Prev | Next

Algorithms To Live By

Brian Christian and Tom Griffiths. 2017.

"Precepts include: Don't always consider all your options. Don't necessarily go for the outcome that seems best every time. Make a mess on occasion. Travel light. Let things wait. Trust your instincts and don't think too long. Relax. Toss a coin. Forgive but don't forget. To thine own self be true."

The optimal stopping problem

How long do you take looking before you go for the best of the bunch? Gather data 37% of the time you have available to make the decision, and then make it.

If you're not sure how a group of people are in their overall rankings but you can tell which one is better than another then a "mathematician would say" you have access to the ordinal numbers but not the cardinal ones. So, that's like having ranks 6, 8 and 12 in front if you, but you only know that one of the candidates is 2 better than another and the third is an additional 4 better than the better of the other two.

Explore vs Exploit

Explore when you have time. Exploit when time is running out. When a company starts only exploiting its assets then it's a sign that it knows time is running out.

What this means practically is that when you move to a new city you should try out lots of places to eat. When you leave do the round of your favourites because you won't see them again.

From an investing standpoint when you see an industry trying to play it safe and only put out stuff that's worked before you know they are headed for a cliff edge. The film industry is trying to build only on established franchises. On Disney+, for example, once I've watched the Marvel stuff, what else really is there to see? This could be a sign that the age of big screen cinema is over. You'll get safe, blockbuster experiences there still but the place for story has shifted online, to streaming, aggregators and indies.

End result of this - try out things, stick with the stuff that works, while it works. When it stops working, or looks like it's running out of steam look for something else.


Sorting is often quadratic or exponential - the more things you have the longer it takes to sort. The most efficient sorts are when you do smaller sorts - put things into buckets and then sort the buckets.

What's even more efficient is putting things into buckets and then sorting only if you have to.

I use a version of this when working on book length projects. I use slips of paper to first jot down all the ideas that could go in the book - ending with 30-50 pieces of paper. Sorting all of them would take too long so I first put them into three piles, beginning, middle and end. Then I sort each pile, comparing each piece of paper with the others to get the ideas in order, which is pretty quick since you're only comparing 10-15 ideas rather than 45-50. Then I can write, a slip of paper a day from start to finish and end up producing 45,000 to 60,000 words.

I've also learned that when writing it's better to write a paragraph at a time rather than a sentence at a time, even online. It's okay to write a sentence at a time if you're never going to edit later, but when you want to pull those things you've written into a bigger piece then all those sentences exponentially increase the amount of time needed to sort the book. There's a good chance I'll never finish my first two projects because the overhead of ordering all those lines is not worth it - that's 120,000 words just thrown away.

It's a good thing I'm still at the exploring stages of figuring out how to write!