Based on GitHub they are actually considering this, so get your refinement tools ready to argue why a particular version is better.

@Eliza @SG paying interest in shares seems so much better than paying it in mana and much less game-able -- interest on the market subsidy is so freaking important -- we WANT people to make markets

@Eliza @SG Yep. These bits are important and low risk. Whether interest gets paid on the cash balance or not is much less important.

@Eliza "we WANT people to make markets"

I want people to make markets they actually care about. That means making it more expensive. Creating a market should be the product that Manifold is selling.

@travis That would certainly kill the site, since that model only works with plenty of good predictors, who will not stick around if there's much less to trade on.

@SimonWestlake Yeah, I think most of the current activity would not be sustainable. I don't know if it would kill the site, but I agree it's probably not a realistic proposal. One ideal of prediction markets is like if you don't know what you're doing, you shouldn't be trading, so it might be inherently anti-engagement.

Currently, if you set the liquidity filter to 100,000+, you'll get a list of markets someone was willing to pay for. They seem to have a decent amount of activity even though some of them are fairly esoteric.

I want people to make markets they actually care about. That means making it more expensive. Creating a market should be the product that Manifold is selling.

One of the problems with this is that it's a classic collective action problem. Often lots of people care about the same market, but it's hard to coordinate and pool resources to make it. Markets with broad interest end up under-created and under-funded. Making it cheap, supplying house funding, and so on is a subsidy that can help, but the collective action problem remains.

The markets are classic public goods: most of the benefit accrues to people other than the creator, because they also care about the question.

I don't have good solutions here, though there are definitely small steps one could try. I'm just saying you shouldn't push too hard in that direction.

@EvanDaniel Yep, I check prices on polymarket every day, even though I don't have an account there so I'm part of the problem. I doubt there is any solution because people have been thinking about this for a long time. Nevertheless, I continue to find useful forecasts on real and play money prediction markets.

In the early days of Manifold, they were subsidizing market creation a lot more, specifically markets with lots of unique traders. At least in my opinion, the site has gotten better since they reduced those. Admittedly it's selfish, I just want to see useful forecasts. I don't know if it helped engagement or whatever.

@travis FWIW I consider myself the biggest advocate of "people should be creating markets they actually want the answers to, rather than just randomly spewing questions into the void". But we also need to recognize that putting 10s of thousands of mana into market pools for years on end is a big ask.

Maybe there's 3 kinds of markets:

1. Markets that are naturally high-engagement (mostly sports and politics).

2. Markets that someone is subsidizing because they actually want a forecast.

3. All other markets.

I guess on Manifold now 20% are type 1, 1% are type 2 and 79% are type 3. I think you could reduce type 3 without reducing engagement much, because it's mostly type 1 that drives engagement. Type 1 markets actually don't need to be subsidized at all. I think a lot of popular markets actually work better with less mana in the AMM. Manifold has a big advantage with type 2 markets because you actually can't do them at all on the real money sites. I guess there's no problem with having infinite type 3 markets, except that those markets are more likely to require moderator attention.

@travis A lot of markets would work better with less in the AMM, but the UI and/or trading incentive structure has to convince people to set up limit orders. They spent a long time purposely de-emphasizing limit orders and it has taken a while to come out of that.

@Eliza Yes, gradually I've learned to use limit orders. They're probably 80% of my profits now so the incentive is there, but it took me a long time to get comfortable with them.

Nice! I think you hit most of the important points here.

• You touched on this a bit, but Manifold has a wide spectrum of users from "very sophisticated" to "this user will surely bet all their mana on one market". The current loan system heavily favors more sophisticated users who have a coherent plan for managing their stack. The proposal for paying interest on positions probably gives a more appropriate treatment to the "average bettor" who doesn't want to think about those things.

• You wondered whether interest should be paid on current value or cost basis. The current 'loan' calculation uses min(investedAmount, currentValue) because the iteration before that was easily gamable.

• I didn't think about this too much yet but Maniswap's moving p would also enable you to pay interest directly in Shares rather than Mana. If users mostly hold Yes shares and the AMM holds No shares, then when you pay Share Interest, the users get more shares, the AMM gets more shares, and p shifts to keep probability the same. (I didn't take any time to consider whether this is actually better, just a comment that it's possible.) -- this at least provides some compensation to liquidity providers for tying up their mana indefinitely in AMMs, which is a severely lacking aspect of the site right now.

• Using interest as a prodding function to get users to reduce balances is exactly the kind of thing @SG is trying to encourage (I think).

---

Now, does it work to print mana like this on a play money website with no mana sinks? Yes.

Would adding more mana sinks make it even better? Yes.

Would users be upset if this reduced or turned off 'loans'? Probably yes.

Would this convince people to pay more money to buy mana? Not sure.

@Eliza

(I didn't take any time to consider whether this is actually better, just a comment that it's possible.)

Alright so some quick things I see:

• Market creators or liquidity providers get some compensation.

• It doesn't directly increase your balance, so you need to trade actively if you want to use it for balance growth.

• It's like automatically re-investing in a market. Does this encourage you to keep the probability more accurate? (Could the system automatically place market orders when your interest accrues if you wanted to take your interest as mana instead?)

• If you want to sell your accrued share interest for mana, you need a trading partner (a person or the AMM), but that also means you get a more appropriate mana valuation?

• If you're a No holder in a 99% market, you don't get to keep betting it to 98% with your daily loan every day (also obviously the size of daily interest would be significantly smaller)?

@Eliza @EvanDaniel come eviscerate me

@Eliza will read and respond more thoroughly later, but:

Did you see my other post?

https://manifold.markets/post/margin-loans-are-good-and-complemen?r=RXZhbkRhbmllbA

Paying interest on assets, in the same asset, works great with AMMs: just pay interest to the AMM pool. Both sides of the pool grow the same amount, liquidity goes up, p value and probability stay the same because it's just a multiplier on the pool size.

@EvanDaniel

just pay interest to the AMM pool. Both sides of the pool grow the same amount, liquidity goes up, p value and probability stay the same because it's just a multiplier on the pool size.

Ah, of course, that's even better naturally. So you're saying "yes, do the thing you said, it's even easier than you thought".

@Eliza Yep :)

I'm just not sure how to effectively continue advocating for "pay interest on positions". We have been trying to make that happen for several years and it's never felt like we got even the tiniest bit of interest.

@Eliza I guess you didn't see this comment: https://manifold.markets/post/loan-changes#ZqPsSy5N

@SimonWestlake That is definitely related to the efforts Evan has been making.

@Eliza I wasn't trying to imply otherwise, so I guess I didn't understand what you meant

@SimonWestlake Basically we have tried things very similar to this post many many times but it never goes anywhere. I do not know what other approaches to try to be more convincing.