The wisdom of crowds
From Yahoo.com:
Online gamers have achieved a feat beyond the realm of Second Life or Dungeons and Dragons: they have deciphered the structure of an enzyme of an AIDS-like virus that had thwarted scientists for a decade.
The exploit is published on Sunday in the journal Nature Structural & Molecular Biology, where — exceptionally in scientific publishing — both gamers and researchers are honoured as co-authors.
Their target was a monomeric protease enzyme, a cutting agent in the complex molecular tailoring of retroviruses, a family that includes HIV.
Figuring out the structure of proteins is vital for understanding the causes of many diseases and developing drugs to block them.
But a microscope gives only a flat image of what to the outsider looks like a plate of one-dimensional scrunched-up spaghetti. Pharmacologists, though, need a 3-D picture that “unfolds” the molecule and rotates it in order to reveal potential targets for drugs.
This is where Foldit comes in.
Developed in 2008 by the University of Washington, it is a fun-for-purpose video game in which gamers, divided into competing groups, compete to unfold chains of amino acids — the building blocks of proteins — using a set of online tools.
To the astonishment of the scientists, the gamers produced an accurate model of the enzyme in just three weeks.
Put those gamers on the FOMC! I wonder what gamers would predict if we asked for the amount of QE required to boost NGDP by 5%?
Tags:
20. September 2011 at 09:36
I saw this story too, and was absolutely shocked. Not that this discovery came from outside a laboratory, but that it was discovered manually, in a game, as opposed to all the shared systems folding during what would otherwise be idle system time (it’s a larger story, but you can sign your home computer up to do this number crunching, “folding”, when you aren’t using your computer yourself. Read here: http://folding.stanford.edu/). The gamers part, though, makes perfect sense to me. A perusal of the top performing folding “teams” shows that it’s mostly gamers who are most generous in sharing their computational horsepower: http://fah-web.stanford.edu/cgi-bin/main.py?qtype=teamstats
For people to unlock this mystery ahead of a global network of machines… I can’t even guess the appropriate analogy (nor factor of that analogy). John Henry outproducing one hundred steam drivers? Beating ten Watsons at simultaneous speed chess? John Connor beating Skynet in an arm wrestling match?
20. September 2011 at 10:13
I wonder what gamers would predict if we asked for the amount of QE required to boost NGDP by 5%?
They would predict infinity as QE has no bearing on demand, lending and NGDP.
http://bilbo.economicoutlook.net/blog/?p=6617
20. September 2011 at 10:43
We’d probably just start spamming “u mad bro”.
20. September 2011 at 11:13
Chacokevy, Yes, very impressive.
OhMy, I suppose that also applies to New Zealand. If so, then do it. New Zealand can buy up the entire world stock of government debt, and become the richest nation on Earth. Go for it!
Daniel, That pop culture reference went over my head, as I don’t keep up with much of anything any more.
20. September 2011 at 11:33
What we need is real pictures of economic flow at all levels and where they actually get backed up. I’m not a gamer but that’s a game I would play.
20. September 2011 at 12:25
The “gamers” are out there readjusting various and innumerable heterogeneous production processes taking more or less time, i.e. they are doing the part of economics that is completely absent from your “macroeconomic models”…
20. September 2011 at 12:42
@Scott_Sumner: There’s a big reason why Foldit efforts to find protein structures succeed, while an analogous one for the economy wouldn’t work at our current state of knowledge. This is because the rules atoms obey are simple and so a purported solution is easy to check; but there is an immense space of possibilities to search through, so you can’t just “brute force” it and harness computer power that way.
In contrast, to merely create the computer model of the economy detailed enough that you could (correctly) recognize a good monetary policy (because of how running the model churns out a success prediction) you would need as much power as the “brute force” option above — so there wouldn’t be anything for gamers to play.
Workable in principle, though.
20. September 2011 at 13:02
Greg Ransom,
How best to model depends on what you want to find out: if you want to find out the underlying logic of a system, then a very detailed system-specific model isn’t going to be useful. On the other hand, a very general model like a macroeconomic model sacrifices specificity for generality.
Where I think macroeconomic models can often go wrong is by having assumption-constrained results i.e. the results of the model are determined by assumptions rather than principles. If you compare a generic physics model to a generic economics model, the number of non-principle assumptions in an economics model is quite staggering and reflects the fact that economists have relatively few uncontroversial principles (within the profession) at their disposal.
20. September 2011 at 13:19
Becky, Easier said than done.
Greg, Agreed, which is why we need talent applied to monetary policy. Monetary errors are making life tougher for all those talented entrepreneurs. We need to stop creating turmoil, and go back to stable monetary policies.
Silas, I don’t want to rely on computer models, I want to stop relying on computer models.
20. September 2011 at 14:40
@Scott_Sumner: Wait, why did you think the gamers would have any insight into monetary policy without a computer model? You realize they didn’t have deep knowledge of biochemistry, right? They were just playing with a molecule that acted per known interatomic force interactions.
Do you actually understand what happened in this protein structure breakthrough?
20. September 2011 at 15:57
Silas,
I don’t think Scott ever meant for this post to be about bio-chemistry. The title of the post suggests this is all about EMH.
But I understand what you are driving at, and agree.
20. September 2011 at 19:14
SS,
You don’t understand the difference between nominal and real. Yes, they can buy all the debt, doesn’t make anybody rich in the real sense.
20. September 2011 at 23:52
Someone needs to write a book called “Packs vs. Herds” about the circumstances that cause some crowds to be far smarter than any individual in them (markets, scott’s example) and what causes them to be stampeding mobs (extraordinary and popular delusions, actual mobs).
21. September 2011 at 01:29
Silas,
It’s funny that you say the rules atoms obey are “simple.” Ever read anything about quantum mechanics?
21. September 2011 at 09:02
Just thought I would point out: This is a problem which, if one has a solution, it is easy to check whether or not is correct ( unlike in economics.) Thus, it is useful to have a large number of people trying different things and maybe someone will get it- and it’s easy to identify all of the failed attempts.
Doesn’t seem to imply anything about EMH, or whether or not we can predict NGDP.
21. September 2011 at 09:06
@John: Yeah, I have read about QM. Its linear, unitary, local rules are simple, just as Newtonian gravitation is simple. The predictive difficulty (in both molecules *and* the n-body problem) comes from the number of interacting entities, not the complexity of the rules. And in both cases, it is easier to verify that a given simulation is obeying the rules than to produce the simulation.
21. September 2011 at 09:07
Also, Andrew is right.
26. September 2011 at 08:43
Silas and Andrew, It was a metaphor for NGDP futures targeting.
OhMy, Only if it creates inflation, in which case your whole argument is wrong.