Comment by AlbertCory | Hacker News Reader

Comment by AlbertCory | Hacker News Reader

Old news with a catchy new name "futarchy".

In 2005 when I joined Google, they actually HAD prediction markets (no $$ involved) on business questions like "Gmail will have 50 million 7-day actives on April 1." It seemed so hip and with-it. Top management can get the real truth about things, rather than just asking their underlings!

The markets didn't take off, they withered. AFAIK they don't have them anymore. A friend of mine actually created an open source project for setting up your own prediction market. We can ask him how many users he got.

We've had betting markets on politics for a long time, too. I lost a few dollars on Amy Klobuchar in the 2020 primaries, although I had a 3x profit for a while (didn't sell, damn it).

So the interesting question isn't "is this a good idea?" but "why hasn't it taken over the world already?"

===== "Richard: Do you put futarchy in a larger intellectual tradition? Because a lot of people when they're coming up with an idea... Did you come up with this term by the way?

Robin: Yeah, and I've been ridiculed for it."

If there was no money involved it sounds like Google actually didn't have prediction markets. This is like saying that because nobody will take care of my lawn in exchange for Monopoly money lawn care businesses can't exist. It might be different if people got paid!

Agree, tangible incentives are important. But non-monetary prediction markets still have other benefits over traditional forecasting:

(1) Traders only make predictions in markets where they think they will gain;

(2) Traders control their risk, so one very confident person can make a larger bet than 10 people who are not very confident;

(3) Markets are continuously updated as new information arises;

(4) Profitable traders trade more, while losing traders exit the market

Prediction markets are probably not a panacea, but Google's various internal implementations of them (there have been several 20% attempts) are far from proof of this. If the people making the predictions don't have any actual skin in the game, the economic incentive to be correct is removed. That is the secret sauce that should supposedly make prediction markets more accurate than other forecasting venues.

People DO bet where "ego points" are the reward, and my own personal proof is the Hollywood Stock Exchange [1]. For several years, I won my movie Meetup group's Oscar pool, just by taking all the predictions from HSX. (They don't do all the minor awards, which is a problem.)

Actually using real money is a major legal hurdle, and that's why most prediction markets don't do it. The political markets have a special "research" dispensation, last I checked.

[1] https://www.hsx.com/

Isn’t this automatically inflationary? Just joining adds $2 mm Hollywood dollars to the pool. And you can join multiple times? But it does work clearly so I must be wrong about something.

You'd be surprised how motivated many Googlers are to be on a leaderboard of top traders, even without $!

Right. asdfasgasdgasdg knows nothing about prediction markets, the vast majority of which don't involve real money.

Not enough to generate routinely accurate, non-trivial predictions, apparently.

Another good example here is the Long Bets project, which I wrote the initial code for: https://longbets.org/

It has been an interesting exercise, and our largest bet was over $1m: https://longbets.org/362/

It has been a success as far as I'm concerned, in that it's still up and running almost 20 years later, and it has generated a lot of discussion about long-term topics, which is our goal.

But at the time I hoped it would turn into something much bigger. Lots of bets! Lots of money on table! Secondary bets! Maybe even options! That was definitely not the case. It has stayed niche, and so have prediction markets generally. I think there are good reasons for that.

Checked it out. I think this is a great idea! Why do think it's stayed niche? There IS real money at stake, albeit philanthropic donations.

My personal feeling: people do not want accountability, and to paraphrase Jack Nicholson in A Few Good Men, they can't handle the truth. I think that's the real reason the Google prediction markets died: managers prefer to do what they want to do, regardless of whether it's going to work or not. Perhaps someone can theorize as to why that's a good thing? I can't.

That is definitely a key component. Getting an actual bet requires two very dedicated and thoughtful people to negotiate terms and commit to possibly looking foolish down the road.

And I think you're also right about what managers want. I was very involved in the Lean Startup movement, which had a few years en vogue a while back. Core to it was being very disciplined about stating and testing hypotheses as quickly as possible. Used right, it can work extremely well.

It got some uptake in startups, but not as much as I expected. And even less uptake in more established companies, although the methods work well there too. Why? My take is that it's not just about managerial desire, although that's important, but also managerial status. For rising through the ranks it's much more effective to talk a big game than a modest and humble one, even though the latter traits are more likely to lead to success.

I hate it, of course. But we can look at examples like Theranos, Uber, and WeWork. None of them ever turned a profit, and it's possible none of them ever will produce a net return. The first CEO got caught eventually and might face some consequences one year or another. But the other two ended up incredibly rich. Who am I to say that bullshit doesn't pay?

Robin has been at this for over 15 years. Some reasons why he failed in the early years were that he used government money for uncomfortable bets about assassinations of US politicians.

I just can't trust a social scientist personally telling me that something offers better predictions, because social science has lost my trust in it's methodology. I'd need to see a data scientist, mathematician or statistician arguing that it does for me to start considering it more.

The good thing though is, he could simply setup a decision market to have people decide the true value of decision markets no?

good news, Robin wasn't always a social scientist, he has a background in physics and statistics

Robin happens to be something of a poster boy for criticizing social science methodology, as it happens.

Yeah, that's one of the few markets where I'm with the alarmists. You definitely can move the market by assassinating someone.

> The markets didn't take off, they withered. AFAIK they don't have them anymore.

Actually there is a new prediction market at Google now called Gleangen. It seems to be fairly active and flourishing despite not using real $$. Although there is a leaderboard, so perhaps the desire to want to rank in the top 10 is what motivates people to place bets.

I'm not looking for any corporate secrets, but generally speaking:

* what sort of things can bets be placed on? Real sensitive business stuff, or more mundane things?

You'll have to join to find out! j/k

Both. It could be something mundane like "Fed raises interest rates in 2020" or something company specific like "Nextgen pixel phone display size is > x inches". Generally the company stuff tends not to be the real sensitive stuff and tends to be more about fun/mundane topics.

The problem is mostly around the legal gray area of using them, and whether it constitutes gambling, instead of general buying and selling of assets.

If there wasn't a legal problem, I think they would have much more of an impact.

Futarchy also isn’t new. I remember it being a popular topic of discussion online 10 years ago, and apparently it dates back at least to 2007.

> no $$ involved

> The markets didn't take off, they withered.

This is surprising?

I'm not sure this is so clear. Google's first prediction market had 1,463 traders (>10% of the company at the time) and produced >250k predictions (though >50% of those were from trading bots).

The market was a surprisingly accurate forecaster too: http://static.googleusercontent.com/media/services.google.co...

Reply on news.ycombinator.com