Every record in the catalogue rests on eighteen small decisions. For each of six balances, we say whether it holds, is tipping, or is lost. The method defines the last of these as “clearly past the limit” and says no more. When we had ten records scored a second time, blind, the second scorer agreed with us on most balances and disagreed on six. Five of the six were the same disagreement: tipping, or lost? This note is a first attempt at an answer, and an account of what is wrong with it.
A first test
Ask three questions of the harm a balance names. Is it there at all, with evidence? Can an ordinary person still get round it, without unusual money, skill, or sacrifice? And is it the normal condition, or an exception?
- Holds: the harm is absent, or the tool is built so that it cannot arise.
- Tipping: the harm is real and documented, but people can still get round it, or it falls on only part of those the record covers.
- Lost: the harm is the normal condition for most of them, and there is no ordinary way round it.
If nobody has looked, the balance is not assessed. That is a different thing from holding.
One example of each
Each balance asks one plain question of a tool. The question comes first here, and then three answers to it from the catalogue’s own records: one that holds, one that is tipping, one that is lost. They are illustrations, not proofs.
- Biological degradation. Does this tool damage the living world, or the body, faster than either can recover? The bicycle holds: it takes little to make and runs on food. The electric car is tipping: far lower emissions than a petrol car, but particles from tyres and brakes remain. The private car is lost: exhaust and land given over to roads are the normal condition wherever it is used.
- Radical monopoly. Can people still meet the need another way, without this tool? Nobody has to own a bicycle. The telephone is tipping: banks and public services ask for a number, so doing without is hard but possible. Compulsory schooling is lost: attendance is required by law. The test between them is whether a law, a built environment, or a credential leaves no ordinary alternative.
- Overprogramming. Does using this tool require training or certification that people cannot give each other? Riding and repairing a bicycle are learned from neighbours. A library run as a teaching device, with staff between readers and shelves, is tipping. Schooling is lost: people come to believe they cannot learn without being taught. Needing a specialist is not enough; believing you cannot act without one is.
- Polarization. Does this tool move power or advantage to a few, at others’ cost? One person’s cycling takes nothing from anyone else. On the fediverse power sits with whoever runs your server, but you can move: tipping. Hosted language models are lost: a handful of companies can build one, and they set the terms for everyone.
- Obsolescence. Does this tool force people to replace what still works? A book printed five hundred years ago can be read today. Telephone carriers retire a network a few times in a generation: tipping. A hosted language model is replaced or withdrawn by its maker as a matter of course: lost. The test is who decides when the thing stops working, and how often.
- Frustration. Does this tool give back less than it takes, counting everything it takes? A bicycle gives back more than it asks. Automated calls have made answering the telephone costly: tipping. The private car is lost on Illich’s own sum: count the hours spent earning what it costs, and it moves its owner at less than five miles an hour.
What is wrong with it
We gave this draft to the same outside reviewer that has read the method three times. It did not survive intact.
- One test does not fit six balances. “The normal condition for most people” fails for harm that falls on a few, on people far upstream, or on people not yet born, and for harm that is rare but cannot be undone. Polarization can be lost exactly because a minority bears the cost.
- The examples are conclusions, not tests. That a bicycle takes little to make does not prove a balance holds; it needs something to be compared with. “A handful of companies” does not by itself separate lost from tipping.
- The polarization test slides into radical monopoly by asking mainly whether people can leave. It should ask who holds the control, the benefit, the burden, and the bargaining power, even where leaving is possible.
- Frustration needs a rule for whose costs count: the user’s only, or also those of workers, people who do not use the tool, and the public.
- A harm can belong to more than one balance. We had proposed counting each harm once. A phone required for proof of identity may be a monopoly, a concentration of power, and a cost that outweighs its return, all at the same time. Each needs its own argument.
The hardest point was the plainest. We asked whether the test would have settled the six disagreements. The answer was that it would clearly settle none of them on the evidence we have. Four of the six were not disagreements about words at all. The records did not contain the fact the judgment turns on: how often a smartphone is required, how long its maker supports it, how a health budget is divided. No test can supply a missing fact.
Where that leaves the scoring
It changes the order of work. A sharper definition of “lost” is worth having, and each balance will need its own. But most of the uncertainty in the catalogue is of a simpler kind, and the catalogue now shows it: a balance with no evidence behind it is marked not assessed, and a record with such balances shows its position as a range or as “at least” a number.
Illich did not give a test either. In Tools for Conviviality he names the balances and says that each must be preserved. How to tell a balance under pressure from one that has gone is left to the people who have to live with the tool. That may be the right place for it. A catalogue can show its working and say what it does not know. If you can see a better test for any one of the six, tell us.
Sources
- Illich, Ivan. Tools for Conviviality. Harper & Row, 1973. https://archive.org/details/illich-conviviality.