I'm in the market for some nerd stuff to read, some sort of treatise.
Stuff I've found thought-provoking in recent years:
The Strategy of Conflict (Schelling)
Leviathan (Hobbes)
Inadequate Equilibria (Yudkowsky)
Information Theory, Inference, and Learning Algorithms (MacKay)
And stuff I haven't:
Debt: the First 5,000 Years (Graeber): too polemic, felt not like something a person with good epistemic hygiene would write
Better Never to Have Been (Benatar): afaict never addressed my main glaring objection (harm/joy asymmetry)
Scout Mindset (Galef): all stuff I was familiar with from Less Wrong etc.
The Signal and the Noise (Silver): I forget!
Each arm of this market resolves as soon as I read the book or decide not to:
YES if I read the book and find it thought-provoking
NO if I read the book and don't (or if I put it down after reading the first ~30 pages and never pick it back up)
maybe PROB if I think a book is borderline
N/A if I decide not to read the book, or on 2024 Feb 29 for any book I haven't read yet
1,000🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ9 |
People are also trading
@OptimizationProcess Something involving randomly resolving the ones you don't read with some distribution? Or you could pay some bonus to traders when you n/a something they traded on
randomly resolving the ones you don't read with some distribution
I think "resolve it with some distribution" is equivalent in EV to "resolve it to some PROB"? And, if that PROB is above my "should I read it" threshold, then there's still no incentive for anybody to bet it down from there; while if it's below my threshold, there's incentive for people to bid it down to the PROB even if that doesn't reflect their true belief about whether I'll find it thought-provoking. ...I think?
Another possible solution: if I decide not to read a book, force myself to read it anyway with 10% probability, and resolve the market accordingly? But this doesn't eliminate the problem entirely.
pay some bonus to traders when you n/a something they traded on
Hmm, I don't think that solves the incentive problem, but it does... seem fair...
@OptimizationProcess You could pre-commit to reading a certain number of books that the market predicts you won’t like, ensuring that there’s always a chance for a “No” outcome to be tested. This could be randomized or based on a fixed percentage (e.g., for every ten books, you must read at least one that the market predicts you won’t like).
@OptimizationProcess I think one correct solution is, for each individual leg of the market:
- 90% probability of resolving N/A, though you might still read the book if the market thought it was good
- 10% probability of you reading it (or reading the first, like 50 pages?) and resolving the market, even if the market predicts that you'll hate it
Then you're only committing to read 1/10th (in expectation) of the books posted while keeping the market incentives correct.
@WilliamEhlhardt It wouldn't be that crazy to just promise to read the first 30 pages of every book?