> Ah well... Perhaps this is becoming a religious argument? If so, I
> apologize. I see this clearly, but find it hard to convey. Perhaps I am
> wrong, or perhaps it is just not something that can be known.
Funny you should write that. The other day, I read another comment from you
with your list admin hat on that referred to this distinction between the
knowable and the unknowable. And that got me thinking... is there really
anything knowable?
If there is, how do I know that I know? To take one off the list up front:
IMO Popper science doesn't count as "known" or "knowable", because by its
own definition it's not known; it's assumed until something better comes
around. Seems to me that thinking something is known is largely the realm
of religion (from the inside), but from the outside that's usually regarded
as "believing" rather than "knowing". So while the person may think she
knows, the others may think she believes... How does one /know/ one knows?
If there is not, what does this distinction exactly mean for you?
To throw in another thought: IMO a very important part of good engineering
is to make sure the unknowable is reasonably contained, "reasonably"
meaning finding a balance between the cost/effort and the (of course
unknown) probability of the unknown :) And there's lots of this unknowable
stuff: imprecision or outright errors in data sheets and component
characteristic fluctuation, the client's mind and how it expresses itself
in requirements, the actual real-life conditions when using a product,
compiler implementations and operating system interactions, my state of
mind at 03:00, ...
Maybe the unknowable does have its place in engineering?
On 11/27/05, Gerhard Fiedler <spam_OUTlistsTakeThisOuTconnectionbrazil.com> wrote:
> To throw in another thought: IMO a very important part of good engineering
> is to make sure the unknowable is reasonably contained, "reasonably"
> meaning finding a balance between the cost/effort and the (of course
> unknown) probability of the unknown :) And there's lots of this unknowable
> stuff: imprecision or outright errors in data sheets and component
> characteristic fluctuation, the client's mind and how it expresses itself
> in requirements, the actual real-life conditions when using a product,
> compiler implementations and operating system interactions, my state of
> mind at 03:00, ...
>
> Maybe the unknowable does have its place in engineering?
>
> Gerhard
>
The unknowable and the knowable are relative. A 1K Ohm with +/-5% tolerance
resistance is of unknown value and unpredicatble exact value, but it is of known
boundary value. It is knowable in one sense yet unknowable in another. ;-)
Mathematics is suppose to be the subject who needs precision. However
a mathematics theory needs only to be coherent (or self-contained)
to make it a valid theory. One have to agree on some atoms/base
theorem and then based the whole theory on those. Again this base
theorem can not be proved or disproved.
Engineering is even more so. All theories are accurate (or even only good
enough) only under certain conditions. That is good enough for people to
make use of it.
> Maybe the unknowable does have its place in engineering?
Of course. Engineering is dealing with the half-unknown, with risks and
assesments, with balancing requirements and implemenation alternatives,
often both (at least partially) unknown. It is much like real life :)
Wouter van Ooijen
-- -------------------------------------------
Van Ooijen Technische Informatica: http://www.voti.nl
consultancy, development, PICmicro products
docent Hogeschool van Utrecht: http://www.voti.nl/hvu
--> And that got me thinking... is there really
anything knowable? <--
Well, since Descartes, it's been pretty clear that you can know that
you exist. It's difficult to be sure of much beyond that, if you are
talking about knowing with certainty (Cartesian certainty.. knowledge
that is simply beyond doubt).
There's also a more relaxed idea of knowledge that seems to work
pretty well: Knowledge is justified, true belief. That is, if you
believe something to be true, and it actually is true, and you believe
it for the right reasons, then you can be said to know it. In that
sense of knowledge, there's plenty that we know.
> Funny you should write that. The other day, I read another
> comment from you with your list admin hat on that referred to
> this distinction between the knowable and the unknowable. And
> that got me thinking... is there really anything knowable?
A long time ago I figured out that the purpose of life is to keep yourself
out of situations where you start wondering what the purpose of life
actually is. Yes, it's a recursive definition, but it still has value. This
question is related and so I will avoid thinking about it.
> If there is, how do I know that I know? To take one off the
> list up front:
> IMO Popper science doesn't count as "known" or "knowable",
> because by its own definition it's not known; it's assumed
> until something better comes around. Seems to me that
> thinking something is known is largely the realm of religion
> (from the inside), but from the outside that's usually
> regarded as "believing" rather than "knowing". So while the
> person may think she knows, the others may think she
> believes... How does one /know/ one knows?
>
> If there is not, what does this distinction exactly mean for you?
As a list admin, it means that people do or do not argue endlessly without
externally verifiable statements of fact. I wish to avoid statements that
are made as if they were solid facts and not just opinion that are not based
in any accepted and duplicated experimental result. In the entire thread on
top vs bottom posting, there was not one reference to an e.g. university
study where one type was shown to cause less confusion or take less time...
There were some examples where you could repeat that for yourself and see if
it worked for you, but there were also a lot of statements of fact without
backup. E.g. "This is better, no that is better" without any WHY it is
better and without any qualifier that this is opinion.
> To throw in another thought: IMO a very important part of
> good engineering is to make sure the unknowable is reasonably
> contained, "reasonably"
> meaning finding a balance between the cost/effort and the (of course
> unknown) probability of the unknown :) And there's lots of
> this unknowable
> stuff: imprecision or outright errors in data sheets and
> component characteristic fluctuation, the client's mind and
> how it expresses itself in requirements, the actual real-life
> conditions when using a product, compiler implementations and
> operating system interactions, my state of mind at 03:00, ...
>
> Maybe the unknowable does have its place in engineering?
>> If there is not, what does this distinction exactly mean for you?
> E.g. "This is better, no that is better" without any WHY it is better
> and without any qualifier that this is opinion.
IMO this is closer to the core of a good discussion. But even here the
limits are not so clear.
For me, the things I can /know/ are not really the things I need most of
the help. I can look up which status bits get set by a certain command.
That's the easy part. The most valuable help comes in areas that are not so
well-knowable: like the experience that maybe in certain (unknown)
situations one or the other bit may not get set as expected and documented.
That may not be a "knowable" in the sense that the one writing about such
an experience could offer clear, repeatable conditions, it's also not
something where a "why" would or even could be offered, but it may be
enormously helpful to get even the idea that something like this may be
going on.
>> Maybe the unknowable does have its place in engineering?
>
> Certainly. But not on the PICList please. <GRIN>
I'd say the question "what is the standard PNP switching transistor these
days?" is definitely an unknowable, as there is not one clear criterion
that defines /the/ standard PNP switching transistor. Yet the answers
provide helpful insight in what's common these days for certain people, and
correlating that with other information about the type and depth of the
work of these people actually can make that information helpful for you
(and me).
For me it's to a large degree this and similar types of the "unknowable"
that makes this list such a valuable resource. Experience can be seen, in a
way, as accumulated "feeling" about the unknowable (as opposed to e.g.
"knowledge" gained from reading data sheets), and the collective experience
that we all can tap into here is something I wouldn't want to miss.
> In the entire thread on top vs bottom posting, there was not one
> reference to an e.g. university study where one type was shown to cause
> less confusion or take less time... There were some examples where you
> could repeat that for yourself and see if it worked for you, [...]
Even though that was not about something "knowable" on the surface, I think
Xiaofan (and of course everybody else :) has gotten a good insight in what
a few more outspoken people think and feel about the issue. In many cases,
that's as good as it gets, and that's good enough then. The fact that there
haven't been cited any studies doesn't mean much. As you say, there were
given examples and reasons you can try out for yourself. You can read the
list messages differently now and start analyzing "ah, this is this style
and it has that effect" and so on -- and make your own study. Better than
citing some study that was created to justify a pre-conceived result :)
Just a few thoughts, not to make your life as admin more difficult, but to
maybe add a bit to the clarification of what could be seen as worthwhile
and helpful and what not -- and that the "knowable" nature of the issue or
argument may not be all (or not even the most important) there is to it.
Maybe we get to the bottom of it one day :)
>> Maybe the unknowable does have its place in engineering?
>
> The unknowable and the knowable are relative. A 1K Ohm with +/-5%
> tolerance resistance is of unknown value and unpredicatble exact value,
> but it is of known boundary value. It is knowable in one sense yet
> unknowable in another. ;-)
This resistor is not really of known boundary value. The spec only means
that the probability that its value falls within the +-5% band is /much/
higher than that it falls outside of it. There are few manufacturers that
guarantee you that the value won't be outside the range; and by "guarantee"
I mean more than just replacing a resistor that falls outside. Think life
support -- would somebody go to jail for life (or on the chair in some
places in the USA) if the value were outside the range and somebody died
because of that? /That/ would be a guarantee. (This is a mere and grossly
simplified thought experiment, of course -- in reality there's more
involved with that.) And there's no manufacturer that guarantees you that
the value remains inside the band in your application, no matter how simple
that application may be. So if not even the manufacturer guarantees it, the
"know" part must be /quite/ relative...
IMO we get away most of the time, and in some cases often enough to make a
living off it, with assuming that it'll be in the target range, whatever it
may be. And the better the engineering, the more often one gets away with
it. But generally I put a higher mark on "knowing" than just getting away
with it...
> Mathematics is suppose to be the subject who needs precision. However a
> mathematics theory needs only to be coherent (or self-contained) to make
> it a valid theory. One have to agree on some atoms/base theorem and then
> based the whole theory on those. Again this base theorem can not be
> proved or disproved.
Exactly. Mathematics has knowables, by definition, because it has no
relation to real life; it is a well-defined, completely contained space of
pure thought. Anything in real life has no actual knowables -- at least
that's what I suspect (I wouldn't know even if I knew it :)
> Engineering is even more so. All theories are accurate (or even only good
> enough) only under certain conditions. That is good enough for people to
> make use of it.
Thing is the conditions are usually not known. Some are known, but I think
not all, usually. So what do we do with something we know is only valid
under certain conditions, but we don't know whether the conditions we think
we know are all there are? Do we "know" that? :)
connectionbrazil.com> wrote:
