Cool and excellent! I hope we can get a bunch of folks.
I want to explain a little bit about what I’m wanting to test, so you all know that we’re being fully transparent and not running a mind control experiment. (Those come after we get out of beta. I’m kidding.)
OK, so I’ve been doing a little bit of sub-contracted work for UNU, which is an artificial intelligence based on human “swarms”. The idea is this: in nature, swarms and small colonies of organisms tend to work REALLY well as a group intelligence. See bees, for instance.
The idea behind UNU is that humans might have some of that. The idea isn’t mob mentality or herd mentality by any means. It’s more and idea that small groups of people may be more intelligent collectively than as individuals.
We’ve tried this out in a variety of predictors. When we do really well, it’s because it’s something that is more than just random chance.
Which…baseball. My gosh. You’re nuts if you bet baseball, because baseball typically makes no sense. Gajillions of variables, right? Crazy stuff happening all the time. Rain delays. Nagging injuries. A ball that lands two inches fair or foul. But…over 162 games, the best teams tend to overcome the random chance and win consistently.
Can we predict single instances of that? I honestly don’t know. But I want to see how close we get.
-
Forget having to know tons about teams or specific games. If you’re an expert, great. If you’re just playing a hunch, great too! The swarm mentality takes that into account, or should.
-
We’re not debating and making points. Sure, some of that may happen…but you’ve got 90 seconds to come to a group consensus. You may start by really trying to push things to the Mariners beating the Athletics…but if the room is hung in indecision slightly favoring the A’s, you might switch and go pick them just to help get a consensus answer. That’s how it’s supposed to work! People switch back and forth all the time, and that’s how it goes.
-
We’ll try to keep things moving fast. I may ask everyone to fill out a quick survey where you all pick the games as individuals, so we can see if any one person outperformed the group. That should go really quickly. No personal information, yada yada yada.
OK, so here’s how the mechanics of this work.
You go to our site, and sign in as a guest. Pick a username. You can pick any username you want. Doesn’t have to be your qt3 handle. That’s it. No email, no other info needed.
Next, go to the “room” I’ll direct you to. I’ll probably password protect it with something easy.
In the room, there’s a chat stream on the left side. There’s an "arena filling up most of the screen. It has a big glass “puck” in the middle. There’s a set of choices marked by terminals underneath the puck.
I’ll put possible answers on the terminals. (“Who wins this game, Athletics or Mariners” and list starting pitchers) One dot will be the Mariners, one dot is the Athletics.
The goal is to pull the puck to center over one of these answers, kind of like an internet Ouija board.
How does that happen? When you put your mouse pointer over the arena, it changes into a little magnet icon. Move the magnet near the puck, and it pulls it towards you. Move the magnet close to the puck, and it pulls stronger.
Here’s the fun: everyone has a magnet, and everyone’s pulling the puck, sometimes in wildly divergent directions. That’s where the need to compromise and come up with consensus comes from.
Here’s a replay of one of our more coherent swarms tonight.
Here’s another.
Not gonna lie–most of the time the open rooms are lightly moderated and tend towards millennials from either reddit’s pol boards or 4chan’s politically incorrect boards (always fun when they get together!) You’re welcome to visit any time you like and play around, but if the room isn’t moderated, you’re going to flee screaming. Don’t say I didn’t warn you. We’re still very much in beta.
Anyway, when we do this for realsies, we’ll ask a couple of questions to get everyone familiar with the interface and how it works, and then proceed to ask the outcomes of 6 or 7 games. I’ll probably ask both straight up and against the spread for all. Then that’s it!
I’ll post our predictions, and we’ll see how we did!
If you know anyone who might be interested, feel free to invite them. The more the merrier, although probably up to 40…but honestly I don’t think we’ll get that many.12-20 will give us a decent swarm, and I’m recruiting at a couple of different places to get a good cross section sample.
