Posts
Comments
Relevant data: https://x.com/wilhelmscreamin/status/1870804383283433731
my current guess is 15-20% of new users are already referred to the site because ChatGPT or Claude told them to read things here
Wow, this is higher than I would have expected! Do you have metrics showing this or otherwise what makes you think so?
Thank you!
Currently tags seem to only be useful for filtering your track record. I'd like to be able to filter the forecast list by tag.
I agree this would be nice. But try clicking on a tag to see all of your forecasts under it!
I think it's still very useful to be able to predict your own behaviour (including in the case where you know you've made a prediction about it).
Things can get weird if you care more about the outcome of the prediction than the outcome of the event in itself, but this should rarely be the case - and is worth avoiding, I think.
Cool idea! I am not sure you'd be able to move to real money betting given that cheating is trivial (just google the text of the article).
Here's an alpha version of a Firefox version!
If you run into any problems, it would be great to hear about them (e.g. by email).
most potentially dangerous capabilities should be highly correlated, such that measuring any of them should be okay. Thus, I think it should be fine to mostly focus on measuring the capabilities that are most salient to policymakers and most clearly demonstrate risks.
Once labs are trying to pass capability evaluations, they will spend effort trying to suppress the specific capabilities being evaluated*, so I think we'd expect them to stop being so highly correlated.
* If they try methods of more generally suppressing the kinds of capabilities that might be dangerous, I think they're likely to test them most on the capabilities being evaluated by RSPs.
We've added a new deck of questions to the calibration training app - The World, then and now.
What was the world like 200 years ago, and how has it changed? Featuring charts from Our World in Data.
Thanks to Johanna Einsiedler and Jakob Graabak for helping build this deck!
We've also split the existing questions into decks, so you can focus on the topics you're most interested in:
This should be fixed now (it was a timezone-related bug!)
I've made a basic version of Fatebook for Discord - you can install it here!
I've also added the ability to import your forecasts from a spreadsheet/CSV file, which I think is also useful for switching tools: fatebook.io/import-from-spreadsheet
This is now added (see below)
I've now added this! You can also see your track record for questions with specific tags, e.g.:
Thanks, fixed!
Cool thanks! For now you can use https://fatebook.io on your phone, do you think a native app would be much better?
Thanks for asking - I'm interested in doing iOS and Android apps! It'd be helpful to hear if other people are keen for this to help prioritise.
For now, one option is to add a homescreen shortcut to the website in Chrome.
Thanks! And thanks for mentioning this bug where the contents of comments were lost - I've now fixed this, comments made from now on should be recorded properly.
I've added a feature to export all your forecasts to CSV, thanks for the suggestion!
Currently it's not - just Slack and web. What would do you think you'd use it for in Discord?
I agree - I think data export is especially important for a prediction platform so you're confident making long-run predictions.
I'm planning to add import/export to spreadsheet, and maybe also to JSON, probably this week. If anyone has thoughts about the format of this data lmk!
Glad to hear it!
For PredictionBook users: you can import all your historical predictions to Fatebook here: https://fatebook.io/import-from-prediction-book
This is a great collection of tips! I think it's also worth explicitly noting that most of these strategies involve slowing down other people's orders, and many involve more inconvenience/stress for the sellers, so it's important to weigh this tradeoff.
If you're right, you will all be dead, so it won't matter
Posting a concrete forecast might motivate some people to switch into working on the problem, work harder on it, or reduce work that increases risk (e.g. capabilities work). This might then make the forecast less accurate, but that seems like a small price to pay vs everyone being dead. (And you could always update in response to people's response).
There are also Twitter Communities, which are relatively new. For example, there's an effective altruism one https://twitter.com/i/communities/1492420299450724353 and a (less active) forecasting one https://twitter.com/i/communities/1494107174519427072
Hits-based befriending and recursive curiosity both sound great!
I really enjoyed this post, it made me excited to try these techniques - which reminded me of my experience listening to a recent Clearer Thinking podcast episode with Steve Dean about developing lots of relationships.