Any counterarguments to Joshua's spreadsheet? Any reason to believe those markets aren't all asking about the same thing, or that any of them were created with intent to make this market resolve YES?

@IsaacKing @JoshuaB

This one seems different:

https://manifold.markets/jonatanmagaji/will-donald-trump-contest-and-win-t?r=SmltSGF5cw

Looks like if Trump wins in a landslide such that he does not choose to contest the results of the election, this would resolve NO

@JimHays I think the word "contest" there just means that Trump will run, because it doesn't say he will contest the results. The first usage example from Google's dictionary actually happens to use it in exactly this way.

@pkpr Not sure if we’ll get a response, but I posed the question in that market to the creator

@JimHays There are still 20 excluding that one though, right?

@IsaacKing 21-29 had various colors representing some degree of difference. I haven’t looked through those to see how much the differences matter

Sorry for the delay in posting: https://docs.google.com/spreadsheets/d/1_p_smr1I5vPVqp7gcGODZajFgnX_1W2gaWDJs_Ebg30/edit?usp=sharing. There’s clearly 20 markets about Trump winning the Presidency

@JoshuaB I think multi markets are distinct. It makes sense to havve both

@Shump Isaac has said they do https://manifold.markets/IsaacKing/by-the-end-of-2023-will-any-realwor#6X8ypKJ85RRYA6hc3MRv

@JoshuaB https://manifold.markets/MrBonnison/will-trump-be-the-next-president seems like it could also be fulfilled by Biden being reelected and Trump winning in 2028 (there is no clarification in the market description or comments).

@Multicore Excellent point! I’ve edited the spreadsheet so that’s no longer counted

@IsaacKing two questions:

1. How much do market resolution timeframes matter?

e.g when Sam Altman was fired, if there had been 20 markets "Will we find out before X date that Sam was fired because he had bad breath?", where X ranged from one to seven years from now, would they be sufficiently redundant to count?

2. Does the real-world event have to take place in 2023? Or are 20 markets about a future real-world event ("will an asteroid kill us from 2024–2100 inclusive") fine?

1. I'd say probably not. Information does tend to come out over time.

2. Future events are fine, the markets don't have to be resolved.

I counted like 10 each of Biden nominated and Biden reelected. I think there's more than 20 "Will Biden live to X date" but not 20 for the same value of X.

Do Multi-choice markets count towards resolution? Do the old parimutuel markets count? What about markets N/A’ed nigh immediately for being duplicates?

@JoshuaB Hmm, I'm not sure how to count multiple-choice answers. This market was made before they existed I think.

@IsaacKing Wait, so have you decided whether they ought count?

Those can count if one of the answers is identical to a binary market I guess.

How would this resolve if there were 10 markets that were will X event happen, and 10 markets will X event not happen?

@JoshuaB That'll count, there's no real benefit to swapping the outcomes.

biden and trump aren't guaranteed nominees until sometimes in 2024

It's not about the Trump markets, but Sam Altman.

Here's a spreadsheet with...
20 markets that would resolve YES immediately if Sam Altman becomes permanent CEO of OpenAI, with nearly identical criteria just on different time frames (much like the Musk example, that was the explicit comparison in the description).
3 more markets with the same criteria, but that have already closed by now
11 more markets that would probably resolve YES or swing heavily if this were to happen (e.g. he's on the board again, he's CEO at end of 2023) or would resolve YES now but also could have done so by other triggers (e.g. somebody else became permanent CEO of OpenAI)
4 more markets that are adjacent in various ways and would resolve, but are not explicitly Sam Altman + OpenAI CEO.
6 more markets that are candidates but that I would lean pretty heavily against counting.

The wording "they just need to be similar enough that placing identical bets in all the markets is probably a decent strategy." from the description seems pretty close to cover about 34-38 of the above-mentioned markets.

I think this resolves YES.

@HenriThunberg Wow, I was vaguely thinking about having a thing in the comment about Altman, and was like, nah, Altman was too recent for so many markets to be made about it. Very nice find!

They all ask significantly different questions and it is plausibly valuable to have such markets. Some are 'will he ever come back', some are 'will he be on date', some are on weeks, some months, some years, some just 'will he return', some 'return as ceo'. The prices clearly vary based on timeframe

@jacksonpolack See the comparison to Musk, that also have very varying time frames.

I didn't include "Will he be on date" in the first 20+3, but think a case can be made that they should. The "by date X" however all resolve at once if he were to be reinstated.

In my 20+3 "Yes" markets, I didn't include questions such as ~"In what role would Sam Altman return", which would be yet more markets.

@HenriThunberg If you put all those markets in a 'portfolio' and bet no on them, you'd be making a very irresponsible financial decision. The true probability for many of those markets varies significantly (will he join by Monday versus will he join by 2028/2100 are very different).

I'd be very hard pressed to say they are duplicate markets.

Edited/deleted

I definitely see some motivated reasoning going on from my side and will try to stay level headed here. 🙃 Also might stop my break from work and come back to this in a few hours, which might help with that.

If they couldn't all be resolved in a single swing, I think this would be different. See for example stock market questions (Will X close higher today than yesterday), that are duplicates over various time frames but do not try to answer the same question – in this case: "Will he be back?".

@HenriThunberg thanks, more food for chooterbot

@HenriThunberg you are a gentleman and a scholar!

I see what you are saying about an event being able to cause them all to resolve yes, and the stock market questions are a good example of where you could see superficially similar titles but not have that be the case.

My argument is:
It would not be a "decent strategy" to place "identical bets" on either No or on Yes. In fact, for several of them you cannot (because they are already closed and will be resolving differently from the later dated ones in the now apparently 67% likely case he gets rehired).

@RobertCousineau haha yeah about that phrasing I am still trying to understand the whole decent strategy bit in the description to begin with. To me it seems like diversification would be higher and it being a better strategy the less similar they are?

Did the wording get wrong, and that it was supposed to be "opposing bets" to have the markets (assuming they have different % to begin with) meet in the middle, i.e. arbitrage?

@HenriThunberg regarding time frames of closing markets, then that will happen to all of them eventually. I was going to say that didn't make it cut either way, but trying to see this from a more neutral standpoint it should probably be "20 markets at any given time", and I now agree with you.

If we keep debating this a few days more, the number could either shrink or grow in the meantime, so I guess I just want to put it to the record that the 20 markets were not dependent on the 3 already closed ones.

@HenriThunberg I interpret "decent strategy" to place "identical bets" as meaning the following statement is true:
If I bet 100 on Yes or No in the following 20 markets, is my EV roughly as high as it would be if I placed 2000 on Yes in 1 markets that had the same liquidity as all 20 of them combined.

I do not think that would be the case in the list of markets you linked.

Agreed with respect to the currently existing number of markets fluctuating as time passes.

@RobertCousineau thanks for elaborating. That seems like a plausible way to interpret it, though not close and shut.

@IsaacKing, thoughts?

A 1km asteroid hitting San Francisco would cause a lot of markets on here to resolve to NO, but that doesn't mean those markets were all duplicates.

Glancing at the list (nice spreadsheet BTW), the time frames look pretty different. I think a reasonable observer could easily think the chance of Sam returning by the end of November is notably different from the chance they return by the end of 2024, so I don't want to count those as duplicates. The intent of this market was about redundant markets, and markets on different time frames for the same event are quite useful. For this to resolve YES, there need to be 19 markets that don't seem to have a clear purpose for existing, and could safely be folded into another market.

@HenriThunberg tiny difference between end of November 2023 and end of year 2024 markets, pretty much exactly what e.g. the "Sweden in NATO before 2024" markets (which indeed are identical) typically has at any given time.

To some extent indicates that people think the questions are actually very similar at this point. Marketd think if he's not reinstated pretty soon, looks very unlikely to happen in most time frames.

(Note that the Nov 22 market seems a bit different though.)

@IsaacKing I'll have to accept your view on this I guess.

A multiple choice market with the question "When will Sam Altman be CEO of OpenAI again?" that included the option Never, would indeed have been a very straightforward question and excellent way to neatly fold in all (?) of my listed markets.

Yeah, with the new multiple choice options, binary markets seem kind of unnecessary altogether.

Basically my position is that Manifold has been talking about adding a way to merge redundant markets together, and if such a feature existed, I do not think that there exist 20 markets about Sam Altman that could reasonably be merged.

@IsaacKing Alright, last attempt at this angle.

Have now marked column L in the spreadsheet for exactly 20 markets I think could be folded into one saying "When will Sam Altman be hired back at OpenAI?". This would mean merging "CEO at OpenAI?" with "Back at OpenAI?", which to be clear isn't the same thing (since there's actually a 80% market on this (I just bought it up to 83)) – but plausibly might pass the bar for resolution criteria not having to be exactly the same.

I in fact believe most question creators did not have this "CEO or not" distinction in mind, as we would probably have seen even more markets with the exact same timelines if that was the case. Maybe if 80% of the existing markets would still amount to 20, it would be more fair to resolve YES.

At this point, I am basically convinced that this should not resolve to YES, but it seems worth it to see if you perceive it differently than I do 🙃