The goal of K-12 CS education

Alan Kay Alan.Kay at squeakland.org
Mon Jul 30 17:05:58 UTC 2001


For a minute let's not worry about satisfying official school 
systems, and worry just about teaching "real math" and "real 
science". (I use these terms because in the US "math" and "science" 
have been "colonized" to actually mean "fake math" and "fake 
science". And let's suppose we are interested in having the children 
actually learn a strong combination of the two.

There are some very interesting implications that arise from these 
goals. One is that we should try to help children develop "uncommon 
sense" (instead of the 80,000 year old common sense that most 
cultures bestow). We especially would like the uncommon sense to 
include both the epistemological stance of science towards knowledge 
and knowing, and also a large toolbox of modern heuristic methods for 
thinking, having ideas, and colliding theories.

If we could come up with a "new math" that included the (a) new math 
of complex internally interacting systems, and (b) a new math for 
differential vector geometry and quasi-linear relationships, whose 
foundations can be taught to children as play, then we might be able 
to make some progress.

Simply put: objects in relationship is a nice way to do (a), and 
turtle geometry with costumes is a nice way to do (b).

Where the play gets applied and transferred is in the gradual 
escalation of the systems that can be modeled. Just the simple car 
with "car forward 5, car turn 5" is the differential vector geometry 
expression of a circle (as can easily be seen if the pen of the car 
is dropped). Each constant differential vector is added (the loop 
acts the part of the integral sign) and the result is an expression 
for constant curvature (a circle) whose derivative, the change in 
curvature, is 0.

Now the few children who even get to calculus do not usually learn 
this in the US because the pathway is Cartesian coordinate, numeric 
and algebraic. What Papert proposed to do years ago was to simply 
start the kids off on the precursors for modern mathematical thinking 
in science.

When the child hooks the steering wheel into the car (via dragging 
the "steer's heading" tile into the "car turn by 5" tile) they are 
painlessly learning about variables and why variables are a great 
idea.

When the child makes the car into a robot car that can follow a track 
by itself, they are learning the powerful idea of feedback that is so 
central to animal and mechanical controls systems that don't have 
complete knowledge. The next stage of this is to make a gradient 
follower and now they can model what salmon, clownfish, and ants do 
in critical phases of their existence.

The combination of science (trying to understand cause and effect 
"out there") and math (trying to model useful and interesting 
relationships) is extremely powerful.

So the possibilities for transfer are both large and successful if 
the further curriculum is also in accord with modern math/science. 
(We have been doing this for 30 years now and have seen much 
convincing evidence.) The difficulty with getting this stuff going in 
the US has primarily been the lack of any kind of math preparation on 
the part of K-8 teachers. (Also, for political reasons the math and 
science groups do NOT cooperate in the US very much.) This is a huge 
problem.

In any case, those who are interested in this stuff will find quite a 
bit in the current Squeak to this end.

Cheers,

Alan

At 4:08 PM +0200 7/30/01, G.J.Tielemans at dinkel.utwente.nl wrote:
>-----Original Message-----
>From: Alan Kay [mailto:Alan.Kay at squeakland.org]
>Sent: maandag 30 juli 2001 16:12
>To: squeak-dev at lists.squeakfoundation.org
>Subject: Re: The goal of K-12 CS education
>
>
>Do you folks realize that every Player (the automatic wrapping of a
>morph for kids) in Squeak is a turtle?  Get the handles by cmd-click,
>then click on the blue eye to see the viewer, then look in the
>category called "pen use". Change pen down to true.
>
>Cheers,
>
>Alan
>
>>  OK Alan, the problem of transfer: Suppose I learn in Squeakland to build a
>car and automatic run a track with the color-under-command. Where can I
>apply this knowledge outsite track-racing or other kinds of mazes (like the
>moving robot.) What can help me to come to better transfer from this
>knowledge-molecule?


-- 




More information about the Squeak-dev mailing list