Calibrating The OT Regulars - Beat My Score

[rdsrds2120]

https://80000hours.org/calibration-training/

Batch 1, 90% confidence. No cheating. Engage with some rds nerd shit.

[DoomYoshi]

I don't understand what this is. Since it requires an account, I'm going to need at least a basic understanding before I'm inclined to investigate further.

[2dimes]

Uh, confidence testing Yosh, I'd say you have just been eliminated.

[rdsrds2120]

Damn 2dimes with the heat.

Basically, it's a project by ClearerThinkingProject. You take the test and it asks you, say, "What year did Elvis die?" Then you enter a range where you're 90% sure that, yes, Elvis did die between say, 1970:1980. The more confident you're asked to be (90% vs 50%) the wider your range should be since you're saying you have a stronger certainty about that range.

It's basically trivia with a belief testing metric. If you absolutely don't know what year Elvis died, you can use what little knowledge you have and just put something like 1900:2020 if you want. You'll still get a positive, but smaller, score for the question as long as it is right. The goal is to not be wrong a lot more than it is to be precise, but it's scaled, so putting in -9999:9999 for each answer will give you 0.

[DoomYoshi]

Ok, I got an answer of "Confidence Intervals, batch 1 results

100 questions

Points per question: 3.72 out of 10

Total Points: 372.07

Percent Correct: 86%

Answered: 100

Skipped: 0"

The problem is that most of the questions are about sports which only dubiously exist. As if a single person on the planet cares about something called "European Championship".

[saxitoxin]

Damn 2dimes with the heat.

Basically, it's a project by ClearerThinkingProject. You take the test and it asks you, say, "What year did Elvis die?" Then you enter a range where you're 90% sure that, yes, Elvis did die between say, 1970:1980. The more confident you're asked to be (90% vs 50%) the wider your range should be since you're saying you have a stronger certainty about that range.

It's basically trivia with a belief testing metric. If you absolutely don't know what year Elvis died, you can use what little knowledge you have and just put something like 1900:2020 if you want. You'll still get a positive, but smaller, score for the question as long as it is right. The goal is to not be wrong a lot more than it is to be precise, but it's scaled, so putting in -9999:9999 for each answer will give you 0.

Was your Magic: The Gathering game canceled this week or something?

[2dimes]

Ha, trick question. Elvis never died.

[rdsrds2120]

Ok, I got an answer of "Confidence Intervals, batch 1 results

100 questions

Points per question: 3.72 out of 10

Total Points: 372.07

Percent Correct: 86%

Answered: 100

Skipped: 0"

The problem is that most of the questions are about sports which only dubiously exist. As if a single person on the planet cares about something called "European Championship".

This actually doesn't matter too much - if you don't know about a certain subject, your confidence interval should reflect that and be wider. Your score might be a little smaller, but it's really only wrong predictions that set you back.

Was your Magic: The Gathering game canceled this week or something?

Online play has survived the pandemic, I'll have you know! The ZNR championship was this past weekend and I'd like to congratulate Barclay on piloting two underdogs through the Standard and Historic brackets this weekend. Good job, Barclay!

[rdsrds2120]

Ha, trick question. Elvis never died.

-9999:9999 is all we know