How attention markets work

You get more attention if & only if you make good use of attention - a virtuous cycle

In February I made a prediction market (using play money) called “Will Ogi Ogas reveal a genuinely novel grand unifying theory on Monday, March 3rd?”. I did that because I felt like I had discovered someone who was brilliant, but overlooked.

I wanted to test:

1. Is this thing actually brilliant, and once people understand it, it will blow up¹?

2. It’s NOT brilliant, once people read it they will say it was NOT worth it

An attention prediction market lets you test this.

If it’s (1), and it does blow up as you predicted, people pay more attention to the next thing that you pitch. If it’s (2), you learn why people didn’t like it, and you lose some internet points. In the above case, it was (2).

---

How does an attention market work?

Say you’ve read (1)

Julian Gough

’s Blowtorch Theory (new model of cosmology that has made accurate predictions about what the James Web Telescope would see, that the standard model failed to make) or (2)

Stephen Skolnick

’s The Schizophrenia Gene (a theory that schizophrenia is caused by a specific gut bacteria, and that it is totally 100% curable). And you want to know, are they legit?

The responses to the schizophrenia article look like this:

“Holy shit you might have actually done it. If you're right about this you deserve a Nobel prize”

“How might I use this information to help someone I know (or test whether they could be helped)? Or test whether someone is at risk?”

For Blowtorch Theory:

“This feels like the foundation for a resynthesis of science and religion, just what humanity needs to burst into a new post-reductionist era.”

“Geoff, you are clearly riding on the breaking crest of the shockwave of what is likely to eventually be recognized as the most profound revolution in science to now”

Maybe they’re right, maybe not. How do we find out?

We can make a market that says: “Will <notable person> endorse this?”

The notable person you want to pick is someone that people trust to “stamp the truth”:

• Scott Alexander / Yudowsky, for the Rationalists/Effective Alturists

• Sabine Hossenfelder, for physics

• Matt Levine / Patrick McKenzie, for finance tech

• etc..

---

What happens if everyone starts doing this? Won’t this create a lot of noise & bother the notable people?

No.

Let’s think through how this plays out for both cases: (1) 🟢 valuable & overlooked idea & (2) 🔴 not a useful idea.

For (2) 🔴, like what happened to my market in February, it unfolds like this:

1. I pick Ben Shindel & Andrés Gómez Emilsson as my “notable persons to review” (I couldn’t go “any higher” because I didn’t have enough trust/social captial)

2. The market starts with ~100% “YES” prediction (because I am the only one betting on it)

3. A few people take a look, and as (almost) every person who looks at it decides it is unlikely that those notable people will endorse, the market falls to 5% hours later

4. Ben writes up a thorough review, which seals the deal.

5. I do not ask Andrés to look. He can save his time. We have our answer.

6. I end up losing all 500 points that I bet on to start the market

Let’s look at what it would look like if it was a case of (1) 🟢:

1. It starts at 100% (because I bet on it)

2. A few people look, they also bet “YES”

3. Now this becomes a “hot market”. A lot of people seem to think <notable person X> will endorse, if they saw it. This brings more attention to it, more thorough review, more scrutinity

If it survives the scrutinity, then we now have a win-win situation: this is either worth the notable person’s attention, and we’ve now established a mechanism to reach them, OR it is NOT worth their attention, but everyone thinks it is. Which means this is a blindspot that the notable person can fix. Then maybe Scott Alexander or Matt Levine will write about it in their next column, “it appears many people think idea X is revolutionary, but it isn’t, here’s why” and everyone learns something.

---

There are a few reasons why this scales really well, even if thousands or millions of people started doing this:

• There’s a very easy filtering mechanism to see the high signal markets: anything that falls to > 90% “NO” is likely not worth your attention (unless you’re looking for potential hidden gems).

• You could sort by who has a lot of market points, so those who have made good use of attention in the past get your attention first

• You need to bet on your own idea to start a market. If you keep putting up bad ideas for review, you will just run out of money / be essentially donating money to the market

• Instead of targeting busy high status people, anyone “under them” who has more time to spare, and wants to contribute can go in & either endorse it and “flag it up”, or reject it & give feedback

  • In fact, the notable person doesn’t actually need to respond at all if others handle it, like in my case with Andrés. Using their name serves as a useful pointer/anchor for the market

• Instead of notable people getting thousands of emails, potentially about the same exact idea, we can just group that all together so they’re more likely to see it, and the response is public so it benefits everyone

---

If you’re thinking about starting one of these, consider pre-reviewing it with a specific person first, so that in your market you can say <so & so> has already endorsed it, now I am trying to get a review from <such & such>.

I am happy to be a “pre-reviewer” if you think lending my name to your thing will help. I’ll either help you reach the notable person, or we’ll both learn why it was a bad idea and take the status/attention hit.

1. I never opened a prediction market for it, but I saw The Telepathy Tapes on twitter 3 months before it went mainstream. Regardless of what you think about its truth value, I was predicting that it had the shape of something that would cause a big stir. The fact that it had like 100 listeners on spotify at the time made me doubt the accuracy of my prediction. But I was right, just too early. I wonder if I had opened a prediction market, would that have caused it to blow up earlier than it did?