[ recap ]

Last week we looked at art + technology, technē, Open Score and AARON.

[ thoughts ]

Sometimes a tool will not quite do what you want. A piece of software has no setting for the thing you need, so you change the way you work instead.

By tool here, I mean something made to help us do a job, from a hammer to software.

In 1973, the Austrian social critic Ivan Illich published Tools for Conviviality.

For Illich, a drill and a school curriculum could both be tools because both had been deliberately designed for a purpose.

Illich wanted "tools to work with rather than tools that 'work' for them". He called such tools convivial. With a convivial tool, the person using it could decide when to use it and what for.

One example was a transport system built around human-powered vehicles, with roads and repair stations to support them. The system could be complex while the vehicles were still powered by the people using them.

Canadian physicist Ursula Franklin reached the subject through ancient Chinese bronze.

Franklin had spent much of her professional life studying Chinese bronzes. In her 1989 CBC Massey Lectures, later published as The Real World of Technology, she described large ritual vessels made through a tightly organised casting process.

A full-size model came first. Workers covered it in clay, cut the dried mould into pieces, fired them and assembled them again around a core. Molten bronze was then poured into the mould. The stages required different kinds of expertise and had to happen in the right order.

Franklin called this prescriptive technology.

The job had been split into separate parts, and someone had already decided how each part should be done. In the bronze foundry, the person making the mould had little latitude for judgement.

A potter working on a single pot has more room to respond. The wall can be made thicker or thinner while the clay is still in their hands. Franklin called this holistic technology: the maker stays with the job from beginning to end and makes decisions as they go.

Franklin wrote that prescriptive systems could be extremely effective and efficient. In those systems, many decisions were made before the worker started.

[ creativity ]

The mathematician and computer scientist Seymour Papert helped develop Logo in the late 1960s, a programming language designed for children.

Children could use Logo to control a small mechanical turtle or one drawn on a computer screen. They typed an instruction and watched the turtle move.

In Mindstorms, published in 1980, Papert described the relationship he wanted as "the child programs the computer". If the drawing came out wrong, the child could change the program and try again.

With generative AI, software that can produce text or images from an instruction, you can ask for a result in ordinary language without writing the instructions that produce every part of it.

A 2025 study watched people use generative AI for image-making and creative writing. During one writing task, the AI suggested a familiar story in which train delays produced anxiety. The writer chose a calmer direction instead.

The study was small and its authors call the findings preliminary, so one writer cannot tell us how everyone works with AI. It gives us one documented example of someone using AI-generated text without following the direction it suggested.

[ making ]

Researchers at Ink & Switch, a software research lab, once worked on a team that that ran its projects with index cards taped to a wall. When the team changed the way it worked, it changed the wall: tape lines moved and new groups of cards appeared. A new idea could be tried in minutes.

Later, remote work pushed the team towards an online project tracker. One of the special areas they had made on the wall could not be recreated there, so they dropped that part of the process. Changes that had taken minutes on the wall could now take hours to set up, if the software allowed them at all.

In 2025, Ink & Switch used that story in Malleable Software, an essay about software people can reshape to fit their needs. The authors describe the problem as being "forced to adapt your workflow to fit your software".

Ink & Switch had already experimented with software that could be changed by its users. Potluck, introduced in 2022, starts with an ordinary text note. Its coffee recipe can be given a slider that changes the quantities and a timer for the brew. A user can add those features inside Potluck and keep changing them there.

By 2025, the lab was also using AI inside Potluck. The AI can draft some of the rules that control what the note does. Those rules stay visible, and the user can change them.

For one recent piece of work, make a simple timeline from start to finish. Beside each step, write me, tool or both.

Some steps you no longer do yourself may be exactly the boring work you wanted to hand over. Others may be where you used to shape the work.

[ sign off ]

When the steps are on the page, you can see where the tool is helping and where it has already decided part of the process.

Stay the one doing the noticing.

Thanks for reading. See you next Sunday.