In a previous post I said:
I'd rather set my nearterm goals low and be achieving them than to set nearterm goals high and be not achieving them. I also like to have a meta goal where I continually increase my nearterm goals.
I said this as if it's just a personal preference thing. As if it's subjective. It's not.
The method I described in the quote is similar to what physicists do when creating models of physical reality. They start by making the simplest model. Then they test it against reality. If they don't succeed, they try other simplest models. When they succeed, they add complexity iteratively while continuing to test against reality.
- nearterm goals = models
- actions = reality
If you try to start with a model that is too complex and it doesn't match reality well, then you can't tell where the error is (which part is causing the error? can't tell. the data doesn't differentiate between them). This also happens if you add too much complexity to an existing model.
This stuff applies in software writing too. I've heard the phrase "test often". I agree with that. And that means that between the test iterations there are successful tests that are being used as stepping stones to get to more ambitious tests.
No comments:
Post a Comment