1/
This is a beautifully written thread about “the wisdom of crowds”, inspired by this piece by @mjmauboussin:
operators.macro-ops.com/wp-content/upl…
This is a beautifully written thread about “the wisdom of crowds”, inspired by this piece by @mjmauboussin:
operators.macro-ops.com/wp-content/upl…
2/
One question I have: is crowd wisdom better at estimating “derived quantities” or it is better at estimating “fundamental quantities”?
One question I have: is crowd wisdom better at estimating “derived quantities” or it is better at estimating “fundamental quantities”?
3/
Best to take an example.
Suppose we have a room and a tile.
We want to estimate how many tiles are needed to tile the walls of the room. Then:
No. of tiles needed = Surface area of room / Surface area of tile.
Best to take an example.
Suppose we have a room and a tile.
We want to estimate how many tiles are needed to tile the walls of the room. Then:
No. of tiles needed = Surface area of room / Surface area of tile.
4/
Here, the “fundamental” quantities are the room’s surface area and the tile’s surface area.
The “derived” quantity is the number of tiles needed — a function of these fundamental quantities.
We have 2 ways to crowdsource this: the “fundamental” way and the “derived” way.
Here, the “fundamental” quantities are the room’s surface area and the tile’s surface area.
The “derived” quantity is the number of tiles needed — a function of these fundamental quantities.
We have 2 ways to crowdsource this: the “fundamental” way and the “derived” way.
5/
The “fundamental” way:
a) Ask crowd to estimate the room’s surface area.
b) Ask crowd to estimate the tile’s surface area.
c) Divide “a” by “b” for each crowd member.
d) Average “c” across all crowd members to get estimated no. of tiles.
The “fundamental” way:
a) Ask crowd to estimate the room’s surface area.
b) Ask crowd to estimate the tile’s surface area.
c) Divide “a” by “b” for each crowd member.
d) Average “c” across all crowd members to get estimated no. of tiles.
6/
Then there’s the “derived” way:
Just ask the crowd to estimate the number of tiles needed. Don’t break it down into simpler estimation tasks.
Then there’s the “derived” way:
Just ask the crowd to estimate the number of tiles needed. Don’t break it down into simpler estimation tasks.
7/
The question is: which will produce the better estimate, the fundamental way or the derived way?
This matters because market prices are complex functions of many variables. Crowds may be good at estimating individual variables, but maybe not at estimating derived functions.
The question is: which will produce the better estimate, the fundamental way or the derived way?
This matters because market prices are complex functions of many variables. Crowds may be good at estimating individual variables, but maybe not at estimating derived functions.
8/
@borrowed_ideas, @mjmauboussin: have there been any studies on this type of crowdsourced estimates?
I’m not aware of any. I’ve read the article by @mjmauboussin, but haven’t had a chance to read Scott Page’s book yet.
/End
@borrowed_ideas, @mjmauboussin: have there been any studies on this type of crowdsourced estimates?
I’m not aware of any. I’ve read the article by @mjmauboussin, but haven’t had a chance to read Scott Page’s book yet.
/End
