FAQ - Frequently Asked Questions #
Why care about forecasting? #
Forecasting is an important skill. Essentially all decisions we take are based on vague gut-level predictions. For example before leaving for the airport you predict how much the traffic will delay you on your way, timing your departure appropriately. You also make a prediction about how likely you think it is your car will suffer a catastrophic engine failure on the road. If you thought this was very likely to happen, you would take a cab instead.
Making such forecasts explicit and tracking your accuracy can be incredibly valuable to hold yourself acountable (i.e. what did I really believe back then?), to communicate more clearly with other people and to discover systematic ways in which you are erring (e.g. you tend to be too optimistic concerning your favorite political party).
Some resources to learn more about forecasting:
- Superforecasting: The Art and Science of Prediction (book)
- The Signal and the Noise: Why So Many Predictions Fail — But Some Don’t (book)
- Thinking in Bets: Making Smarter Decisions When You Don’t Have All the Facts (book)
- Superforecasting | Philip Tetlock (video/talk)
A future event either happens or it doesn’t, does it make sense to attach a probability to it? #
Many people find the idea of attaching a probability to a future event to be counter-intuitive. That is because one way of thinking about probability is when repeating an event many times. For example if you flip a coin 10.000 times it will land about 5.000 times on heads. Therefore we say there is a 50% probability of getting heads. Some events can’t be repeated however. What is the likelihood of the US president stepping down in the next year due to a conflict of interest with the rice industry? How do you repeat this event 10.000 times?
A more helpful way of thinking about probabilities is belief (this is called the Bayesian interpretation of probability). How likely do you believe something is going to happen (0 being you are certain it will never happen and 100 you are certain it will definitely happen)? If I flip a coin, how strongly do you believe heads will come up? Given your prior experience in the world you will probably believe 50% probability of heads. Or maybe you think I am the kind of person who has trick coins in their pocket and then you would only believe 10% probability of heads. The same can be applied to any future event. The strength of you belief can depend upon many things, and much more importantly, will vary over time because you get to see new data (evidence)! So maybe your belief that the US president will have to step down due to a conflict of interest with the rice industry is 0.001% . When you learn that he in fact owns large amounts of stock in rice companies your belief will possibly increase to 5%. Cleodora will help you keep track of your beliefs and once you find out the truth, how often you were right or wrong. This will aid you in forming more accurate beliefs in the future.
Yet another way of thinking about such probabilities is, how much money would you be willing to bet on it? For example if you believe there is a 5% chance the US president will step down in the next year due to a conflict of interest with the rice industry, then you consider the following bet fair: You bet $1 on YES and your friend Mindy bets $19 on NO. If it turns out you were right, you get her $19 and if it turns out she was right she gets your $1. You would not take the bet if she was only offering $15 and you would definitely take the bet if she was offering $20 . How this is calculated exactly (5/95 = 1/19 = 1:19 odds) does not matter, only that you have this intuition that if you think something is very likely to happen you would be willing to bet a lot of money on it and vice versa.
Why the focus on running Cleodora on your own computer? #
To enable you to make deeply personal predictions (e.g. will my uncle find a new partner within the next 6 months?) without worrying about who gets to (accidentally) read your thoughts. Also it emphasizes the fact that the tool is about YOUR forecasts, not other people’s.
Given that it’s a web application you will be able to host it yourself e.g. on small home server and access it from any device within that network.
Later versions of Cleodora will also have a public instance with multiple users, but the focus will always be on personal forecasts.
Can’t you just use a spreadsheet (Excel, Calc, Sheets)? #
Yes and no. If you just want to represent a list of forecast and for every forecast the same possible outcomes (e.g. Yes/No) with a probability then it’s trivial to implement in a spreadsheet because it’s two-dimensional data. But what if for each forecast you want to predict different outcomes? And if you want to update the probabilities you estimated over time and keep a record of the previous probabilities? And if you want to sort the forecasts into different categories? You end up with multi-dimensional data that is not trivial to represent.
It also becomes harder to focus on specific forecasts, searching and filtering, seeing forecasting ability within a certain category and such features that are or will become part of Cleodora.
Why the name Cleodora? #
Cleodora (/kliːəˈdɔːrə/) is a nymph from Greek mythology who was able to predict the future by casting stones. She and her sisters Melaina and Daphnis are known as the Thriae and were associated with the Oracle of Delphi.
Cleodora is the mother of Parnassos (eponym of the famous Mount Parnassos, Greece).
She is mentioned in “Description of Greece” by the traveler and geographer Pausanias (c. 110 – c. 180).