Wisdom of crowds - the update
steve at openmoko.com
Tue Jun 3 19:34:47 CEST 2008
well I am a big fan of the wisdom of crowds.
I am a big fan of the army of davids
That said, I also understand the limitations of these methods.
My method is brutally numeric. this is my training. You begin with a demand
forecast. Lets say "X per week."
actually, it's more sophisticated than that since products go through an
initial SURGE and then they settle into
a regular run rate. For example. at launch you might see 10X units per week
for 1-3 weeks, and then it would
taper off to 5X, 3X, 1X. Engineers here will recognize this response
function. response to a impulse.
basically I track the data on weekly basis and use that to guess the entire
demand curve. more data equals better estimate.
When a product hits the run rate ( linear) part of the demand curve, then a
monkey can do the math. hey boss I sold 1 million
last month. I bet I'll sell a million next month!" thanks monkey. here's a
The big risk is the guessing the intitial impulse. make sense?
The other thing that everybody misses is the markets they don't know about.
I have a forecast for Linux Mobile developers. Call that number X.
Then I get a call from University XYZ. they want Neo.
Then I get a call from medical company PDQ. they want Neos.
and the list goes on. adjust demand curves.
So the crowd guessed 4000 GTA01? for linux developers only?
pretty damn close. But supply constraints biased the result.
From: community-bounces at lists.openmoko.org
[mailto:community-bounces at lists.openmoko.org] On Behalf Of JW
Sent: Tuesday, May 27, 2008 6:24 PM
To: community at lists.openmoko.org
Subject: Wisdom of crowds - the update
hi steve & community
once upon a time a long, long time ago tim kersten set up a site to track
"the wisdom of crowds"  with regard to the question "how many freerunners
will FIC sell in the first two months"
the answer can be seen here (click the stats tab)
http://openmoko.hobby-site.com/ and is remarkably consistent at a median
value of 4000 after over 1000 people voted. In fact this number has
stayed very close to 4000 since the low hundreds of votes.
so are you going to ask the webshop/pulster/trisoft/whoever to keep track of
the real number for you? .....after all, if it turns out to be accurate you
can use this method for planning your next product!
and are you going to tell us how accurate it was :-) ?
actually the rules for the "wisdom of a crowd" to be accurate demand that
each guess is independent..... and here we have a little pollution in that
it is possible for you to see others votes before you vote yourself
(although many will not have bothered)
anyway.... it gave us something to do while you folks were busy making
i reckon it saved at least 12 useless messages to list by frustrated
geeks...... (and oops created only 9 more....)
actually next time we could always use it to guess the release date for son
of freerunner ....
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the community