The Bulb That Never Went Out: What a 1925 Cartel Actually Proves

In a fire station in Livermore, California, there is a light bulb that has been on since 1901. It was made by the Shelby Electric Company in Ohio, hand-blown, with a thick carbon filament designed by a French engineer named Adolphe Chaillet. It has its own webcam. People make pilgrimages.

It is the single most shared piece of evidence for the belief that things used to be built to last and are now built to fail.

It is not evidence of that at all. But the belief it is used to support has a real, documented, minuted case sitting behind it, and that case begins in a room in Geneva.

Geneva, 15 January 1925

On that day a company was registered under the name Phoebus S.A. Compagnie Industrielle pour le Developpement de l'Eclairage. Behind it were most of the electric lamp industry on earth: Osram of Germany, Philips of the Netherlands, Tungsram of Hungary, Associated Electrical Industries of Britain, and General Electric of the United States.

They agreed, among other things, to standardise the working life of an incandescent bulb at 1,000 hours. The typical bulb before this had lasted somewhere around 2,500.

The part that removes any doubt about intent is the enforcement. Members sent sample bulbs to be tested, and manufacturers whose lamps lasted longer than the agreed life were fined, with the penalty calculated in Swiss francs according to how many hours over they had gone. A company could be punished for making a product too good.

Anton Philips stated that the point was to prevent "a period of long life lamps", and that it mattered not to supply lamps with a very prolonged life. In 1949 an American court, looking back at the arrangement, found that General Electric's main consideration in setting bulb lifespans had been profit. Prices went up, because with the market carved up there was nothing to stop them.

The cartel intended to run until 1955. It fell apart in 1939 when the war began.

So the conspiracy is not a theory. It happened, it was written down, and the people involved were reasonably candid about why.

The part the meme leaves out

Here is where an honest account has to slow down, because there is a genuine engineering argument buried in this and leaving it out would be its own kind of dishonesty.

Filament life and filament efficiency pull against each other. Run a filament hotter and it gives more light for the same electricity, and it also burns out sooner. Run it cooler and it lasts longer while giving you less light per unit of power and a dimmer, more orange light. There is no setting that wins on both. Every bulb ever sold sits at some chosen point on that trade.

Regulators looking at the cartel acknowledged this: shorter-life bulbs can be brighter for the same wattage. A 1,000-hour standard is not automatically a swindle, because somebody has to pick a number and 1,000 hours is a defensible place to pick.

What makes Phoebus a scandal is not that a number was chosen. It is that it was chosen in secret, by everyone at once, enforced with fines, and accompanied by higher prices with the competition removed. The crime is the cartel, not the filament.

The bulb in the photograph proves the opposite of what people think

Now back to Livermore, because the famous bulb is doing none of the work it is credited with.

It was originally around 60 watts. It now draws about 4, roughly a nightlight. It has been left on almost continuously for over a century, which matters more than anything else, because what kills a filament is the thermal shock of heating and cooling every time somebody hits a switch. Add a thick hand-made carbon filament rather than a thin modern tungsten one, good brass and glass, and a high-nitrogen atmosphere inside the envelope.

What the Centennial Light demonstrates is that a bulb run very gently and never switched off will last a very long time. That is a fact about physics and about how it has been treated. It is not a fact about the moral character of manufacturing in 1901, and it would tell you nothing about Phoebus even if the cartel had never existed.

Where the phrase came from, and the surprise in it

Two other meetings matter, and neither is what people assume.

In 1932 a New York property man called Bernard London published a pamphlet titled Ending the Depression Through Planned Obsolescence. The twist is who he wanted to do it. London was not describing a corporate plot, he was proposing a government policy: the state should give goods legal expiry dates, after which they must be surrendered and replaced, in order to force spending and restart a stalled economy. It was a Depression-era stimulus idea, published openly, and it went nowhere.

The phrase entered ordinary speech in 1954, when the industrial designer Brooks Stevens used it at an advertising conference in Minneapolis. His definition is worth reading closely: instilling in the buyer the desire to own something a little newer, a little better, a little sooner than is necessary.

That is not a description of things breaking. It is a description of wanting. Stevens was talking about desire, and most of what people now call planned obsolescence is closer to his version than to the cartel's.

The survivors are not a sample

There is also a plain statistical reason the past looks better built than it was.

The 1970s refrigerator still humming in somebody's garage is famous precisely because it is unusual. Its thousands of siblings failed years ago and went to landfill, and nobody photographs those or posts about them. What survives to be admired is, by definition, the tail of the distribution, and comparing that tail against the average of everything sold today is not a comparison at all.

Some things have also genuinely got better, and the honest version has to say so. When Japanese cars with longer service lives arrived in the American market in the 1960s and 1970s, domestic manufacturers were forced to build more durable products to compete. Competition, in that case, made things last longer rather than shorter. And an ordinary LED lamp today will outlive the 1,000-hour standard Phoebus fought to impose many times over.

What actually changed

The modern version of the problem is real, but it is not usually a filament specification. It shows up in three forms.

The first is cheaper construction at a lower price point, which is a genuine trade the buyer is partly making themselves. The second is the prevention of repair: glued assemblies, proprietary screws, batteries sealed into the case, and spare parts that are simply not sold to anyone outside an authorised network. The third is newer than anything the cartel imagined, which is that a device can be physically perfect and still be finished, because the software support has ended and it will no longer take updates.

That last one is the true successor to Geneva. Phoebus had to physically shorten the life of an object to retire it. A support window retires it by announcement, at no manufacturing cost, and without touching the hardware at all.

The bulb in Livermore has now been burning for more than a hundred and twenty years, which is long enough that the company that made it, the cartel that would have fined it, and the entire technology it belongs to have all gone. It is still just about the only object in the building that nobody has had to think about.

Sources

• Wikipedia, "Phoebus cartel" (the registration in Geneva on 15 January 1925, the member companies, the 1,000 hour standard against roughly 2,500 previously, the testing and fines in Swiss francs for lamps that lasted too long, Anton Philips on avoiding long life lamps, the 1949 American court finding on General Electric's motive, the price rises, and the collapse in 1939).
• Wikipedia, "Centennial Light" (the 1901 installation at Livermore, the Shelby Electric Company and Adolphe Chaillet's carbon filament, the fall from about 60 watts to about 4, and the technical explanations for its longevity including continuous operation, filament thickness and the nitrogen atmosphere).
• Wikipedia, "Planned obsolescence" (Bernard London's 1932 pamphlet proposing government-imposed expiry dates, Brooks Stevens popularising the phrase at a 1954 Minneapolis advertising conference and his definition of it, the categories of contrived durability, prevention of repair and programmed obsolescence, and the effect of Japanese competition on American car durability).
• IEEE Spectrum, "The Great Lightbulb Conspiracy" (Markus Krajewski's account of the Phoebus archive, the testing regime and the fine schedule, and the argument over how much of the 1,000 hour standard was engineering and how much was commerce).
• Centennial Bulb, "the official site of the Livermore bulb" (the bulb's history, its continuous operation and its webcam).
• Wikipedia, "LED lamp" (rated lifetimes far in excess of the incandescent standard the cartel set).
• Wikipedia, "Right to repair" (glued assemblies, proprietary fasteners, sealed batteries and restricted parts supply as barriers to repair).
• Wikipedia, "Incandescent light bulb" (the trade-off between filament temperature, luminous efficiency and service life).