Re: critiques on https://astralcodexten.substack.com/p/mantic-monday-82823, another way to boost the "voting power" of superforecasters would be to adjust the loan policies based on factors like:
how long has the user been on the site
what's the user's current profit
what's the user's current calibration grade
This idea was proposed in discord here: https://discord.com/channels/915138780216823849/1145904687170723933/1145904687170723933
Related market with some relevant discussion: https://manifold.markets/SG/will-manifold-lower-the-loan-rate-f
This market will resolve YES if the Manifold developers say the loan rate varies on a per-user basis at some point between now and January 1, 2024 (even if this variability is later turned off or if the conditions are unrelated to what's proposed above).
resolved this to N/A due to reasoning in https://manifold.markets/AndyMartin/will-manifold-use-a-loan-rate-that#LOodr7wGd8zHB9TjDpPb
@AndyMartin I think this was good! I know people try to save questions and resolve them YES or NO when at all possible, but I kind of think N/A should be on the table more often in cases like this, I think this was a good use of it.
@AndyMartin, the loan rate itself is 4% for everybody, but the amount of mana eligible to be loaned from each market is (since a few months ago?) capped per market, such that if you've made a loss on that market, you won't get all your mana loaned back. But of the amount you will get, you'll still get 4% of it each day.
Also, if you're in negative net worth as a whole, you don't get any loans - though I don't know if that was already the case when this market was created or not.
Could you clarify if any of that is relevant to this market?
@chrisjbillington Hmm yeah good question
Happy to hear feedback/disagreement or give "yes will be relevant" / "no would not contribute to a resolution" answers to other examples, but here are some general thoughts-
> the amount of mana eligible to be loaned from each market is (since a few months ago?) capped per market, such that if you've made a loss on that market, you won't get all your mana loaned back. But of the amount you will get, you'll still get 4% of it each day.
It seems like this is driven by the user's participation in an individual market, which is far enough away from the user themself that I'd lean towards "no doesn't not contribute to a positive resolution"
> if you're in negative net worth as a whole, you don't get any loans - though I don't know if that was already the case when this market was created or not.
Oh interesting - I didn't know about this policy. When I originally made this market, I was interested in whether they'd make it more dynamic and possibly raise the top rate, but it's harder for me to rule this one out.
It seems reasonable to not count this if it was in place before I opened the market but count it YES otherwise, since "what's the user's current profit" is pretty similar to looking at the net worth.
@AndyMartin Ah, so looking into it, actually it was previously based on net worth (portfolio value plus balance > 0), and now it's based only on portfolio value needing to be positive. So that means you stop getting any loans if your investments are worth less than what you already owe in loans, regardless of how much balance you have.
That change was made in September:
https://github.com/manifoldmarkets/manifold/commit/4bd39ef77b9198efcacbceb93bb86d1f1f745edd
@chrisjbillington I was going to say because that's based on only the user and not specific to individual markets, that seems good enough for a "yes" resolution, but it seems like the prior logic (balance + investmentValue) would meet this criteria too and had been in place before this market started
leaning towards N/A or saying this market covers significant expansions beyond the previous & existing logic, but I'm open to people's feedback
@AndyMartin I interpreted the question as "will manifold adjust the percentage given based on user data", and I don't think that has been fulfilled. Entirely up to you though.
@AndyMartin up to you, I can see the argument for YES and for NO, and am not averse to markets NAing in circumstances like this either. Appreciate the effort to try to make it fit, writing resolution criteria is hard.
@chrisjbillington What is the per market cap? That's an extremely impactful change that people really need to know about.
@IsaacKing The amount eligible to be loaned was changed from cost basis to the minimum of the "sum of bets" (not quite the same as cost basis), and the mark-to market value of your investment. So if your investments are falling in value, you get less loans, and you get none from a market where your shares are worth less than the amount already loaned out.
The "sum of bets" part exists for good reason (to prevent you being able to inflate your cost basis), but has the unfortunate effect that if you sell shares to realise profit, and then buy back in again later, you won't get loans for the newly made bets.
@chrisjbillington I know, I'm asking about the hard cap that you mentioned above.
the amount of mana eligible to be loaned from each market is (since a few months ago?) capped per market
@IsaacKing That per-market cap is the same as what I described in my most recent comment - the amount of mana eligible to be loaned back from a given market is capped to the mark to market value of the shares. (And also capped to the sum-of-bets number, whichever is smaller).
There is also an overall check such that a user doesn't get any loans unless their total portfolio value (net of loans) is positive.
@chrisjbillington ...But that number can be arbitrarily high? By that logic, "you can't get lent more mana than you put in" is also a cap, so loans have always been capped.
@IsaacKing Perhaps my use of the word "cap" was misleading, but since eligible loan amount per market is min(condition_a, condition_b)
, I'm thinking of both conditions function as caps on how much mana you can get out compared to what you put in. Sum-of-bets kind of replaces the cost basis as used previously, whereas the "can't be negative value" requirement is more conceptually new. But both can limit how much mana you get back compared to what you put in, whereas you used to eventually get back 100% of the mana put in.
Perhaps "cap" to you implied something that would always be a limiting factor, instead of dependent on the context which is the case.
@Loppukilpailija great question - no, A/B tests won't count. If they first A/B test a variable rate system and later make it the default/standard/active policy for all users, it will be the latter date that triggers a "yes" resolution.