22 Apr
2012
22 Apr
'12
3:57 p.m.
On Apr 22, 2012, at 12:06 PM, Tony Duell wrote:
No, all I can do is look at the data and do analysis on it. Look, when the company
I work for is producing 20-30,000,000 "widgets" per quarter, the interval on the
line
is 3-6 per second (not in line with things like aluminum cans which are 1-3,000 per
minute) but quite high given the complexity of the devices we're talking about you
can't look at each device in detail.
Statistics (and good design practices) are the only way that high volume
manufacturing works. You look at trends in the test and reliability data. Trust me
when I tell you the attention gets pretty intense when failure rates go up by as little
as 0.01% and line shutdowns occur at 0.5% (then you don't want to be anywhere
around).
And of course then there's the warranty data. Which again is all statistically
derived
based upon life tests during the development (and spot checks during production).
Again we're talking ridiculously large sums of money at stake.
Because it's anecdotal. BTW here's the definition:
(of an account) not necessarily true or reliable, because based on
personal accounts rather than facts or research
Again, when you're dealing with large populations, looking at a few in isolation,
is I'm sorry to say just anecdotal.
Oh come now. And you expect us to believe comments based on data that's
not even available? Sorry, I'll stick with what my experience as taught
me, modifying it on the basis of more experience.
Fine. There's lots of things that I've started to say on this list to deal with
an
argument and couldn't because it would violate various NDA's that I have to
live under (I kinda like my job). Arguments on this (and other fora) while
fun aren't worth sacrificing my livelihood.
I'm sorry if you can't/won't understand how large companies work and the
restrictions that are put on employees and the products (though some have
alluded to here). Because if you did, I'm guessing that you wouldn't have
made the comments you did.
TTFN - Guy
On Apr 21, 2012, at 7:08 PM, Mouse wrote:
Except that your reality is a *small* sample set. When you look at data
for 1,000,000's of components you'll see different results. Until you look at
And you have _ersonally_ looked at millions of components? Sorry, I've
seen too much bogus data already?
Extrapolating long tail effects from single samples does not yield
accurate results.
Neither does assuming that theory matches reality. truly large data sets and do the appropriate
statistical analysis all you're
really talking about is anecdotal evidence.
How coome 2 of use here (at least) seem to have contradictory findings? It
contradicts tony's experience. It contradicts mine as well. Do you
have experience pointing the other way? (I would tend to assume not,
since if you had I would expect you to have mentioned it by now, but
assumptions don't make for good data either.)
Unfortunately I can't really point to it since it's proprietary to the companies