PG sounds like it's okay to invest time,
effort, code, money, etc. in a project
without having any idea why it might work,
and his justification is Microsoft,
Facebook, Google, etc.
Okay:
I grew up not far from a nice public golf
course. It had some par 3 holes. Sure,
on such a hole, it's possible to make a
hole in one.
Okay, for a year track all the hole in one
successes and see how many were by expert
golfers and how many were by people lucky
to have a round at 10 over par.
So, conclude from this that for a hole in
one it doesn't help to be good at golf?
Nope! Instead, on that course there were
so many more poor players than expert ones
that, likely, net, just from luck, there
were more successful hole in one shots
from the poor players than from the expert
ones.
Or, case #9,284,387,437 of how to lie with
statistics!
But, let's consider some other examples:
Ike wanted some pictures of the USSR
and sent some U-2 airplanes. Too soon one
got shot down, and then someone called
Kelly Johnson at Lockheed, and he noticed
a new engine developed by Pratt and
Whitney in their Florida operation: The
engine was a turbojet until about Mach 2.5
at which time it became a ram jet and,
thus, could go on to Mach 3+ without
overheating the engine.
So, get some titanium, design some
aerodynamics especially good for
supersonic flight, and get an airplane
that could fly at Mach 3+, at 80,000+
feet, for 2000+ miles without refueling.
Then the US got a lot of good pictures,
and the Lockheed plane, the SR-71, never
got shot down.
There are many more examples from GPS, the
F-117, the first Xerox copier, etc.
These examples are cases of hole in one
with high payoff and low risk from just
routine work after the initial planning
and review of the plans.
Commercial version?
Okay: Xerox, the IBM System 360, the
Intel 386, Cisco IP routers, the Bell Labs
work in optical fibers and Ga-Al-As
heterojunction lasers, the Sony work on
Blu-ray optical disks, etc.
Lesson: It really is possible to pick a
good target, plan carefully, execute with
low risk, and be successful. For just one
team, poor work and luck are not better.
Okay:
I grew up not far from a nice public golf course. It had some par 3 holes. Sure, on such a hole, it's possible to make a hole in one.
Okay, for a year track all the hole in one successes and see how many were by expert golfers and how many were by people lucky to have a round at 10 over par.
So, conclude from this that for a hole in one it doesn't help to be good at golf? Nope! Instead, on that course there were so many more poor players than expert ones that, likely, net, just from luck, there were more successful hole in one shots from the poor players than from the expert ones.
Or, case #9,284,387,437 of how to lie with statistics!
But, let's consider some other examples:
Ike wanted some pictures of the USSR and sent some U-2 airplanes. Too soon one got shot down, and then someone called Kelly Johnson at Lockheed, and he noticed a new engine developed by Pratt and Whitney in their Florida operation: The engine was a turbojet until about Mach 2.5 at which time it became a ram jet and, thus, could go on to Mach 3+ without overheating the engine.
So, get some titanium, design some aerodynamics especially good for supersonic flight, and get an airplane that could fly at Mach 3+, at 80,000+ feet, for 2000+ miles without refueling. Then the US got a lot of good pictures, and the Lockheed plane, the SR-71, never got shot down.
There are many more examples from GPS, the F-117, the first Xerox copier, etc.
These examples are cases of hole in one with high payoff and low risk from just routine work after the initial planning and review of the plans.
Commercial version?
Okay: Xerox, the IBM System 360, the Intel 386, Cisco IP routers, the Bell Labs work in optical fibers and Ga-Al-As heterojunction lasers, the Sony work on Blu-ray optical disks, etc.
Lesson: It really is possible to pick a good target, plan carefully, execute with low risk, and be successful. For just one team, poor work and luck are not better.