Connecting the dots

If I could go back in time

September 19, 2020

Imagine one of your friends comes to you asking you to invest in their startup. You really believe in the idea, think it can change the world and think you can make some money by getting in early. You don’t want to regret later on that you passed an opportunity to invest in the “next Facebook”. On the other hand, as with most startups, it could fail and you could end up losing it all.

If the startup does end up becoming the next Facebook, you would become a billionaire, this decision becomes the best decision of your life. You pat yourself on the back for a job well done and start giving talks and writing books, giving sage advice to other aspiring angel investors.

If the startup ends up failing, you resolve that you will never invest in startups again. You begin to question your decision making ability (and everything else you do :p).

There is a over-correction happening in both of these cases because of outcome bias.

Outcome bias

Outcome bias happens when you judge the quality of your decision based on the outcome it produces instead of focussing on how you arrived at a particular decision. In the above example, because the startup became successful, your decision turned out to be right.

We can see outcome bias all around us - the sales person with the highest sales numbers gets promoted but the sales person who took a calculated risk, which ultimately did not not pay off, gets no credit.

But thinking about situations from this angle can be useless for a few different reasons.

Even the best process can produce unfavourable outcomes

The best algorithm for the secretary problem is the 37% rule (under certain conditions) and even there, it fails 63% of the time!

For most decisions in life, we don’t get to roll the dice multiple times till we get a favourable outcome, even if we are using the best possible algorithm to make our decisions.

Inaccurate data

Even if you have the best algorithm in the world, you probably don’t have the most accurate data to give to your algorithm. You can’t predict how the competitive landscape is going to change exactly. Probably there are a few other teams working on the exact same idea as your friend’s.

Unpredictability

Modeling the real world is hard. Even if you had accurate numbers for all parameters in your model, you can never truly account for the unpredictability of the real world. Your friend’s business plan definitely did not have a section on “How the business will survive a pandemic”.

What should you do instead?

Don’t beat yourself about it

As we have seen, there is a lot of reasons why something may not work out as you expected it to, even if you did the best you could at that time. Once you internalize this, it becomes easier to not beat yourself about “bad” decisions you took in the past. Over-correction based on any one incident is not a good thing.

Separating the process and the outcome

This doesn’t mean that you don’t learn from your experiences. But what you learn from your experiences is to see if you can change your decision making process somehow. Remember that this retrospective analysis should be done even for decisions that had a good outcome! You just might have been lucky that particular time.

It is pointless to think “If I could go back in time, I would not have invested in my friend’s startup”. The exact same situation at the exact same time is not going to happen to you again anyway. Instead these are more useful questions to ask yourself

  • How would the weightage to different factors I considered be different?

    Now that you have gone through one concrete manifestation of the different possibilities, you might realize that the weights you had given to certain factors might need some tweaking.

    I didn’t invest in the startup, but I am regretting passing up on the opportunity more than I thought I would.

  • What factors I had not even considered before?

    Each decision produces more information and some of that may be due to factors you had not accounted for before. You can use this to make better decisions as you learn more about yourself and the people around you.

    My friend feels betrayed that I did not invest in their startup and I feel bad about that which I had not considered before.

Thinking about all possible outcomes

One way of not falling for the outcome bias is contemplating about alternate outcomes instead of just the outcome that happened. It is fascinating how much weightage we give to our own experiences. At each point in our life, things could have gone so many different ways - some because of the decisions we took, but mostly because of the events happening around us which we have no control over. The lives we are living now is just one manifestation of infinite possible ways in which it could have gone. But still we take almost all of our decisions based on how things have gone till now. It is like we are adjusting our weights to our neural network with just one training sample (#sorrynotsorry).

So your learnings ideally should be = learnings assuming the startup failed + learnings assuming the startup succeeded. Since we can’t live out all alternate timelines, the closest we can come to is by actively thinking about what your learnings would have been in the other timelines.