I am using forecasting to worry and ruminate less

This post also includes a graph of everything I have been worrying about for the past year (which feels vulnerable, but probably helpful, to post publicly).

Key points

• I was worried about having my stuff stolen so I made a quantitative prediction about the chance of this happening.

• This gave me a clear sense of how likely I thought it was that the thing I was worrying would happen; making me worry less and make a (probably) better decision.

• I started doing this for my many other worries!

I went surfing and couldn’t find my friend, this left me with a dilemma 

A few years ago I went surfing with a friend. Except they didn’t want to go surfing. So they found a spot on the beach to read while I surfed. We agreed to meet back at that point when I was finished. Leaving my phone and car keys with them, I went surfing.

After an hour or two I was tired, so I left the water to meet back up with my friend. When I got back to the place they had been sitting, they weren’t there. I considered waiting in that spot, but I was cold and the most likely explanation for their absence was that they had gone back to the car. So I marched the 200m or so back up the beach to the car. It felt like a long way and the surfboard was heavy.

When I finally got back to the car I leant the surfboard against it and rested. But my friend was not here either. So what next…  I was still cold, and was now even more tired and had no way of communicating with them. I had ruled out the most obvious hypothesis, now the most likely thing was that they had left their spot on the beach for [some other reason], and we're back there, waiting for me…

So now I had 3 choices.

1. Stay put and wait for them to realise that I’m not getting out of the water (seems bad, could be ages)

2. Walk back down the beach with my surfboard and wait at the agreed spot (it would take about 10 minutes)

3. Leave my surfboard by the car, and walk back to meet them unencumbered.

I was worried about leaving my surfboard unattended 

Option 1 could mean waiting ages so it seemed clearly the worst. So the real question was “do I take my surfboard with me, or do I leave it by the car” I didn’t want to walk with my surfboard because it was really heavy. And I didn’t want to leave it there because it was expensive and could be stolen. It was a very difficult situation. 

I had recently learnt about the idea of “expected value” (EV). EV is (to simplify) the value (or cost) of a thing happening multiplied by the probability of the thing happening. So in my case the EV of me leaving my surfboard by the car was:

[Probability of someone stealing the surfboard before I get back] x [Cost of theft] 

If I knew this, I could then work out whether it is greater or lesser than the cost of having to carry my surfboard up and down the beach. (Given how tired I was, I would have been willing to pay around £40 to not have to carry they surfboard)

I predicted the chance of surfboard theft was 2% so I left my surfboard by the car.

The cost of the theft was relatively easy to figure out. The surfboard cost £500 new, so maybe £450 now (I’d used it a few times). I would also have to go buy a new one. I would probably pay £50 to not have to do that. So the total cost of the theft is around £500.

There might be some psychological/social costs too. But believing in the model I am using here is a significant defence against these. The times I have had things stolen were slightly annoying, but only because of the inconvenience. I suspect I would be much more traumatised by being the victim other crimes, and would want to account for that - but theft like this is basically fine for me…

So the main unknown is just the [probability of surfboard theft]. The cost of carrying the surfboard was £40, so the EV just needs to be less than -£40 to be worth leaving the surfboard. So the crux was: is the probability of theft bigger or smaller than 40/500 (8%)? I looked round the carpark, saw the families, and other surfers and built a vague intuition for how many people might come nearby in the next 30 mins or so. Not many, maybe 10. And what proportion of these people would actually opportunistically steal a surfboard? Again, surely that’s rare?! Maybe 1 in 500 people? This results in a probability of theft of 2%. So the expected cost of leaving my surfboard was £10. Lower than £40. I left my surfboard in the car and went to find my friend.

I found them, and when we got back to the car, the surfboard was still there! (yay)

Possibly more importantly, because I had done the maths, I didn’t worry. I had already paid the £10 cost emotionally. Part of me was still a little concerned, but that part of me could be reassured that I had made a very-sensible-maths-based-decision.

I worry a lot and making specific predictions consistently reduces the worrying

There is a reason that insurance companies sell “peace of mind” around sudden unexpected costs. Worrying about things really sucks. Real fear and rumination can suck joy out of any situation. If (like me) you find it relatively easy to find reasons to worry, then tools to mitigate worries are really powerful.

I started doing the above every time I left my things unattended, and it has consistently been reassuring (and never resulted in my things being stolen). But I don’t just worry about petty theft…

About a year ago, I decided I wanted to be more self-aware about what I was worrying about. So every week (as part of my weekly review) I listed around 2-5 things I had been worrying about that week. I recently categorised these ~200 worries and put them on a graph.

Figure 1: The number of weeks in the past year that I have worried about that category of thing (where I worried enough to remember by the weekend)

This week I will probably add a new bar to this graph, something along the lines of - “people now think I am super neurotic because of all the things in my worries graph.”

This wasn’t just a data collection exercise, I also used the above prediction technique to assess these worries. i.e. what is the probability that the thing that I am afraid of will actually happen? Here are a few specific things I predicted:

• I will feel bad about what I have achieved when I present my work at our quarterly review (<5%)

• People will not like and/or not engage with the article I posted on impact in the Civil Service (<200 views by end of Sept) (20%)

• We won't' get accepted as tenants for a new flat due to difficulties proving income (25%)

• I will get offered [job that I wanted but thought that my application might be embarrassing] (20%)

• I will be explicitly public embarrassed by something that I post online in the next year (someone whose opinion I respect highlights something that is embarrassing enough that I need to remove it or even post some kind of apology) (5%)

I haven’t gone into the details of how I came up with these numbers, but I tried to roughly follow good forecasting protocol as explained in books like Superforecasting. And in case you want to know: The quarterly review was good, the article got plenty of positive attention, we got the flat, I didn’t get the job, I haven’t been publicly embarrassed (yet).

In all these cases, having an actual number on the worry was immediately relieving. Like buying insurance. I’ve already accounted for the risk, no more rumination needed (I still ruminated on these things a little, but definitely less).

Those familiar will notice that the above is a form of Cognitive Behavioural Therapy (CBT). CBT is a mental health intervention that asks patients to uncover and challenge beliefs about themselves or the world that are causing them distress. 

By exposing my beliefs and putting numbers on the probabilities, I can make an actual decision as to whether I want to do anything about the risk or not. By making the beliefs explicit I can challenge or accept them. I believe this has made me ruminate less and generally feel better.