Published on January 10, 2025 2:44 AM GMT
I'd like your practical advice and learned experience: how do you go about phrasing predictions when you ask people "Is X going to happen?" - in order to get answers that reflect the actual thing you're trying to get a read on?
Now I'll use a fictitious country with a fictious sitting president running for reelection - let's say I ask for the likelihood of:
"Will Sum Bodee[1] win the Ruritania Presidential Election this year?"
Even this isn't as straight forward as it seems because it could be motivated by either
a. a desire to know if he will remain President of Ruritania for whatever reason
b. a desire to know what do the people you're asking believe will be the outcome of the election, because you're not interested in the actual administration of the country, but you're interested in sentiment.
See what happens when I ask an intimately related question:
"How many votes will Sum Bodee win in the Ruritania Presidential Election this year?"
I think (and if you disagree - please speak up!) this makes it much more clear that the emphasis is less on "who will run the country?" but more on "who do people think will run the country?". Let me imagine some context, let's say that Ruritania was recently caught up in a spying scandal where the equally fictitious country of Buranda was caught red handed spying on members of Bodee's inner cabinet. Now this is where I get confused, because I feel like the phrasing I've given could be asking many different things in this context:
a. what is the impact of the spying scandal on Bodee's electability?
b. what do people think is the impact of the spying scandal, and how influential do they think topics like that are?
c. maybe I'm aware of some other scandals or issues, and curious about the awareness of people about them
d. Maybe I'm posing the question to Bodee loyalists and want to see if the scandal has changed their optimism at all?
And so on and so on... any given phrasing could have multiple intentions - so how do you match the intention to the right prediction?
Indeed, how should you format the prediction is an indication of what you want to know: how do you decide which format to choose?
I wonder if in fact I'm asking this question the wrong way around: perhaps I should come up with an example of a prediction I'd like to make, such as "How many music video commissions will I get this year?" and then ask "Is this how you would pose that prediction?" - Note! I'm not asking you to predict how many commissions I will get - I'm asking: how should I pose the question (to myself).
HOW DO YOU MAKE A PREDICTION
And that's just posing the question.
Being the start of the year there recently were a few prediction threads . and while the 'reasons' one states for giving a certain level of confidence are one thing, I would very much like to learn how to generate those reasons - how to get there. How do you do it? For example, put yourself on the other end of the prediction questions above - is this you?
"hmmm... yes, there are some interesting candidates running against Mr. Bodee in the Ruritanian presidential race this year? I wonder how many votes they will take from him?
Or not!? More generally when you're asked a question like "How much do you expect [politician's name] to win/lose by?" how do you go about generating your response? How do you generate your predictions?
What is your first instinct? How do you break down this question? I'm not looking for references or instructions. I'm looking for narratives of internal thought processes. How it looks like in practice.
- ^
Sorry about the lazy fake name. I'm alluding to an old political axiom “You can’t beat somebody with nobody” which I understand to mean that a personality or figure will always beat out even a well intended policy that can't be sold politically with a particular personality. Indeed, it seemed like it would be more effective to pose an example with a fake name and a fake country than to use nondescript variables.
Discuss
