Gone on a bit of a mad one this week, looking at risk, precision, and recall.
Not my area of expertise by any stretch. I wanted to know the impact of the different types of errors on decision-making.
First up, here’s some context: My structural editor, B, has signed off on my book content; it’s time to apply the polish. Only, it turns out polish is not cheap.
Here are two outcomes for you:
- The “Positive” Outcome: You sell enough books to make a profit.
- The “Negative” Outcome: The book does not sell, and you lose the £3,000.
Now, if we look at the average outcome, which is a smart move when you are taking a look at how to ground yourself, we know the average author sells 250 books.
If you sell a book at £9.99, you can expect a £5 profit. Not great news if your operating costs are £3,000 ( 250 x £5 = £1,250).
Knowing what the average author sells is useful, but not helpful, since it is likely that in a room full of authors, one author will have outsold all the other authors put together. We need to be a little more cautious about our prediction.
Let’s take a look at the possible errors:
Type I Error (False Positive): You predict the book will be profitable when it won’t be
- You spend £3,000 on polish and lose money
- This is your main downside risk
Type II Error (False Negative): You predict the book won’t be profitable when it would be
- You don’t polish/publish and miss out on potential profit
- Opportunity cost varies depending on how much you actually would have made
The asymmetry is crucial here. Our Type I error has a fixed, known cost (£3,000 loss), while our Type II error cost depends on the unknown upside.
The question is, what would you do? Which error are you going to accept?
I find this interesting – in a world where we are after perfection, it’s humbling to know even simple low low-key decisions like this contain error.
What did I do?
I can’t tell you! You’ll have to wait and see 🙂