Applying this to life
I believe a similar approach can be very effective when we’re trying to improve areas of our lives.
It could look something along the lines of:
- assessing our current position, and any areas of dissatisfaction
- deciding on a small change we could try, which we think would improve it
- making that change
- re-assessing the situation, and repeating
Retrospectively, I realise this was an approach I took to my career. I wasn’t 100% fulfilled while working as a doctor*, so I looked for a small step I could take to improve things. The first small step was learning to code, and I enjoyed it. That spurred me on to do an online course in machine learning, then to work for a healthtech start-up, then do my own projects, etc, etc. and after each step I felt more fulfilled.
In machine learning, there are other methods that let you ‘jump’ to the point where the model is least wrong (where the green arrow points in the above diagram). However, gradient descent is preferred when the problem you’re trying to solve is very complicated, because the ‘jumping’ method doesn’t work as well.
I’d argue that our lives are pretty complicated, and problems we want to solve can’t be done overnight, which is why this approach may be well-suited.
(Of course, life doesn’t always follow a straight path, which is perhaps where the analogy falls down.)
But I guess, in summary, what I’m thinking (as I over-analyse this method of computation) is that; for complex decisions such as how we live our lives, it makes sense for us to take an iterative approach, where we assess the outcome each time and use it to decide our next step.
This might not be a particularly novel insight - but it was a fun analogy I thought of this week.
Let me know what you thought :) Did you enjoy this more experimental approach to the email? Or did you prefer the previous style? After all, I guess this is part of my email gradient descent…