How do you know what you know? (1 Viewer)

[Libre]

Doc Man-
So we agree that whether or not the 99.9% solution is acceptable, depends largely on the type of investigation being conducted.
In a pure research setting, ANY discrepancy between the observed and expected results could yield entire new branches of knowledge and therefore should not be neglected. The time, money, and effort that has been budgeted also have to be taken into consideration - yet, I would venture that a great many treasures have gone undiscovered because the money ran out ant the treasure hunters gave up and went home just before checking that one little area one more time. The Australopithecus named "Lucy" was discovered in just such a way.

Mechanical tolerances aren't so cut and dried as you imply, by the way.
There are various ways to evaluate them. A mere check off that all the individual components are within tolerance is not a guarantee that the entire machine will assemble and function. (This is my field, by the way). To an "outsider", it may seem that the experienced pros are "splitting hairs". The pros don't see the exercise in the same way.

Kraj:
Steve's post was entirely accurate.
Frothingslosh has not shown himself to be particularly sensitive to others or very open to disagreement. On the contrary.

See - I don't get you. You start by saying you want to "smooth things over". Then you write a thoughtful post in reply to my question - then you wag your finger at me because I resent the treatment I had received earlier.

I apologized because it costs nothing and I'd rather stay friendly here (but I do not deserve your finger wagging, Kraj. I don't believe that I - in ANY way, have overreacted in this thread. If you're going to step in to a squabble than at least be fair minded. You say you can understand why I would feel slighted - BUT...

I may have been sarcastic and dismissive - in response to being disrespected and shouted at. And it's not the first time. I answered in kind that's all. If you don't think so I'm fine with that.
But I certainly don't need to be coached on what constitutes open minded discussion and what constitutes belittlement and argumentation.

 

[The_Doc_Man]

Then again, when we are talking tolerances, I look at electronic component tolerances and how things have changed since I started working with computers. A lot of the resistors and capacitors and inductance coils had 5% tolerance and yet things functioned pretty well, if a bit slowly. They relied on laws based on statistical distributions of electrons in various thermal states forming a "current pulse/cloud" traveling down a conductor. They worked on current flows in the milli-ampere range.

These days with printed microchips and ULSI (Ultra-Large Scale Integration), you get little bitty capacitors, inductors, resistors, etc. that work on PICO-ampere currents. But if you remember what one ampere means, these circuits work on 16,000,000 electrons, give or take a couple.

That is small enough that the Law of Large Numbers doesn't fully apply - i.e. when using current-flow laws based on Gaussian distribution concepts, that means that we are working on statistically insignificant numbers - yet my desktop computer works happily at speeds in excess of 3.2 GHz.

I'm not in the right field of study to figure out how this computer still works when clearly it should not. When I first saw a computer many years ago (we'll politely avoid saying HOW many...), to me they were F***ing Magic. Now, seeing the way they work in an apparent flaunting of the laws of statistics, they are again in the FM category. History repeats itself.

 

[The_Doc_Man]

A mere check off that all the individual components are within tolerance is not a guarantee that the entire machine will assemble and function.

True for watches, internal combustion engines, Rube Goldberg devices, ... it is a matter of (a) being in tolerance and (b) being assembled correctly so that when you are finished, you didn't leave out any parts (and, according to my dear old Uncle Maurice), (c) knowing where to kick it to prove that you mean business

 

[Libre]

True for watches, internal combustion engines, Rube Goldberg devices, ... it is a matter of (a) being in tolerance and (b) being assembled correctly so that when you are finished, you didn't leave out any parts (and, according to my dear old Uncle Maurice), (c) knowing where to kick it to prove that you mean business

Another thing is - somebody figured out those tolerances. They could have screwed it up - seen it happen many many times.
Another thing is - you don't always design for worst condition tolerances. There are statistical approaches that allow interference or non-assembly in certain cases.

 

[Kraj]

I answered in kind that's all. If you don't think so I'm fine with that.

The thing is, 'answering in kind' is what everyone thinks they are doing. "He started it" is a child's reason for perpetuating a conflict.

But considering Frothingslosh is the only person on your friends list and you haven't seen fit to use the forum's ignore feature, I suspect your relationship isn't as adversarial as you claim, so I'll just excuse myself from this.