3 ms·

One thing this makes me think of is "Oracles" that could potentially decide a market. Someone could write an oracle, and open source the code, and MM could then

One thing this makes me think of is "Oracles" that could potentially decide a market. Someone could write an oracle, and open source the code, and MM could then run the oracle for the lifetime of the market. The oracle would be responsible for resolving the bet and would turn an ambiguous statement into a well defined one. i.e. What you are betting on, ultimately, is how the linked oracle will resolve the bet, and everyone can see the code. The first bet I see is about Russia invading Ukraine. We could have an oracle that checks Wikipedia's list of Russian military engagements every day and if there is a new one in Ukraine and it stays for five days that might be a sign. Another sign might be scraping New York Times headlines for a list of terms you might expect to see in an invasion. Maybe there's a website tracking Russian troop deployments, or Russian flagged planes and ships and those could become signals. Then, the oracle collects and judges based on its signals.

akrolsmir 5y ago
Yeah, oracles are a common solution to the question of "how should the prediction market get resolved". We think there are some usability problems with oracles in practice, which is why most prediction markets (e.g. PredictIt, Kalshi) still end up relying on platform resolution. That said, we'd be very happy if anyone was interested in setting up an oracle on our Manifold! "User-resolved" includes the possibility of "user publicly commits to resolving all markets via an automated, predefined oracle".
jahooma 5y ago
Oracles are cool, but I humbly submit user-resolved markets as a cleaner and better mechanism. As akrolsmir says, someone could commit to resolving according to an Oracle. But mainly, it allows for a diversity of approaches to serve different user and subject-matter needs. Some markets can be resolved subjectively, others need to rigorously specify each case. Plus, the real world is often ambiguous. You need someone you trust to make the right judgment call.
- ALittleLight 5y ago
  User resolved simply won't work if people care about winning their currency. And people will care if they ever use real money or become popular. Imagine - I create a market "Donald Trump will be speaker of the House in 2022" and people bet on that. I, through sock puppet accounts if necessary, bet that he will as much as possible. Then, I just resolve the bet so that I win. Cheaters will have all the currency and cheating will make the predictions useless or worse than useless.
  - jahooma 5y ago
    It sounds plausible, but our experience so far is that cheating is very rare. Out of hundreds of markets, we had our first clear incorrect resolution yesterday. But the argument is not simply that people won't cheat often. The argument is that people will choose to bet in markets where the creator is trusted. Even if the majority of creators cheated, users would react appropriately and flock to the ones who don't. It's just like how you purchase goods and services. They often have a chance to keep your money without providing what they promised. But in practice, they can't run a long term business that way. By the way, even in our play-money currency, the market creators earn a fee on the traders' profits.
flak48 5y ago
I bet this specific example will be tricky, invasions/incursions now a days seem to be usually denied by the aggressor in various ways, i.e. there might actually be a debate, 'Has Russia actually invaded Ukraine yet?'
noduerme 5y ago
>> turn an ambiguous statement into a well defined one A well-designed series of forms would take away the friction of having some dude come in later and tell a moron what they meant to say. Unfortunately, a form would also deter morons from creating contracts. Ah. But what we have here is a would-be casino operator with a short-term view, who would love to have their site full of contracts, even if they're free-written by morons. God-for-fucking-bid we deter some idiots from starting contracts by making them write down logical rules. Let's get as many idiots as we can! It ain't well thought out. It's not designed to be able to deal with the public. If it makes any impression at all in the media, it will just cast more negative shade on prediction markets, because the people behind it didn't think for a second about how humans fucking work. Actually, they're a prime example of dumbass greed themselves, since they did it as a result of a stupid idea winning a grant to build...a stupid site. Wonderful. GO MAKE SOME BETTER GENETICALLY MODIFIED RICE, YOU OVERPAID LAZY FUCKS.
HWR_14 5y ago
> We could have an oracle that checks Wikipedia's list of Russian military engagements every day and if there is a new one in Ukraine and it stays for five days See, that's the issue. Wikipedia already has the Russian-Ukrainian war, because Russia has been occupying Crimea for years. Wikipedia as a source of truth when money or power is on the line is also difficult. That's why topics frequently get locked. The NYT would work as a source, but keyword matching could be iffy. A nominally third party site (again, if real money is on the table) could easily be a fake set up by someone heavily invested in one of the two answers. That shouldn't be used to say "ground truth isn't available", I'm just not sure it's machine readable and scalable yet. It's possibly I'm missing something obvious.

akrolsmir 5y ago

Yeah, oracles are a common solution to the question of "how should the prediction market get resolved". We think there are some usability problems with oracles in practice, which is why most prediction markets (e.g. PredictIt, Kalshi) still end up relying on platform resolution. That said, we'd be very happy if anyone was interested in setting up an oracle on our Manifold! "User-resolved" includes the possibility of "user publicly commits to resolving all markets via an automated, predefined oracle".

jahooma 5y ago

Oracles are cool, but I humbly submit user-resolved markets as a cleaner and better mechanism. As akrolsmir says, someone could commit to resolving according to an Oracle. But mainly, it allows for a diversity of approaches to serve different user and subject-matter needs. Some markets can be resolved subjectively, others need to rigorously specify each case. Plus, the real world is often ambiguous. You need someone you trust to make the right judgment call.

ALittleLight 5y ago
User resolved simply won't work if people care about winning their currency. And people will care if they ever use real money or become popular. Imagine - I create a market "Donald Trump will be speaker of the House in 2022" and people bet on that. I, through sock puppet accounts if necessary, bet that he will as much as possible. Then, I just resolve the bet so that I win. Cheaters will have all the currency and cheating will make the predictions useless or worse than useless.
- jahooma 5y ago
  It sounds plausible, but our experience so far is that cheating is very rare. Out of hundreds of markets, we had our first clear incorrect resolution yesterday. But the argument is not simply that people won't cheat often. The argument is that people will choose to bet in markets where the creator is trusted. Even if the majority of creators cheated, users would react appropriately and flock to the ones who don't. It's just like how you purchase goods and services. They often have a chance to keep your money without providing what they promised. But in practice, they can't run a long term business that way. By the way, even in our play-money currency, the market creators earn a fee on the traders' profits.

flak48 5y ago

I bet this specific example will be tricky, invasions/incursions now a days seem to be usually denied by the aggressor in various ways, i.e. there might actually be a debate, 'Has Russia actually invaded Ukraine yet?'

noduerme 5y ago

>> turn an ambiguous statement into a well defined one A well-designed series of forms would take away the friction of having some dude come in later and tell a moron what they meant to say. Unfortunately, a form would also deter morons from creating contracts. Ah. But what we have here is a would-be casino operator with a short-term view, who would love to have their site full of contracts, even if they're free-written by morons. God-for-fucking-bid we deter some idiots from starting contracts by making them write down logical rules. Let's get as many idiots as we can! It ain't well thought out. It's not designed to be able to deal with the public. If it makes any impression at all in the media, it will just cast more negative shade on prediction markets, because the people behind it didn't think for a second about how humans fucking work. Actually, they're a prime example of dumbass greed themselves, since they did it as a result of a stupid idea winning a grant to build...a stupid site. Wonderful. GO MAKE SOME BETTER GENETICALLY MODIFIED RICE, YOU OVERPAID LAZY FUCKS.

HWR_14 5y ago

> We could have an oracle that checks Wikipedia's list of Russian military engagements every day and if there is a new one in Ukraine and it stays for five days See, that's the issue. Wikipedia already has the Russian-Ukrainian war, because Russia has been occupying Crimea for years. Wikipedia as a source of truth when money or power is on the line is also difficult. That's why topics frequently get locked. The NYT would work as a source, but keyword matching could be iffy. A nominally third party site (again, if real money is on the table) could easily be a fake set up by someone heavily invested in one of the two answers. That shouldn't be used to say "ground truth isn't available", I'm just not sure it's machine readable and scalable yet. It's possibly I'm missing something obvious.