SF Worldbuilding: Relationship Prediction Market

[DataPacRat]

Originally Posted by Gardner_Dozois

The job of a science fiction writer is to notice the car and the movie theater and anticipate the drive-in – and then go on to predict the sexual revolution.

I've literally just woken up from a nap with a SF idea I don't recall encountering elsewhere - and I'm trying to work out as many potentially plot-worthy consequences as possible. The local hive-mind has often served me well in this sort of thing, so I'm throwing the floor open to any further thoughts on this basic idea...

Real Background: Prediction markets are a way of harnessing the 'wisdom of the crowds' to get estimates of future probabilities that are more accurate than any individual or small group can manage on their own. For one standard method, if event X happens by date Y, and you hold a "share" of that event, then you're paid $10; if date Y goes by and it doesn't happen, you get $0. At any given time leading up to date Y, the trading price corresponds to the market's prediction of event X - if the price is $7.50, then there's a 75% confidence that X will happen.

Fictional Background: In my standard hard-SF future, the revolutionary Blues used such prediction markets to improve their public policies. After the Blues lost the revolution and became a mere political party, even suggesting such methods became enmeshed with "being a Blue", bringing the whole swathe of politics and political identity into whether or not it's actually a good idea. It's been near a decade since the revolution fizzled, and so the Blues have built up enough political capital to start de-criminalizing some types of prediction markets... Oh, yes, and just about every light-post, stop-sign, and bill-board includes a government-owned camera, tiny computer, and wireless router.

Idea: Some bright boy looks at the deregulation, and forms a start-up to open prediction markets in a previously unexplored sector, one that's the least amenable to computer analysis and requires the use of the greatest amount of human-level brains to consider: interpersonal relationships. Instead of buying shares that "Oil prices will rise above $X by DATE" or "The Oilers will win by X points on DATE", this market's shares are more along the lines of "Person X and Person Y will be in a close relationship on DATE", with 'in a close relationship' measured by any number of objectively verifiable criteria, such as "continue to live in the same household", "continue to make out in the movies weekly", and so on.


The above is the idea I woke with - working out the consequences is the fun part.

One simple consequence of the above: When two people are in love and want to spend the rest of their lives together, and they think they have inside information that nobody else has (ie, what they feel for each other), they can proclaim their confidence that they'll stay together by buying shares that they'll stay together at 1, 5, 10, 25, etc, years. Assuming somebody comes up with a clever way to deal with inflation for long-term contracts, then this could be as viable a way to store wealth as a bank-account or shares in a market-index fund, at least for those people who really do stay together that long. The financial incentives could encourage people who make such investments to work harder at their relationships than otherwise, thus encouraging long-term family stability even in the face of various crises, such as infidelity, which could be used to promote the market for people who prefer 'traditional, conservative family values'. However, the market's association with revolutionary Blueism could lead to Blueists abandoning 'traditional' marriage ceremonies entirely in favour of simply buying a block of stock in their continuing relationship. How those two threads interact... I've yet to figure out. :)


A more complicated consequence: the rise of what's effectively a "privacy tax", though not a government-mandated one. Reasoning: Insurance companies hate using pooled risk, and prefer to use every available scrap of information to personalize the premiums every individual pays. Once the relationship market has expanded enough to be better than any one individual's guess, then the market becomes a useful tool for predicting individuals' behaviour in certain regards, and thus can be used to finely slice the insurance company's predictions about who is destined for a messy breakup that could involve throwing dishes and increased risk of heart attacks, and who will stay happily together for decades. However, people who tend to keep to themselves rather than expose their personal pursuits in public will present less information for the market to act on, meaning any particular piece of information will swing the prediction more strongly, meaning their market price will be more volatile. One of the standard ways to treat volatile prices safely is to pretend it's a higher price than it's officially stated to be. Thus, insurance companies will treat privacy-oriented people as being riskier than they really are, raising the premiums charged to them even when those peoples' behaviour is otherwise identical to more extroverted individuals.


A random consideration: Some people think that one of the real spurs to species such as humans, chimps, parrots, and dolphins becoming especially smart compared to their neighbouring species involves a runaway evolutionary process, where there's an advantage for each brain to be as smart as possible compared to each other brain in the species; and that that runaway process is based on modeling the minds of other members of the group, such as, among other possibilities, to detect sexual infidelity. Coming up with computer algorithms that can compete with human minds in the area where human minds have been competing with each other for millions of years is likely to be one of the trickiest problems worked on yet - and should decent solutions be found, could very well be the arena that leads to the development of truly sapient 'general' AI.



... And that's as far as I've gotten so far. Anyone care to take any of these ideas and run with them? :)

[johndallman]

Originally Posted by DataPacRat

Instead of buying shares that "Oil prices will rise above $X by DATE" or "The Oilers will win by X points on DATE", this market's shares are more along the lines of "Person X and Person Y will be in a close relationship on DATE", with 'in a close relationship' measured by any number of objectively verifiable criteria, such as "continue to live in the same household", "continue to make out in the movies weekly", and so on.

Sounds like a good way to make a small fortune, if you start with a large one.

I can see this replacing lotteries, based on prediction of celebrity relationships, but I find it hard to believe that large numbers of people would be traded on this market. There are huge numbers of people around, and the number of possible relationships is so vast, and the fraction of them that any given person knows anything about so small, that there's a real problem with trying to find rational choices to invest in. How many pairings do you imagine a participant in this market typically having shares in?

[Flyndaran]

Very cutthroat as complete strangers will find ways to force couples to stay together... or break up at specific times.
Mother in Laws don't just dislike the husband. They have money riding on his screwing up. And if it takes hiring a woman to seduce him, or faking it, then she will.

[DataPacRat]

Originally Posted by johndallman

Sounds like a good way to make a small fortune, if you start with a large one.

The same can be said of just about any financial market, n'est pas?

I can see this replacing lotteries, based on prediction of celebrity relationships,

I'm not sure about replacing entirely, given the value in having a truly random lottery; but as a nice complement, maybe even a more popular one, sure.

(And we've got our first extra idea out of the gate. Have I mentioned that I love the local hivemind? I don't think there's a single idea I could post here that I couldn't get suggestions for improvement for. :) )

but I find it hard to believe that large numbers of people would be traded on this market. There are huge numbers of people around, and the number of possible relationships is so vast, and the fraction of them that any given person knows anything about so small, that there's a real problem with trying to find rational choices to invest in. How many pairings do you imagine a participant in this market typically having shares in?

It might be worth importing the idea of the 'monkeysphere', that a certain piece of a human brain is the right size to have a social group of up to 150 individuals. Given near-ubiquitious levels of top-down surveillance and bottom-up sousveillance, it shouldn't be that hard to identify the members of any particular individual's monkeysphere; and out of that group, many will be too old, too young, too closely related, the wrong gender and/or orientation, or otherwise have about as much chance of entering a relationship as a total stranger. As a semi-educated guess, I wouldn't be surprised if the average number of potential romantic partners at any given time is under 20, or under a dozen.

As for knowing enough to invest - I suspect that outside of computerized algorithms (somewhat equivalent to today's High-Frequency Traders) setting a rough baseline, most people would invest in predictions about the people they do have personal information on - those within their own monkeysphere. And if most of the folk in your monkeysphere have shared their knowledge about your relationships, in the form of bids in the prediction market... then that's plausibly the best prediction that can be made at all.

[DataPacRat]

Originally Posted by Flyndaran

Very cutthroat as complete strangers will find ways to force couples to stay together... or break up at specific times.
Mother in Laws don't just dislike the husband. They have money riding on his screwing up. And if it takes hiring a woman to seduce him, or faking it, then she will.

(The 'faking it' might be tricky - imagine if most folks' cellphones were constantly recording audio and video, stored in an offsite data centre, cryptographically secured so that only your permission or a government warrant could release the data. (I'm still working on the ramifications of this idea, but it's not the focus for this thread.))

This does bring up a thought: How large will the pool of shares in any particular relationship be? Eg, if it's a married couple who live out in the boonies without many neighbours, then there won't be many people who have any particular insight, and so there won't be much incentive to invest there when there's a potential for much greater profits in the highly-integrated urban environments. How much would anyone be willing to risk in regards to any particular relationship? This might limit how much effort anyone would be willing to exert to manipulate the markets' results...

Which brings up a possible further extrapolation: meddling mothers-in-law who invest against sons-in-law they simply dislike will fare poorly in the markets, and thus the social influence of that class will be reduced in direct proportion to the amount of cash they lose.

Which brings up some thoughts on further possible extrapolations: derivatives markets based on the relationship market. I haven't even checked Wikipedia for anything on present-day or historical derivatives yet, so don't have any particular ideas yet.

[Anthony]

In general, you have the problem that prediction markets don't work for everything -- wisdom of crowds helps eliminate bias, but it doesn't create information from nowhere. The other problem is that this particular market is basically designed for insider trading and fraud.

[DataPacRat]

Originally Posted by Anthony

In general, you have the problem that prediction markets don't work for everything -- wisdom of crowds helps eliminate bias, but it doesn't create information from nowhere. The other problem is that this particular market is basically designed for insider trading and fraud.

Hm... maybe pull a tulipmania (or Bitcoin, depending how that plays out) style boom, bubble, and bust, over a period of some years?

[Anders]

You can also make a killing by entering a committed relationship, betting a lot of money on it breaking up and then break up. Live apart for a month then move back together again. You probably couldn't pull that more than once though.

[Flyndaran]

Originally Posted by Anders

You can also make a killing by entering a committed relationship, betting a lot of money on it breaking up and then break up. Live apart for a month then move back together again. You probably couldn't pull that more than once though.

Springer made lots of money on exaggerations and lies such as those.
It's not always easy to tell acting from reality when some willingly commit embarrassing cringe inducing acts in public.

And many really do miss obvious cheating right under their noses.

[johndallman]

Originally Posted by DataPacRat

Hm... maybe pull a tulipmania (or Bitcoin, depending how that plays out) style boom, bubble, and bust, over a period of some years?

I'm having trouble believing you could reach critical mass to get the market running. But I also have trouble believing the stuff that people apparently put on Facebook.

Which suggests a way to start it going: introduce it gradually in your setting's equivalent of a social networking system. Pick one that's used by people with no sense about how much harm they can do to themselves.

Also be aware that there will be people who refuse to have anything at all to do with it.