EA Concepts: Inside View, Outside View

post by aarongertler · 2018-09-18T22:33:08.618Z · score: 2 (1 votes) · EA · GW · 2 comments

Contents

```  Forecasting with the Inside and Outside View
None
```

Hello, readers! I'm trialing for the Content position at CEA; as such, I've been asked to draft a couple of posts for the concept map.

These are meant to be close to the current style (no links in body text, fairly concise). I'd love to hear your feedback on this post. Specific questions:

1. Do you prefer "we" or "you" as the pronoun of choice?
3. When you imagine this from the view of a reader who is newish to EA, and clicked on a link to read about something called "outside view", does it make sense? Is it clear why this dichotomy could be useful?

Thanks for your help! The other concept draft is here.

Forecasting with the Inside and Outside View

When we make predictions, we often imagine the most subjectively “likely” way that something could happen, and then judge whether this scenario seems reasonable. This is called “inside view” thinking.

For example, if we want to predict whether we’ll get to work on time tomorrow, we might plan out a morning schedule, then judge whether we think we’ll be able to follow it.

However, another method -- "outside view thinking" -- tends to be more reliable. In the case of getting to work, we're more likely to make an accurate prediction if we simply think about all the other mornings we’ve gone to work, and estimate how often we were late.

(This is most helpful if tomorrow is in the same “reference class” as those other days, without unusual factors -- like an important meeting or a blizzard -- that increase or decrease the odds of lateness.)

The outside view typically works better because it “automatically” includes data about unpredictable circumstances. If we’re late to work, it could be for many different reasons: oversleeping, missing the bus, forgetting our keys, etc. Calculating the odds for each individual reason would be very difficult, but looking at past workdays means that we can avoid doing so, by summing them into a single number. If we are late for any reason half the time, and today is a typical day, we probably have about half a chance of being late, for any reason, tomorrow.

If you ever find yourself trying to forecast about something in a common “reference class” (that is, you have information about a group of similar things), try using the outside view. The history of past events is often more reliable than your own judgment.

Hanson, Robert. 2017. Beware the Inside View.

Wikipedia. 2018. Reference Class Forecasting.

comment by Risto_Uuk · 2018-10-03T09:36:27.093Z · score: 3 (2 votes) · EA · GW

1. I prefer "we".

2. I'm not sure what kind of references you are supposed to add here. Should they be accessible to everyone or can books, etc. be included as well? If the latter, then I'd add Daniel Kahneman's book Thinking Fast and Slow to the list. There are good parts about these concepts in the book. (e.g. Kindle version location 4220)

3. To me, it seems that the definitions of "inside view" and "outside view" are not clear enough, whereas the examples are very good. https://www.hybridforecasting.com/ had nice slides about this, however, I'm not able to find their material to share here. Anyway, their definitions and explanations are the following:

• Inside view: focus on the unique qualities of the case at hand.
• Outside view: connect the case at hand to a reference class and rely on base rate information.
• Reference classes refer to similar events from the past.
• Base rates are relative frequencies of an outcome given a defined set. For example, the chance of selecting a red card from a deck of cards if 50%.
comment by aarongertler · 2018-10-31T06:35:14.923Z · score: 1 (1 votes) · EA · GW

Thanks for this feedback, especially the suggested definitions. Your thoughts will definitely be incorporated into the final version. (Also, we should probably be linking to Thinking Fast and Slow all over the Concepts site -- that was a useful reminder!)