49. WHAT IT SAYS: the hosts attribute to "Markus Krajewski, writing for Spectrum" that this was the first cartel in history with truly global reach, then say that where bulbs had run to 1,500 or even 2,000 hours, Phoebus decreed 1,000, that members worked to shorten lifespans while driving the price up, and that every factory had to send bulbs to a central testing facility in Switzerland. WHAT IT CITES: Krajewski and IEEE Spectrum, out loud, inside the sentence that carries the claim. It is the only one of the five that names its source at the moment it makes the claim.
E5. Unlearning Economics, "Planned Obsolescence Will Kill Us All," 22 September 2023, at 5:41. WHAT IT SAYS: he calls Phoebus "an actual, demonstrable capitalist conspiracy," dates the first assembly to 23 December 1924, and at 9:12 says the American light bulb companies "were found to have violated antitrust laws in 1949, which included stipulations about them shortening the lifespans of the bulbs." Then he does something none of the others do: at 9:35 he cuts in an editor's note recorded later, answers the Technology Connections argument directly, says the 1949 court "seems to accept that there is this inverse relationship" while still holding that the cartel had the market power to set the standard, and concedes the economics is "a little bit up in the air." WHAT IT CITES: Krajewski's IEEE Spectrum article and the Justia text of United States v. General Electric, 82 F. Supp. 753, which are path items 01 and 05 on this page, plus Giles Slade's "Made to Break," and both E1 and E2 above, all itemised in the description.
NOT EMBEDDED, and the reason, because a hole prints its refusals too. A full copy of "The Light Bulb Conspiracy," path item 08, sits on YouTube on an account that is neither the director's nor a distributor's. The desk will not embed it: hearing and other public-record footage can be cited from whoever posted it, and everything else is cited from the SOURCE channel, which a stranger's copy of a commercial documentary is not. Item 08 stays OPEN and unlinked until an official upload exists. The desk also looked for a university or museum lecture on the cartel and reached none.
The excerpts above are five people telling you the same story. This is where each of them got it. It is the second question and almost nobody asks it out loud.
| the excerpt | what it cites | where the citation is |
|---|---|---|
| E1, Veritasium | Krajewski, IEEE Spectrum 2014 (path item 01); Bernard London 1932 (item 06); Giles Slade, "Made to Break"; NPR Planet Money; a third-party upload of the Dannoritzer film (item 08) | the description, listed; nothing named on the record, including for the 200-franc figure |
| E2, Technology Connections | Wikipedia's "Phoebus cartel" (item 02); the UK Monopolies and Restrictive Practices Commission report, 1951 (item 09); the Washington Post, 1981 | on the record, quoted aloud at 17:58, 18:33 and 21:24; nothing in the description |
| E3, Today I Found Out | nothing at all for the cartel; Bernard London 1932 to 1935 (item 06) later in the video | on the record at 6:10 only; the description names no source |
| E4, Stuff They Don't Want You To Know | Krajewski, IEEE Spectrum (item 01) | on the record at 1:49, inside the sentence making the claim |
| E5, Unlearning Economics | Krajewski (item 01); United States v. General Electric, 82 F. Supp. 753, the same Justia text as item 05; Giles Slade, "Made to Break"; E1 and E2 above | the description, itemised with links |
What the web shows, read on 2026-08-28. Three of the five trace back to one magazine article written by one scholar in 2014, which is item 01 on the path and is also where this page's own load-bearing numbers come from. Two of the five, E1 and E3, name nothing at the moment they make the claim. The only excerpt that reaches past Krajewski is E2, and what it reaches for is an encyclopedia entry and an excerpt of the 1951 report, which is the artifact the desk has been unable to obtain in full, so the counter-argument travels on a quotation of a quotation exactly as the argument does. E5 is the only one that cites a primary document, the 1949 opinion, and it is the only one that visibly changed its mind in public and left the change in the video. No excerpt shows the 1929 fine table. One of them prints a number from it. Two new leads for the desk, neither read yet: Giles Slade's "Made to Break" and the NPR Planet Money episode.
Status: CHECKED 2026-08-28.
The record shows a convention signed in Geneva in December 1924, a Swiss testing laboratory, and a fine for bulbs that burned longer or shorter than their rating. It establishes an enforced life standard; it does not establish why.
Artifact: the eleven sources on THE PATH below, read by the desk on 2026-08-27 and 2026-08-28. Ruling by the desk, 2026-08-28. The desk rules on what the record shows right now, dated, never on true. A verdict here is alive: when the record moves, this line moves and the old one is kept in the log.
Eleven items, in the order you should take them. No item repeats another item's job. Total: 297 minutes, four hours and fifty seven minutes, of which 117 minutes are items the desk has read to the end and 180 are items still open. CHECKED means the desk opened the artifact itself and it is on the receipt. OPEN means the desk has not held it, and says so rather than borrowing somebody else's summary of it.
01. The spine. IEEE Spectrum, "The Great Lightbulb Conspiracy," Markus Krajewski, 24 September 2014. Article, 20 minutes. CHECKED 2026-08-27. Establishes the signing on 23 December 1924, the membership, the 1,000 hour figure codified by early 1925 against the 1,500 to 2,000 hours previously common, the Swiss testing laboratory, the two-directional fine, the quota fines, reference bulb life falling from 1,800 hours in 1926 to 1,205 in fiscal 1933-34, stable prices against falling costs, and Osram's sales collapsing from 63 million bulbs in 1922-23 to 28 million the next year. Does not establish any figure from the 1929 penalty table, that any consumer bulb was destroyed or withheld, or that the cartel caused the life of any modern bulb.
02. The index to everything else. Wikipedia, "Phoebus cartel". Article, 10 minutes. CHECKED 2026-08-27. Establishes the legal name, the Geneva incorporation of 15 January 1925, shares held in proportion to lamp sales, the existence of a 1929 table of Swiss-franc penalties scaled by exceeding hours, the Anton Philips quotation about not sinking back into the same mire of long-life lamps, and where every downstream retelling gets its citations. Does not establish anything on its own, and it conflicts with item 01 on the pre-cartel bulb life. Read it as a map, never as the ground.
03. The physics, which is the part the story usually skips. Maurycy, "You have been mislead about lightbulbs," 30 July 2026, updated 7 August 2026. Article, 12 minutes. CHECKED 2026-08-28. Establishes the trade-off in numbers: a filament would need about 5,700 C to put its peak emission in visible light, tungsten melts at 3,422 C, so every incandescent bulb radiates mostly infrared and a small rise in filament temperature buys orders of magnitude in efficiency while the metal evaporates and creeps away faster. Establishes that ordinary incandescent bulbs still run 500 to 2,500 hours today, a band that contains 1,000. Does not defend the cartel: the author writes that they absolutely engaged in price fixing and other anti-consumer practices and that he suspects profit shaped the number.
04. How the counter-argument is travelling right now, which is a different fact from whether it is right. Boing Boing, "Phoebus cartel myth: why long-life bulbs are dim," 2 August 2026. Article, 4 minutes. CHECKED 2026-08-28. Establishes that the physics argument in item 03 reached a general audience this month and the shape it took there, including the concession quote. Does not establish any fact of its own: every number in it is item 03's.
05. The court record, and the one place a judge in the United States wrote about this under oath. United States v. General Electric Co., 82 F. Supp. 753 (D.N.J. 1949), Justia full text. Court document, 45 minutes for the sections that matter, and the opinion runs to roughly 98,700 words if you take the whole thing. CHECKED 2026-08-28. Establishes that the Phoebus contract is Exhibit 2278-G in this record, that it grew out of negotiations between International General Electric's Mr Woodward, Philips and Osram after an earlier price-regulating cartel broke down in 1924, that the Government introduced it in rebuttal rather than in the complaint, and that the court found in the flashlight lamp and Christmas tree lamp situations "a paramount concern in what would afford the maximum return in profits to the manufacturer." Establishes, in the same paragraph, that the court rejected the Government's contention that General Electric could have produced a 60 watt lamp giving 24 lumens, and wrote that "the life of a lamp is inextricably related to the power of its light." Does not establish a finding about the household 1,000 hour bulb, which is what the popular retelling of this case says it establishes. See DEAD ENDS.
06. Where the idea itself comes from, six years after Geneva. Bernard London, "Ending the Depression Through Planned Obsolescence," 1932, Project Gutenberg. Document, 20 minutes. CHECKED 2026-08-28. Establishes that planned obsolescence was proposed in public, in print, as government policy, by a New York real estate broker who wanted goods given a legal death date and taxed after it. Does not establish any connection to Phoebus, and there is none in the document. It is here because it dates the doctrine and shows what an open argument for it looked like, which is the measure you should hold the Geneva convention against.
07. The film that made this a story your uncle has heard of. Wikipedia, "The Light Bulb Conspiracy". Article, 6 minutes. CHECKED 2026-08-27. Establishes the film's facts of record: Cosima Dannoritzer, 2010, Franco-Spanish, 75 minutes, also released as "Pyramids of Waste" and "Comprar, tirar, comprar," Best Documentary at Spain's Television Academy Awards 2011, and the named criticism from historian Robert Friedel that the telling is simplistic, that the cartel was not secret, and that the Livermore bulb's long life comes from running under-volted. Does not establish a single primary fact about the cartel.
08. The film itself. "The Light Bulb Conspiracy" (Cosima Dannoritzer, 2010), 75 minutes. Film, 75 minutes. OPEN. The desk has not watched it end to end and will not summarise a film it has not seen. It is on the path because it is the reason most people are here, and because its thesis needs the cartel, which is both a reason to watch it and a reason to check it.
09. The regulator who looked at this and said no. UK Monopolies and Restrictive Practices Commission, "Report on the Supply of Electric Lamps," signed 31 August 1951. Document, 60 minutes estimated. OPEN. The desk searched for a free full text on 2026-08-28 and reached none: what is online is Hansard debate around the Commission and a paywalled 1951 law-review note on its first report. Every retelling of the line that the Commission called the planned-obsolescence allegation misconceived is quoting a quotation. Until the desk holds the report, so is this page.
10. The table with no numbers in it. The 1929 Phoebus penalty schedule, Swiss francs scaled by exceeding hours of life. Document, 15 minutes estimated. OPEN. Named in items 01 and 02 and repeated in every secondary write-up the desk reached, and not one of them prints a figure from it. This is the single most-cited artifact in the whole subject and the desk cannot show you a line of it.
11. Where the internal paper lives. The Philips Company Archives, Eindhoven. Archive, 30 minutes estimated. OPEN, and deliberately unlinked: the desk has not verified a public catalogue address for it and will not print a URL it has not opened. The Anton Philips line about long-life lamps and the same mire is the most-quoted internal sentence in this whole story and it travels everywhere without a document number. A corporate archive is where a document number comes from. The desk has not requested one yet, and until it does, that quotation is a quotation of a quotation.
One mine inside the mine. The Livermore bulb is the exhibit everybody reaches for and it is the weakest thing in the room, which makes it a better hole than the story it is used to prove. Three items, 31 minutes.
The double's own verdict, dated 2026-08-28: the record shows a hand-blown carbon-filament bulb from the 1890s that has burned nearly continuously since 1901 and now draws about four watts. It establishes that a bulb run far under its rating can last more than a century; it does not establish anything at all about what was signed in Geneva.
What the desk chased and could not stand behind. Kept, labelled, dated.
| the dead end | what it is | label |
|---|---|---|
| The 1929 fine amounts | Every reachable source says a 1929 table set Swiss-franc penalties by the exceeding hours of life. No reachable source prints one figure from it. Six independent secondary pages describe the table; all six trace to the same two sources; none carries a number. NEW LEAD 2026-08-28: excerpt E1 states on camera that a bulb lasting past 3,000 hours cost 200 Swiss francs for every 1,000 bulbs sold, and says records of the fines exist. He names no source for that number on the record and his description lists five sources, none of which is the table. The figure is a claim until the desk holds the document behind it. | NOT IN THE REACHABLE RECORD, 2026-08-27, rechecked 2026-08-28; one uncited figure now on the board, CHASING |
| "A 1949 US court found GE's main consideration in setting bulb lifespans was profit" | The desk carried this line on 2026-08-27 from an encyclopedia summary. Reading the opinion itself on 2026-08-28: the profit finding is written about flashlight lamps and Christmas tree lamps, described in the same paragraph as "lines of lesser importance," and the same passage rejects the Government's claim about a 60 watt household lamp and states that lamp life is inextricably related to the power of the light. The finding is real. It is not a finding about the 1,000 hour household bulb. | MISREAD BY THE DESK, corrected here 2026-08-28 |
| The pre-cartel bulb life | Item 01 gives 1,500 to 2,000 hours. Item 02 gives 2,500. The desk uses item 01 in the episode and prints the conflict rather than smoothing it. | CONFLICT, UNRESOLVED |
| The Livermore bulb as evidence about Phoebus | It predates the cartel by about a quarter century and runs far under its rating. It is a counter-example to the idea that a bulb cannot last. It is evidence of nothing about Geneva. | DEAD END, and the episode says so on frame |
| Intent | The record shows a standard, a testing regime, a two-directional fine, a falling measured life and rising margins. It also shows a 1951 regulator calling the obsolescence charge misconceived and a historian calling the popular telling simplistic, and it shows a physical trade-off that would push the number in the same direction with no conspiracy required. | THE DESK DOES NOT RULE |
| The 1951 report itself | Searched 2026-08-28. No free full text reached. | NOT REACHED |
Who is this, in the plainest public role, with a receipt you can open. Nothing invented. Money is a possibility flag and never a finding.
| name | who they are | role in this record | receipt |
|---|---|---|---|
| Phoebus S.A. Compagnie Industrielle pour le Developpement de l'Eclairage | A Swiss company incorporated in Geneva on 15 January 1925 to administer an international incandescent-lamp agreement signed 23 December 1924. | The cartel itself. Set the standards, ran the testing, collected the fines. | Wikipedia, Phoebus cartel |
| Osram | German lamp manufacturer, founding member, shares held in proportion to lamp sales. Its head William Meinhardt was the first to propose the arrangement. | Signatory, and the company whose sales collapse from 63 million bulbs to 28 million in one year sits immediately before the founding. | IEEE Spectrum, Krajewski 2014 |
| Philips | Dutch lamp and electrical manufacturer, founding member. | Signatory, and a named defendant in the 1949 United States case. Its executive Anton Philips is the source of the most-quoted internal line on long-life lamps. | United States v. General Electric Co., 82 F. Supp. 753 |
| General Electric | American electrical manufacturer, participating through International General Electric and the overseas subsidiaries rather than as a direct domestic signatory. | Signatory by subsidiary. Defendant in the 1949 case, where the Phoebus contract is Exhibit 2278-G. | United States v. General Electric Co., 82 F. Supp. 753 |
| Markus Krajewski | Media-studies scholar. Wrote the standard English-language account of the cartel. | Author of the article every load-bearing number on this page traces to. | IEEE Spectrum, 24 September 2014 |
| Cosima Dannoritzer | Documentary director. | Made the 2010 film that put this story in front of a general audience, and tied it to the Livermore bulb. | Wikipedia, The Light Bulb Conspiracy |
The mechanism, and where the record thins out. Possibilities are flagged as possibilities.
Dated lines. When the desk opened the hole, what it read, what changed.
| date | what the desk did | what changed |
|---|---|---|
| 2026-08-27 | Opened the hole to write GNOME VS HISTORY episode 001. Read the IEEE Spectrum article, three Wikipedia entries, and ran a web search for the 1929 fine table. Built the episode's provenance card. | The subject entered the record. The fine table was logged as unsupported on the first day and stayed that way. |
| 2026-08-28 | Widened it. Read the full text of the 1949 opinion, the tungsten physics piece, the Boing Boing note that carried it, the Livermore bulb's own site, and the Bernard London pamphlet. Searched for the 1951 UK report and reached no free copy. | Three things moved. The 1949 "profit" line was corrected: it is about flashlight and Christmas tree lamps, not the household bulb, and the desk had it wrong for one day. The physical trade-off between filament temperature, brightness and life entered the record as a competing explanation that needs no conspiracy, stated by a 1949 judge and by a 2026 engineer in the same words. The Livermore bulb was moved out of the main path and given its own hole, because it proves nothing about Geneva and deserves to be interesting on its own terms. |
| 2026-08-28 | Page staged, unpublished. The gem door on the home page is built and gated off. | Nothing public. |
| 2026-08-28 | Went looking for who else is telling this on video. Checked eight searches of candidates, pulled the caption track for six of them (metadata and captions only, no media), read the transcripts for the minute where the cartel or the 1,000 hour standard is named, and read every description for a source list. Five were embedded at the timestamp, from their own channels. | THE EXCERPTS and THE SOURCES' SOURCES entered the page, and with them the first map of who cites whom: three of five trace to Krajewski, two cite nobody at the moment of the claim, one cites the 1949 opinion and openly changed its mind on camera. One new lead on the fine table, a 200 Swiss franc figure stated on camera with no source named, logged in DEAD ENDS as a claim to chase and not as a number. Two unread sources added to the desk's list, Giles Slade's "Made to Break" and an NPR Planet Money episode. A full third-party copy of the item 08 film was found and refused, because the source channel rule is the whole of the embed law. |
Five of the fourteen items on this page are still open, two of them the most-quoted artifacts in the whole subject, and one of them a fine table that every account describes and no account can show you. That is not a complaint. That is the hole.
Dig. Verify. Bring the gem.
Dig. Verify. Bring the gem.
