Scientific Thinking LO21982

Richard Karash (
Thu, 24 Jun 1999 16:33:21 -0400

Replying to LO21955 --

Although this thread is captioned "Scientific Thinking," it interests me
for what it says about thinking and knowing more generally, about how we
know (or learn) anything.

(In this writing, I'll be talking only about understanding the world. For
me, knowlege = capacity for effective action, and understanding the world
is part of knowledge, but not the complete picture. At another time, I
want to explore what else is necessary to have the capacity for effective

I think there is a stage missing in the process proposed so far...

At 5:40 PM -0400 6/21/99, Kent Quisel wrote:

>At has proposed that scientific thinking be described with 3 stages devised
>by Karl Popper:
> -observation
> -speculation
> -falsification
>While learning communities are in general not the same as scientific
>networks, a good description for one group may be very important to the
>other. So I am looking for words that can be applied broadly. To me, one
>of the most robust descriptions of learning is "diverging" and "converging".
>If we can help organizations do these two tasks well, a lot of learning
>potential can be realized.
>The stage of "speculating" matches "diverging". At pointed out the
>resulting static equilibrium if this stage is weak. Too much "diverging"
>results in chaos.
>The word "falsify" does not seem to have a broad meaning outside of science.
>In the more fuzzy arenas of organizational issues, we rarely have general
>statements that are rigidly true or false. We are mostly looking for
>insights that are only a partial viewpoint on a complex reality.


>I would suggest there is a useful fourth stage of "promoting". This would
>mean an effort and a plan to overcome resistance to new ideas by
>convincingly spreading the learnings for communal benefit. This would
>include appropriate implementation plans.
>I have struggled most with the stage called "falsifying." At had seen this
>as a complementary dual of "true" or "false" and he was also looking for
>another dual, triad, or n-ad. I suggest a respectful complimentory triad:
>"reconciled", "fragmentary" and "illuminating transient (no longer needed)".
>Several examples of respectful scientific progress come to mind. Albert
>Einstein was uncomfortable with Quantum Theory (which is still a respected
>view), but he called QT "incomplete", rather than false.

My background is science (physics) and most of my career has been as a
high-tech entrepreneur. In my current organizational learning work, I have
a deep belief that the essence of the scientific method, if properly
applied, can be used by anyone as an important source of new, practical
knowledge. That's the topic for this note.

I'm writing based on my practical experience, not formal study. This topic
is properly in the field of philosophy, where I'm only a casual reader.

I have been thinking about At's review of the classical scientific
stages... And, thank you Kent for your thoughtful remarks. I, too, had
trouble with the falsification stage.

I think there is a stage missing.

First, these process are are processes FOR WHAT? They are processes by
which we come to believe something. That is, by which we come to see
something as true, valid, and useful.

The falsification step distinguishes the scientific method from other
processes for fixing belief. In good science, there is a rigorous effort to
eliminate false theories, leaving the good theories to rise to the top.
Non-scientific process also have winning theories rising, but When there is
no systematic effort to eliminate the bad theories.

There is a common misunderstanding that "science proves (some truth)." On
the contrary, in the falsification stage, "science shows (some theory) to
be false," leaving us to have more confidence in those theories which
cannot be shown false. In the natural sciences we cannot prove something to
be true.

In statistics, to test the theory that there is a significant relationship
between two variables, x and y, we formulate a null hypothesis (usually
that x and y have certain distributions and are unrelated) and then show
that we can reject that null hypothesis. It's a round-about process, and it
usually not presented this way; a good result provides more confidence in
the theory we favor, but doesn't prove anything.

All this is in complete contrast to the much more limited world of
mathematics in which theorems can be proven.

So far we have:

- Science: Observation -> Speculation -> Falsification

- Non-science: Observation -> Convergence on winning theories

What's missing in the "Science" process is the stage in which the strongest
theories rise from among other theories which haven't been rejected. I want
to combine the two into one process for understanding the world.

And, we need to say something about how the convergence process works.

To do so, let's take an example from a recent trip of mine:

- I can't make my Macintosh Powerbook work with the nifty high-speed
internet facility at the Garden Court Hotel in Palo Alto.
- I've done this before and know what I'm doing. Because I'm on Mac, I
have to adapt from PC-centric instructions. There are two ways to configure
for the hotel system. I try both. Neither works. It worked last time, but
there have been some changes. (all these are Observation.)
- I ask the hotel staff person, "How do we know the service is working in
the hotel at all?" (Speculation... maybe the hotel system has crashed.) She
responds, "I'm using it right now." (Rejection... But, notice how much this
- OK, then it's either the wall plug, the cable, or my computer. Since
there are two ways to configure and neither works, it's probably not my
computer. (Tentative convergence... and hopeful thinking.)
- I ask, "Is there another wall plug and another cable I can try?" (Design
of an experiment.) She replies, "Yes, you can plug-in your computer here in
the office." My computer and cable work just fine there. So, the plug is
the room must be bad and my cable is OK. (Rejection of the "bad cable"
theory and the "bad computer" theory. Convergence on the "bad plug"
- Solution: move to another room. Works fine there. (Further convergence
on the "bad plug" theory.)

Notice I said, "the plug in the first room MUST be bad." Of course, this is
exaggerating my certainty; I haven't PROVED that the plug is bad. Because I
can think of lots of ways in which the wall plug could be bad (not
connected on the patch panel, loose wire, never tested, bad port on the
hub, etc.), I don't invest in testing more alternative theories (e.g. that
the TV interferes in the first room, or that an evil spirit is there, or
whatever). Also, it seems to me that if the plug went back, it might not be
noticed for a long term (since only a minority of hotel guests ever try the
internet connection). Yeah, it's probably a bad plug!

And, notice that the "bad plug" theory WORKS, even though we can't PROVE
it. It leads to a solution for me (working connection in another room). In
addition, it indicates next steps for the hotel staff person (check that
plug!), so it works for her.

(See Charles Sanders Peirce's famous essay, "The Fixation of Belief," which
argues that belief is determined by what works for us. "...there is such a
thing as truth, which is distinguished from falsehood simply by this, that
if acted on it should, on full consideration, carry us to the point we aim
at and not astray...". I find
Peirce fascinating, though difficult.)

There's another element which is not illustrated in my example. It is the
principle of Ockham's Razor: we should favor simpler explanations if they
work; extra complexity in the explanation should be worth it's weight.

I propose that the predominant features of stage 4 (Convergence) are: The
theories that rise to the top are those which are simpler, appear more
likely, and appear to work.

So, my general process for understanding the world is:

1. Observation (including experiments and new measurements)
2. Speculation (divergent efforts to create good explanatory theories)
3. Falsification (rigorous attempts to reject every result of stage 2;
when done right it involves the design of experiments which are powerful
rejectors of candidate theories, and it includes rigorous tests of the
logic; all theories are suspect!)
4. Convergence (in which the theories that rise to the top are those which
are simpler, appear more likely, and appear to work.)

We converge, but can never prove our explanations. There's an important
step in science when a theory becomes generally accepted, and not just one
of many theories. This is the Convergence step in large form. Kuhn
("Structure of Scientific Revolutions") says a major problem at this stage
is difficulty in letting go of the old paradigm in favor of the new.

This process seems very general, useful, and effective to me. To me, it is
the Systems Thinking component, and when coupled with other disciplines
(such as the other four Senge describes), we have the stuff required to be
a learning organization. But, that's for another writing.

In the casual version of this process, the Speculation part is unconscious
and the Falsification part is skipped altogether. We could label this as
jumping to conclusions.

- Observation -> Convergence

For example: "My boss is ignoring me... Was really short with me in today's
meeting. She must think I'm incompetent. Oh, woe is me!"

In an even more casual version, we short-cut the observation part as well.
It sounds odd, but I believe a lot of beliefs are held becasue they seem to
work without any serious observation. Simple exercises convince me that we
could all be better at the Observation part.

- Convergence. Period.

For example: the long held theory, "In autos, bigger means safer." This
theory sticks around in spite of findings of modern auto testing that
airbags and other features are more significant than mass.

Let's examine other processes for knowing...

There's the misunderstood picture of science as a field of "discovery" in
which the data indicate the answer and provide proof.

- Observation and measurement -> Proof of fact

For example: "Scientists have discovered that the universe originated in a
Big Bang umpety-ump billion years ago."

A more accurate statement would be: "Scientists now generally accept the
theory of the Big Bang as explaining a number of features of the universe
that are very difficult to explain otherwise."

This misunderstanding leaves many groups, in my opinion, schizophrenic
about the role of data and proof in knowing anything. For example, "You
believe that xx is causing the quality problem. Where's the data? Can you
prove it?"

Research by Synectics in group problem solving shows that most groups
tend to gang-tackle the first reasonable-sounding solution; therefore most
groups would be better off investing more time in the divergent portion of
the process. The faulty group process looks like:

- Limited Observation -> Minimal generation of alternatives -> Convergence
on the first reasonable-sounding approach -> Actions, Names, Dates,

Such a faulty group process would be greatly improved by attention to each
of the four stages before moving into action planning.

This thread began a while ago with the mention of a "crap detector"... I
just connect my thought to the idea of "detectors" of various sorts. To
support my four stages, I would like to see:

- A device to call us to awareness. I understand there is a eastern
tradition of ringing a bell periodically. The bell calls us to pause and
come to a heightened state of awareness.
- A stimulus to divergence. Edward de Bono's creativity devices are examples.
- A "crap detector"?... Well, what I really want is a device to enforce
rigorous examinations of every output of stage 2. When I managed a unit of
young bright consultants, my role was often that "Rigor Police"... I would
constantly ask, "I know this looks good, but how could it be wrong?"
- Some kind of a signal to say, "Out with the old, in with the new" and to
do so at the right time.

I hope this is helpful. Comments please. And, I would appreciate knowing
how to ground my thinking on this more rigorously.

-=- Rick


Richard Karash ("Rick") | <> Speaker, Facilitator, Trainer | "Towards learning organizations" | Host for Learning-Org Discussion (617)227-0106, fax (617)523-3839 | <>

Learning-org -- Hosted by Rick Karash <> Public Dialog on Learning Organizations -- <>