[IAEP] changes in outlook with Sugar (was Re: Comments on David Kokorowski, David Pritchard and "Mastering" Educational SW)

[Alan Kay]

A wonderful question that I feel quite inadequateto give a deep or even reasonable answer.

A "user interface that could mentor learning" has been a goal in computing for the last 50 years or so (perhaps McCarthy's "Advice Taker" proposal in 1958 was the first real articulation of this idea -- Lisp was actually invented by McCarthy to create a programming language in which the Advice Taker could be made).

When I thought about a "Dynabook" as a children's computer in 1968 there were many movements afoot and good funding to support efforts along these lines. I really wanted to have such an interface because I didn't like the weak contexts of schools and parents -- the idea was to make something that was both very powerful and subversive (like the printed book, but dynamic and even more subversive because it could teach and get around the difficulties of adults). 

Because these interfaces weren't there yet -- and because I didn't think I had any special talents or ideas wrt AI technologies -- I decided to design a "tool based theatrical" interface for the "interim Dynabook" at PARC, with the idea of switching over when "those smarter AI folks" came up with the real deal. This didn't happen and we are still using the old tool based theatrical scheme (which is OK as far as it goes, it just doesn't go far enough for what a computer really is as a knowledge, outlook changing, and learning technology).


When Nicholas decided to do OLPC I was both thrilled and depressed. The latter because in every context, but especially in developing countries, it's absolutely critical to have an "interface agency" which can be better than no teacher, better than a bad teacher, and do some subset of what good teachers do. And computing had not yet come up with it, and today, except for a few small exceptions, the topic is not even being funded.

Two of the somewhat polar perspectives on this are (a) start with making mentoring systems for knowledge that is so well understood that computer agents can "understand" enough of the content to not just give good advice and guidance, but gentle versions of these, or (b) to try to help the learner with a larger "master skill" like reading and writing, which allows materials of many kinds (most of which won't be mentored) to be part of the mix.

One of the big problems in even having worthwhile discussions about these issues is that (for example in the US) so many who are interested in education get so excited when a little progress seems to be possible, but lack the outlook to see that for lots of things there are thresholds which have to be reached to make the learning powerful enough to make a real difference. For example, one survey in the US indicated that while most people in the US can read a little, only about 20% of adults can read well enough to deal with the content of being a citizen, to understand the scheme of governance, etc. However, most testing of reading in the US does not test for this at all, but looks to see how students who are below threshold ebb and flow (but never get to fluency). This is a version of "air guitar education".

We can see this with the XO and OLPC also. Much is made about "getting started", but considerable history shows so many ideas that seem to get started don't actually make it even as the supporters retain their enthusiasm. This is because only about 5% of humans everywhere are interested in new ideas and enthusiastic about the possibilities. (OLPC and IAEP are made from these folks.) But about 80% of humans do not make decisions or changes based on the worth of the idea. This very large group instead decides via a kind of social concensus which tips over when almost all of them almost agree. On small issues this often takes decades, and for larger ones often much longer. This is an outlook problem, but it is a deep one because it seems to be not so much cultural as a deeply human trait.

I've been going around talking to possible funders about this issue. It is bigger than any public computer research area that is being funded today, but I think it is not as big an effort as was the "get a man to the moon within this decade" one in the sixties, and is much smaller than the costs of any war. I believe that it is philosophically, scientifically, and technical possible. It's just really difficult.

I've advocated that for (a) above that an absolutely brute force approach be taken for one course (maybe 8th grade science) where the purpose is to solve the UI look and feel and environment and pedagogy, but where there might not be much of an AI at all, but something more like the John Seely Brown and Richard Burton "BUGGY" and "SOPHIE" systems of the sixties (lots of explictly (and expensively) built) relationships. This would be pretty expensive, but very useful in its limited form, and would also be a target for "new AI" researchers to replace the explicit wiring of the coaching logic with much more general and advisible mechanisms (this is absolutle necessary or else the expence of such systems would be astronomical).

The (b) one has always seemed harder to me, but more and more recently it keeps on reappearing with the siren's song of "but for children you really want them to start by reading stories and there are a lot of things you can do to use the child's own intelligence to augment the limited abilities of the AI". So I've started looking for people of talent who have already had this insight and have been working on it while I've been pondering interfaces for (a).

My main observation over the last 45 years of computer science research is that for hard problems it is best to find talent and fund it -- and talent is rarely general. 

So my "vision" here is let's try to find supertalents in this area wherever in the world and try to fund them.

Very best wishes,

Alan


________________________________
From: Tomeu Vizoso <tomeu at sugarlabs.org>
To: Alan Kay <alan.nemo at yahoo.com>
Cc: K. K. Subramaniam <subbukk at gmail.com>; iaep at lists.sugarlabs.org
Sent: Thursday, July 2, 2009 5:37:09 AM
Subject: changes in outlook with Sugar (was Re: [IAEP] Comments on David  Kokorowski, David Pritchard and "Mastering" Educational SW)

On Thu, Jul 2, 2009 at 13:49, Alan Kay<alan.nemo at yahoo.com> wrote:
[snip]
>
> But, if I were trying to make things happen with IAEP, I would try to do
> just a few main things, and one of them would be to make a
> program/user-interface which could do a great job of teaching a child to
> read and write their native language without requiring any more from the
> adults around them than a little encouragement. Part of the desired changes
> in outlook could be made part of the stories and other materials that the
> kid would encounter along the way (and part of the big change in outlook
> that we are a part of is fluent reading of non-story materials in general
> and about outlook changing ideas in particular).

Do you have already any vision about how to make that happen? I have
seen lately several people interested in working on better tools for
reading, may be a very interesting opportunity.

Regards,

Tomeu

> Best wishes,
>
> Alan
> ________________________________
> From: K. K. Subramaniam <subbukk at gmail.com>
> To: Alan Kay <alan.nemo at yahoo.com>
> Cc: iaep at lists.sugarlabs.org
> Sent: Wednesday, July 1, 2009 12:12:59 PM
> Subject: Re: [IAEP] Comments on David Kokorowski, David Pritchard and
> "Mastering" Educational SW
>
> On Wednesday 01 Jul 2009 9:03:26 am Alan Kay wrote:
>>Your last sentence is somewhat parallel to what many business types like to
>>say about how hard it is to measure Return On Investment for research
>>funding. But in the business case, this is actually a form of dissembling,
>>since an enormous percentage of all the GNP (and in fact GWP) comes
>> directly
>>as return from research.
> The top 10% of learners don't need school. The bottom 10% need more than a
> school. For the middle 80%, learning in school should be demonstrably better
> than that out of school. A school is relevant only if it can detect and weed
> out contexts that hinder learning (or have nothing to do with learning) as
> quickly as possible. Otherwise people will vote with their feet.
>
> In one of our local public school parents meet, a poor farmer challenged,
> "Why
> should I retain my kid in this school? Can you show me one student who
> studied
> here and became somebody in life? In the field, I can teach him to raise at
> least one crop a year". For him, there was lot more of physics in the farm
> than in school textbooks.
>
> Subbu
>
>
> _______________________________________________
> IAEP -- It's An Education Project (not a laptop project!)
> IAEP at lists.sugarlabs.org
> http://lists.sugarlabs.org/listinfo/iaep
>



      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.sugarlabs.org/archive/iaep/attachments/20090702/6320be6f/attachment.htm