Ravdeep Singh

❤️ building.

Testing a new idea!

Thought experiment: can I write 300 words everyday on a new thing I learnt?

Yes of course. Why though? Seems like a ton of effort.

To aid the selfish pursuit of being a better thinker. I’ve realised I think better when I write and I’ve recently spent a lot more time on Twitter / Instagram than I like. Hence, uninstalled both and refocusing that time towards reading / listening / thinking / writing.

Core idea: All models are wrong but some are useful. I came across this phrase in a podcast and it verbalised another idea simmering in my head.

The statistical point this quote makes is that every attempt at modelling a system is an approximation. However, understanding situations where approximation errors become tiny give us a better understanding of

  1. What is the smallest number of assumptions we can make to reduce errors?
  2. Which of these assumptions are risky and might lead to increased noise?

The tangential idea that clicked into place for me was: if a physical book costs €x, how much should the kindle version be?

Model: 80-85% of x.

Assumption: Consumers expect kindle books to be cheaper because they don’t cost anything to produce. Despite the immediate delivery benefit users get, it ‘feels’ unfair if there’s no discount.

We can debate endlessly on what x should be. I’m currently writing an entire book on it. There would be a much smaller debate on the model.

In which situations does this assumption produce noise?

  1. Too low price: For a €2 book, you might as well charge the same price. Consumers compare such low expenses to things they do in everday life. “Every good book is worth more than a cup of coffee”
  2. Physical undesirability: E.g., I’d love to get the Feynman lectures books once again but they take up space and are heavy. I’d rather overpay for the kindle than buy the hardbacks.

For 95%+ books (number pulled out of thin air), the model can still be really useful.

Fin. End of experiment Day 1.

Go Back