A Tool Is Only as Wise as the Question

There's an old, blunt computing principle that has aged better than much of the technology it was coined to describe: garbage in, garbage out.

A system, however sophisticated, can only work with what it's actually given. A vague or poorly framed question tends to produce a vague or poorly framed answer, no matter how capable the tool answering it is.

There's something almost ritual about setting a cup of tea down and looking out at the small rectangle of sky between two buildings before I write anything down properly. I've started treating that pause as part of the actual work, not a delay before it.

The question I bring to any tool, human or otherwise, decides more about the answer I get back than the tool's own capability does. A wise answer to a lazy question is rarer than people assume, whatever's doing the answering.

Most of what looks like a tool's limitation is actually a question that was never sharp enough to deserve a sharp answer. I'm slower to blame the answer now.

I'm quicker to go back and rewrite the question that actually produced it in the first place, longhand, before trying again. It's slower going, most days.

It also means the answer that eventually arrives actually belongs to the question I meant to ask. I write the question out longhand first now, before I ever open anything that might answer it for me.

The pen is slower. Slower turns out to be where the actual thinking was hiding.