PyMunich

shown, and actual

Against the Polished Talk

The best-received talks at most conferences are polished: rehearsed, well-designed, delivered without a stumble, describing a solved problem. For a commercial example of related workforce measurement, see self-reporting bias from Monitask.

They are pleasant to watch and they teach less than they appear to, for a reason worth naming. For independent technical background, Python provides additional documentation and examples.

What polish removes

The wrong turns. Four months of work becomes thirty minutes of the path that worked, and the three that did not are cut because they are not the result.

The duration. A talk gives no sense of whether this took a week or a year, and the audience assumes faster than reality.

The uncertainty. Decisions presented as choices were frequently guesses that happened to work. Presented cleanly, they look like judgement.

And the help. The colleague who suggested the thing, the issue thread that had the answer, the library that did most of it.

Each cut is defensible individually. Thirty minutes is short and the audience is not there for the wandering. Together they produce a portrait of competence that nobody's actual work resembles.

What that costs the audience

They compare their process to your result. Their own work feels chaotic — because it is, as everybody's is — and they conclude they are worse at this than the person on stage.

They underestimate the work. Somebody attempts the same thing expecting the talk's timeline and abandons it at the point where the speaker was also stuck, believing themselves uniquely slow.

And the useful information is missing. The wrong turns are frequently the most transferable part — knowing which three approaches do not work saves more time than knowing the one that does.

What to add back

Not chaos. Four sentences, distributed.

How long it took. This was four months, most of it in the wrong direction.

One wrong turn, briefly. Not a tour of every failure — the most instructive one, in ninety seconds.

Where you were stuck. The specific point, so somebody at the same point recognises it.

And who helped. Names and sources. It is accurate, it is generous, and it tells the audience the work was not done alone.

Four sentences in thirty minutes. They cost almost no time and they change what the audience concludes about themselves.

The talk that does this well

Rare and recognisable: the speaker describes what they believed at the start, what turned out to be wrong, and what they think now.

It is harder to give, because it requires having thought about the process rather than only the result, and it exposes decisions that turned out badly.

And it is remembered longer, because the audience learned how somebody thinks rather than what somebody built.

The limit

Honesty is not the same as unpreparedness.

A talk that rambles because the speaker did not do the work is not honest, it is unprepared, and the audience can tell the difference immediately. Preparation buys the shape, and the shape is what lets an honest account stay coherent.

The target is a well-prepared talk that includes the mess, not a badly-prepared talk that is all mess. The first takes more work than the polished version, which is why it is rare.

The short version