And today's episode, I want to talk about a paradox that we face in our work and our lives. They are probably going to face today. My name is Jonathan Cutreller, listening to Developer Tea. My goal on this show is to help driven developers like you find clarity, perspective, and purpose in their careers. This paradox is one that will cause frustration between, let's say, a long time engineer and a product manager. It might cause frustration on a team. It might cause frustration between a data analyst and a junior engineer. This paradox is very simple. It is the constant belief or the constant pressure to use empiricism to drive our actions, and use our experiences, use data measurement, use some kind of feedback loop to drive our decision making. That's one side. The other side is the simple fact that most of our experiences are anecdotal. Most of our experiences are one-offs, or the sample size is so small that there's nothing meaningfully statistically significant in those experiences. This paradox can be hard to deal with because we have to hold both of these truths in our minds at the same time. The first truth is that empiricism is probably the most rational way to behave. Using some kind of input from your environment and from your experiences as one of your factors and decision making is a rational way of thinking. This is why we learn from iteration. But we shouldn't treat this as the only thing to consider. Instead, on the flip side, we should recognize the possibility that what happened in this case was erratic. What happened in this case could have been dumb luck. Or possibly the result of something that we haven't measured well. What happened in this case may not be repeatable. It may change with context. When we take into account empiricism, learning from iteration, bringing that data back into our experience and trying to make adjustments based on that, we shouldn't make that our only input. Instead, we should be seeking outside views. What is an outside view? Well, in some cases, it might be some base rate. How often does this happen? Let's say on other teams like ours. That's a simple example. If we see people, let's say, churning from a particular team in our organization, should we imagine that the whole organization is in trouble? Or we could take an outside view, look at other teams. It's possible that this team is doing something different than the rest of the other teams. Maybe there is some kind of overwork scenario or this particular manager is micro-managing. Who knows what the problem is. The important part here to understand is that if we only look at a specific experience and we try to generalize that to all of our other experiences, we're going to experience the pain of the second argument here, which is to say we're going to wrongly generalize very often because those experiences are by definition anecdotal. That said, that doesn't mean that those experiences don't mean something. We need to always be pushing, always be pushing for context and always be pushing for an outside view. It's critical that we understand this basic concept. The underlying statistical concept here is seeking a base rate. What happens in cases like this? What happens in the broader population or when events like this one occur? How can we compare this thing to another thing? Are we actually experiencing the base case or is this an anomalous event? When we only look at the event itself, we assume that no event is anomalous. This is critical to understand. We must inspect what happens. We must learn from iteration. But we can't look at events in isolation and imagine that we are immune to anomaly, hold both of these things in your head at once. The way that you learn from iteration is going to improve drastically. Instead of imagining that your iteration is a test tube experiment where all you're doing is changing variables and everything is isolated perfectly, you instead recognize that there are so many variables, there are so many influences, and there are so many possible outcomes that there's no chance that you can practically eliminate those influences. So instead, you have to look at the whole picture. Look at the picture from the outside view where those influences are also impacting. In this way, you're practicing a technique of isolation, which sounds once again like an oxymoron, but what you've done is you've isolated the environment to the scale where the data becomes reasonably important, where it becomes statistically significant. When you look at your anecdotal experience in context, you excavate meaning. You might excavate the meaning that this was a fluke, or the meaning that you wouldn't expect it to have gone this way, there's some kind of influence that you haven't figured out yet that's caused it to go this way. Or maybe you excavate the meaning that things are exactly as you expected them to be, that your anecdotal situation just so happens to also line up and be generalizable, that those other teams are also losing members to their team. But without the outside view, you have none of that. And only the outside view you can't iterate. There's nothing to iterate on. So hold both of these things in mind as you review, if you're in a retrospective or a post mortem. Hold both of these things in mind. And I guarantee that your iteration will be much, much more effective. Thanks so much for listening to today's episode of Developer Tea. If you enjoyed this episode, please subscribe and whatever podcasting app you're currently using, that way you don't miss out on future episodes. And if you enjoyed this conversation and would like to take it a step further, you can join the Developer Tea Discord community. Head over to developer.t.com slash discord. There's nothing to pay for over there. If you've ever used Discord, you know this is generally true about Discord. But we're not trying to sell anything there. We're not pushing any product or anything like that. It is purely a community for people like you who are connecting and improving every day in their lives. Talk about getting an outside view. That's a place you could do it. Thanks so much for listening and until next time, enjoy your tea.