Please consider a donation to the Higher Intellect project. See or the Donate to Higher Intellect page for more info.

Observations About Children And Computers

From Higher Intellect Vintage Wiki
Revision as of 12:31, 30 September 2020 by Netfreak (talk | contribs) (Created page with "<strong>Observations About Children And Computers<br> </strong> <strong><br> </strong> <strong>Alan Kay<br> </strong> Advanced Technology Group<br> Learning Concepts Group<br>...")
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search

Observations About Children And Computers

Alan Kay
Advanced Technology Group
Learning Concepts Group

May, 1994

ARL Research Note #31

Alan Kay was asked to speak at a gathering in Boca Raton, Florida sponsored by a group called "For a New Social Science". The meeting's focus was to explore practical applications used in schools to further our understanding of human learning and development. The following is an essay written by Alan for that occasion.

Apple Internal Use Only
Research Note #RN-94-31

I got interested in children and computers in 1968 as the result of visiting the early LOGO experiments of Feurzig, Papert, and Solomon in several Boston suburban schools. The idea of teaching powerful ideas, especially those in math, science, and computing, by getting children to "read and write computer" seemed terrific then, and still does today. One of the most striking things about the last 500 years has to be the difference in "thinking leverage" one gets from learning a framework like calculus as opposed to just trying to be smart with native IQ. Understanding this as a metaprinciple gives still more leverage, in part, because it can lead to purposeful creation of new and better frameworks.

I had been doing an early object-oriented personal computer and this visit completely changed my focus to children's computing where it has remained until this day. This led to the idea of a "notebook computer" for children, and a bunch of us started to work on it at Xerox PARC bolstered by the history of printing, particularly the saddlebag-sized book of Aldus, and the promise of Moore's Law: that silicon would improve exponentially with a hefty time constant. Spinoff results from those days were workstations with overlapping window interfaces and Smalltalk, a completely object oriented language (originally done for children, it has long since been corrupted by adults).

Framing Ideas
1. New media take their initial content from the older media they are starting to displace. This is a McLuhan idea and it applies well to computing. This is in spite of pioneers being able to show what the new ways will and should be. Let me just take 3 of them from the sixties as examples. Sutherland (Sketchpad) showed that interactive graphical simulations of complex nonlinear models could be constructed by end users and would facilitate understanding in a way that no paper or classical mathematical model could. Engelbart (NLS) showed that the promise of hyperlinking and networking could indeed qualitatively improve the possibilities of thinking, communicating, and group cooperation. Papert (LOGO) showed that children could be real mathematicians by giving them a forum in which actual mathematical thinking and symbolizing would have the same kind of import and meaning as it does for adults. Thirty years later, there is still no question in my mind that these are three of the most important new and powerful ideas that the computer will bring to humanity. Our mistake was to think that they would be adopted immediately just because they were so obviously good. We forgot the "McLuhan Viscosity" of the larger world which is not driven by ideas at all. The result is that computers are mostly used to automate old paper processes, almost no model building is done outside of 19th century spreadsheeting, OOP after 25 years is less than 5% penetrated in Fortune 500 companies, there is still almost no hyperlinking, networks are just getting started and are mostly text based, and the predominant use of computers in schools is either as a modern Skinner box, or as vocational training in word processing and accounting. This is pretty discouraging.

2. Powerful ideas create needs only they can fill. This is the essence of paradigm shifting and as Kuhn noted takes longer than one would think or hope (he mentions 25 years for a change in science because it takes that long for the older generation to die off!). As with printing, the new computer media is subversive. Now that it is becoming increasingly pervasive and much cheaper and easier to produce content, we will see, as happened with books, that those with an inclination to learn will be able to find plenty of the "new good stuff" amongst the fluff and junk. In short, the changes predicted by the pioneers will indeed come to pass but will be less generally taken up than we would like to believe and in a less timely fashion.

3. It is not clear that formal education has ever succeeded in educating more than a small percentage of those put through its mill. But even with a small percentage of success--and from the standpoint of democracy a distressingly small percentage at that--getting more of the population in contact with the possibilities of learning coupled with the enormous powers of postrenaissance thinking has been sufficient to change the world around. I believe that the problems of education in a democratic society

with communications technologies have always had much more to do with trying to keep the "80%" within hailing distance of the "10%", than in trying to "teach".

4. Each medium has stories it tells well and those it tells badly. For example, television is not good, nor can it be, at dealing with long abstract arguments such as those presented in Common Sense and The Federalist. The computer is more complex in that it can imitate other media and eventually will pull all existing media into itself. But is this imitation done well enough to preserve the values delivered by the displaced media?

For example, current computers are very biased towards low-resolution moving images--in large part because of the needs of our foveal-oriented vision system to saccade in order to see, and that our peripheral vision is very low in acuity but high in brightness sensitivity in order to warn us of movement behind us. The result when trying to read text from a CRT up close (itself invented to be looked at across a room with a small visual angle) is for people to print out anything longer than a few lines. Of course the resolution, crispness and contrast ratio are all poor, but the real culprit is that the 60-70 cycle flicker seen out the corner of the eye causes many gratuitous saccades which seriously interferes with the sequential scanning required by reading. Some people even experience vertigo. Thus putting current day machines willy nilly into schools and especially trying to teach reading on them is somewhere between ludicrous and disastrous, but quite typical of the majority trying to use technology without understanding that each has a very definite epistemology with regard to what its effects are likely to be. Television should be the last mass medium to be invented naively; modern media may need a kind of Surgeon General's warning!

The flicker problem and readability of text in general will eventually be remedied by flat-screen displays such as the new Xerox PARC display which has 6.3 million active pixels (about the resolution of a laser printed page) and a contrast ratio of 150:1 (paper is only 100:1). These are still about five or more years away.

There are many similar content issues with regard to children. Young children do much of their learning kinesthetically and it is almost certain that removing this will remove much of their foundation for assimilating new ideas. Current day computers do not provide force-feedback--sometimes called haptic interaction--(nor does television) and I believe that young children need a lot of it. Over the years haptic devices have been invented and used (at Apple we have developed haptic mice in which one can feel the weight of what one is dragging), but they are inadequate to the task even if they were available.

At a higher content level, the stories that computers best tell are those that the pioneers early on understood (such as simulation, hyperlinking, and powerful ideas via building models) and these are just what scientists tend to use them for. But most schools do not understand these three areas very well and there is little software that makes these powerful ideas available. For example, Interactive Physics has so much physics history embedded in it that one cannot even try out other laws of gravity (such as inverse cube). SimCity will not show students its model, nor can they try city generation rules of their own. This is not educational software!

Mosaic is at long last making the Internet look a little bit like Englebart's vision, but there is no child's version, nor will there be for some time, nor is the Internet at all well set up to be able to do much more
than look at interesting things: there is no OOP interoperability that allows interesting things other than text or photos to be brought to one's own machine and run.

Seymour Papert and a few others can use LOGO to do just what was promised long ago, but there is no Seymour packaged with the LOGO diskette, and most attempts by teachers who did not understand the

pedagogy or epistemology were failures. (They naturally concluded that the LOGO idea was false or a lie, instead of them being inadequate to the task!) Now, it is clear from observing random programmers that programming per se does not come close to elevating their thinking. If anything, it only makes them more narrow and careful, because they can't get their heads beyond engineering problem solving and the quick fix. This is the "playing scales fast" bug from music transferred to computers. Writing doesn't do it automatically either. Witness the amount of writing that lawyers, etc., of all types have to do. As a class they are singularly boring and constrictive.

On the other hand, in the spirit of Zen, many endeavors can be used as a vehicle and catalyst for boosting enlightenment. Art, meaning esthetic yearnings and satisfaction, has to be the center of the pursuit, or just mechanical cranking results. Art was the most important part of what Seymour, Wally, and Cynthia were trying to do, and this is the part that was missed the most in the attempts to transfer LOGO to schools--in large part I believe because most teachers have never experienced any kind of mathematics as art.

5. The music isn't in the piano. The "80%" who aren't predisposed directly for playing an instrument usually decided they aren't "talented" enough and don't pursue it further, even if they love music. Moreover "applying" students to a musical instrument usually has disastrous results. Casio did a survey in the early 80s which showed that 90% of those forced to "take" piano as children were turned away from music for the rest of their life. So at best an instrument is an amplifier, but there has to be a signal to amplify! In this case an inner musical impulse (which is best developed via singing and dancing, both low tech activities).

At worst, and more often, we have the instrument as a prosthetic being applied to a healthy but underdeveloped limb which then starts to atrophy. Trying to get students to "take" computer will turn them away from it as surely as they have been turned away from reading for pleasure fluently, and as they have been turned away from music.

What is most sad here is both that the 80% can learn music (pedagogy is really more for those that are not "prewired" and it is known how to teach almost everyone to be a musician), and that most of them would like to (again according to a survey). Society doesn't think this is so sad because music is not thought of as important, but society does get a little worried about other areas of pedagogy not working out. Quite a bit is known about art learning and sports learning as well. What about core academic subjects? I personally do not see any real differences here in the basic approaches to learning and teaching, but there are real differences in the system.

The really frustrating thing here is that the interactive and reflective nature of the computer should allow some of its "music" to be put in it--that is, it should be able to do some of the things that a piano needs an enlightened human being to reveal, and for which books can do somewhat for themselves about themselves. This has not happened yet, though it has been one of the most anticipated if not promised new values that the computer should bring.

6. Much if not most real learning happens in a "club". This is a Frank Smith observation. He points out that when people get interested in something for their own reasons, they don't often go to school to learn it, but instead find a group of fellow enthusiasts. A club. At a tennis club one could have a match
with a 12 year old (and probably get whipped), learn fine points of strategy from an 80 year old and even take a lesson or two from a pro. School has none of these features. Ages are not mixed, real things are not done, and the teachers are certainly not pros at what they are trying to teach. Most have never done any science or math, few are fluent readers and fewer write, and worse, as a group they are remarkably shy about learning new things for themselves. The exceptions really stand out and are few in number.

The only catch here is "democracy". First, not everyone has the same access to clubs. And, more troubling perhaps is that democracy needs enough commonality of outlook at least to argue nonstupidly about issues. If we think of formal schooling in its noblest light, it is there to provide more equality of opportunity for everyone, and to create a higher probability that most citizens can think about the important issues of society. Neil Postman would say that the real reason for school is to be the antithesis of common culture. In other words, if the culture is radical then school should be conservative; if the culture is stodgy then school should be radical. If the culture is trivial; school should be profound. Discourse is at its best when it is done in a climate of criticism and opposition.

It should be one of our main tasks to make sure that some of the clubs that are developing within the worldwide networks have lots of content without being stuffy. There are no age barriers here, except those raised by poorly designed user interfaces (for example, four year olds who can't yet write are shut out even though there are many pictorial (even programmatic) things of interest that they could retrieve (and learn to read and write in the bargain).

7. Children's learning is predisposed by nature in a variety of different ways. It is not as important to subscribe to a particular multi-mentality theory (Piaget, Bruner, Gardner) as it is to realize that our mind-brain system is modular enough to have a variety of ways to know and learn, that these are not well-integrated, and that children do not have access to all of them in equal measure and under their control. In brief, telling is not the same as showing is not the same as doing. In the other direction, learning by doing does not necessarily produce analogy or useful generalizations.
Another effect which many who directly work with children have started to notice is a vast change in outlook in most children around the ages of 8 or 9. Howard Gardner in his study of children's art around the world noted that regardless of the culture their approach moved from experimentalism to deeply conservative stereotypes at about the age of 8-9. In our seven year project at the Open School we saw a whole cluster of natural mathematicians, aged 6, in two years become less able to think flexibly and creatively about orderliness (even in many absolute terms) at age 8. We saw children, who at age six could take a block of wood (or any object) and pretend it was anything else, move in the same two years to a state in which they were "frozen" by already used objects and insisted that they could not be used for any new purpose. The video of my talk at the Chicago Arts and Sciences shows one group of children arguing over this. It's as though nature has said to the children: learn your culture's reality by age seven and commit to it for the rest of your life.

A few children preserve their 6 year old outlook and these, by in large, are the ones that continue to do well in math and find computer programming easy and obvious. The larger majority on the other hand find both almost incomprehensible. As far as I have been able to find out, though this phenomenon has been noticed by researchers here, in Europe, and even in New Zealand, no one has suggested any strong ideas for helping the majority of children make a softer landing and avoid this early and disastrous "hardening of the categories".

Related to the above is that much learning that is most easily done by children and adults is situated, but the very situatedness of it may make it almost impossible for them to use it in other contexts. Nonsituated learning is difficult for most. On the other hand, I would say that it is precisely the
nonsituatedness of mathematics, for example, that has made it so astoundingly useful in many different areas. And that the difficult "powerful ideas" that have been so hard to invent and retain are both powerful and difficult because they are metanotions that can attach to and be used in many contexts. My tendency therefore has been to say that we should not retreat back to comic books or their computer equivalent just because they are "easy", but instead should redouble our efforts to learn how to better teach the tough, unnatural and powerful ideas that have made the last 500 years so different from the last 500,000. In a constructivist learning environment, learners need challenge and difficulty; "user

friendly" may be the worst thing for them, akin to putting frets on a violin! Of course, gratuitous difficulties (like those in MSDOS, etc.) are to be avoided at all costs. The difficulties chosen should have a big personal payoff for surmounting them.

I have been very influenced by the ideas of Maria Montesorri, Tim Galwey, Shinichi Suzuki, and Mike Csikszentmihalyi, particularly the "flow" model of the latter. I think of modern thought as being multiview buttressed arguments rather than as single viewpoint dogma. Helping children be comfortable with a certain amount of uncertainty is a very important and difficult part of a learning environment to build. Mike's flow model shows "flow" happening when challenge is balanced against ability. When challenge is higher, anxiety is experienced; when skill is higher, boredom is the result.

Galwey especially has shown that the flow area can be widened in several ways. More challenge can be dealt with if safety is increased (this is the origin of "undo" in the overlapping window interface). More ability than the task demands or the perception of sameness (such as the tennis ball appearing the same) both which ordinarily bring boredom, can be dealt with by increasing awareness, for example, by noticing more about the situation, introducing esthetics, and so forth.

These ideas apply in any learning situation in a multitude of ways. Most classrooms are neither safe nor interesting, etc. Personal computers were invented in part to provide a safe private and interesting forum for thinking. I do not believe that a networked system in which institutions have unlimited access to one's one content is "safe", nor will the children regard it as safe. There needs to be a Bill of Rights for children's content in school. Most institutionally prepared material is not "interesting" and is not likely to get much more so when rendered on the computer. A boring author is a boring author in any medium.

Specific Issues Of The Conference
What would it be like for every child to have a computer on their desk in five years? My first response is that the computer won't be on their desk but under their arm! That has been clear for more than 25 years and it is amazing to me that Moore's Law and the conclusions drawn from it especially by the 1960s ARPA researchers are still not understood. The late sixties experiments of ARPA regarding "packet radio" and how it would extend the ARPANET (now the Internet) are coming to pass as I write, and we will certainly see networking extended (especially via wireless) progressively until it becomes pervasive.

The easiest prediction to make about formal education is that it will not change materially in the next 5 years, or next 15! Television will not yet be generally regarded as actually harmful for everyone, especially children. Therefore we can expect very similar effects that we now see with the book. Most teachers and parents, and hence, most children will not achieve fluency in the important discourse that the computer can carry. Most children will be turned away from what was supposed to be "their machine" because of institutional blundering and heavy-handedness. Most teachers will still be searching for "methods" for teaching a subject instead of understanding it and practicing it. Most administrators will try to use networked computers to "control" the educational process. Most classes will still be graded by age rather than project and club oriented. Most parents will continue to use television, and now home computing, as a baby sitter rather than directly interacting with their children. And so on. This kind of prediction is as dreary and accurate as predicting that there will still be hoards of COBOL programmers long after the millennium has passed.

My reaction was not warm to the materials sent to me about "Standard Teacher-Student Network Interactions" and "Benefits Of Aggressive Computerization". It's rather analogous to the seductive notions about social planning and even the desirability of a benevolent dictatorship--it would be so great if people were high-minded and smart enough to pull it off. But the best social plans, such as

those of the American revolutionaries, were rightly suspicious of central planning and control, and not just because they doubted the controllers' good will. They also doubted that organizations themselves could be trusted and restrained.

If someone could just show me any piece of CAI that actually understands both the area of study and the student that is studying it, that also has a great model of learning and is able to flow and ebb with the student! I have seen bits and pieces over the years, and perhaps I have seen some that are better than the worst teachers that I have ever experienced, but, unless I am very mistaken what is needed doesn't yet exist. Instead the easiest thing to do is what is usually done: automate the current processes of assignments, strands, grading, tests, bad pedagogy, ignorant teachers and administrators, provide imitation television as "edutainment", and so on. Yes, I know that "new ideas" will be added, but even with the state of the art of human modeling (as in Lenat's CYC program), it would be close to impossible at the current time to make a system that could really recognize a student's creative act in mathematics: reasonable but unorthodox. Most teachers can't either, but most mathematicians can and do. It is probably much better for the next decade or so to treat the computer as an instrument that extends our reach in many dimensions rather than as a better instruction sequencer. I do not doubt that a system could be built in the next five years that is "better" (using current measures) than many teachers. But this misses the point. Relative judgments don't count in art nor do they in learning. There are thresholds in both. We have to get the kids over the thresholds for them to be successful and I just don't believe that anything along the traditional models of schooling, learning, tracking, stranding, etc., is going to do it.

By the way, I don't have any philosophical problem with the idea of eventually being able to do schooling outstandingly well via computer, but most people who have worked in this and related areas over the years have just not been mentally tough enough about what constitutes real success to pull it off. The models just aren't really good enough to do it well enough yet.

At PARC I did not think we were smart enough to do a really good intelligent interface (and we weren't!). Instead I opted for a "theatrical approach". Coleridge said that people go to bad theatre hoping to forget and to good theatre "tingling to remember". In the trade, theatre is called "the magic mirror" because when you are doing it well you are reflecting the audience's own intelligence back out at them. You can't "tell" people about the deepest crannies of the human heart in a few hours in an artificial room with cardboard scenery, but you can evoke their humanity to fill in the thin spots. Theatre, like all of our great symbolic systems is a vast simplification of what it is mapping, but the maps are cunning enough to reinvolve and reinvoke. I thought that a theatrical approach to the complexities of the computer might work as well if couched in metaphors that were already part of people's experience. Now the trick in this is to make sure that "good theatre" is what results. I think we achieved that because we had a lot of powerful and useful content in our story about simulations via object-oriented languages and the kids sensed it and ran with it. Today the commercial version of this interface is similar--the stage is the same--but the show is mostly weak imitations of paper. Soon it may be weak imitations of schooling or television. I think that what is most needed at your conference is a reexamination of the "show' you are trying to put on with the tough question"is this good theatre or bad theatre?" emblazoned constantly in the air. Otherwise the temptation with technology is always to do something just because it can be done--I call this "inverse vandalism".

Alternatives? I do think that there will be more alternatives to state schooling, and that home and club schooling will be more popular, especially augmented by networking and "multimedia" (after more suitable displays, etc., appear). The use of collaborative simulation environments (known as MUDs or MOOs) is particularly popular with children and will continue to grow. A number of schools are experimenting with these in a variety of contexts. The logistics of home schooling will not be possible for many who need it most, and democracy will not be furthered thereby.

One of the things I have been interested in for many years (and Seymour as well) is the possibility
of giving young kids direct access to interesting ideas via an interface that does not require the mediation of adults at all. I have never forgotten the story of my college roommate, one of the smartest kids in our school and now a successful doctor. He was the son of illiterate immigrant parents in Hoboken New Jersey and the first book ever brought into his house was brought in by him at age 5. He had grokked what they were on first sight and that was all he needed. The public library was his information highway and he pretty much ignored both his parents and the local schools as he educated himself. Now, I kind of think that "getting educated" really means educating yourself, both from the standpoint of motivation and as a general believer in constructivism. There is no question that sympatethic and exciting adults are what children need most, but their highly developed curiosity around age four, coupled with the "universal access" that the current administration is talking about, with a really good interface and enough new ideas to really take advantage of the active and interactive nature of the computer, might be able to help bridge the gap that my roommate was able to leap all by himself.

In my own self education as a child I discovered that the WorldBook Encyclopedia was too shallow to learn anything from, and the Britannica was too hard, but that getting an overview from the WorldBook first helped greatly with puzzling out the "real stuff" in the Britannica. This multiview approach coupled with reasonable displays and a decent model of users and how they grow and regress will help many to learn much by themselves.

I also discovered for myself what I think is the most useful heuristic in self learning (which is also mentioned by Richard Feynman and Leon Lederman in their respective books). As Lederman put it, when you read something you don't understand, don't worry and read it anyway! The next time you look at the book, it will be hauntingly familiar. Feynmann would take anything he thought was important and keep on reading it, perhaps over several years, until he understood it. I think this approach is rather common amongst self learners tackling serious stuff; once you realize that it doesn't matter if you understand it right away (this is 100% opposed to the school way which wants you to understand trivial things at a particular time), but you can understand almost anything eventually via pondering, more reading, asking questions, etc.,--in short all of the things that very young children and scientists do naturally, but are off limits in most schools.

Related to this is one of my main objections to preset curricula--and that is the "coverage bug". Even the AAAS in its Project 2061 and "Benchmarks for Science Literacy" miss the mark here. It is like trying to get a prospective musician to play all of the instruments! The most important thing in modern learning is to have the experience of going deep on a topic of interest to the "good stuff". Most deep routes will lead there, and once there, many possibilities open out. Music is like this too. I will never forget one of my first choir experiences as a very beginning musician learning to sing the "Crucifixus" from Bach's b Minor Mass with a very exacting conductor. He worked us giddy teenagers to the good stuff and I've never been the same since. Once one has the experience of the enormity of human response to great art, it completely shifts one's entire world and the sense of purpose within it.

For children, it shouldn't much matter what the subject matter is as long as it is made real and there are resources and motivations to go deep enough to the basic questions of knowledge and human thinking. I think the same applies for teachers as well. I always try to find out what they really like to do and have gotten good at. It usually has nothing to do with their official job.

And they have not learned it in school! But whether it is gardening, cooking, painting, model trains, fixing cars, etc., this is where they need to start: with topics that are close to their heart and that they still feel romance about. (Because the children respond in the biggest way to adults that are romantic

and love something.) Then I help them take their interests deeper and deeper and guess what? They start to find connections with other parts of modern thought, especially science and mathematics.

Again we can see that computers and networks will eventually be incredible vehicles to the wonderful world of the deep, but what an incredible amount of work will have to be done before it will be possible. Libraries are by far the best bet today, and will be for some time in the future. Until then, we can think of computers and networks as bringing commercials for the good stuff (just as do some of the PBS NOVAs, etc.). As long as we don't fool ourselves into believing that the commercials are the content (as they are on network TV) we can work towards getting the good stuff online and in a usable and flexible form.

Finally, Neil Postman has pointed out that there are vast forces contending for the minds of children and much of it is being done via media. One of his "school as a thermostat" suggestions is that we should seriously consider teaching children how to recognize propaganda whether for athletic shoes or for ideology--that is, in his words to help them to become "media guerrillas"--"epistemological commandos"?

Why learn programming in the 21st century? Why learn math if you are not likely to become a mathematician or scientist? As with reading and writing, if "information" is what you are after, formal systems seem to be overkill as opposed to the village agora. We see this happening in spades on the Internet. Formal systems are hard and make people's heads hurt. They will abandon them whenever they get the chance.

I have no interest in hurting people's heads. What interests me is what they become if they get fluent enough so it it doesn't hurt. They are now very different! This is what Frank Smith means when he says that reading and writing aren't/shouldn't be primarily about transfer of information, but about the kind of thinker you become by learning them fluently, particularly the constructive process: writing. For him, this happens when you read and write about ideas, that is literature, as opposed to junk communication.

Programming in this sense should also involve reading and writing about Ideas. They need to be powerful, compelling and illuminated by a particular kind of formal treatment. They could involve things that people have always wondered about, but not really understood. They could be about organizations that are not even expressible in common sense terms. All in all, the beauty of formal treatment is a kind of elegant pragmatism, not so much in solving a real-world problem, though this happens often enough, but a pragmatism in representing and expressing in so compact a fashion as to not only illuminate the idea, but also to leave some brain cells free for more thinking.

All the creativity literature I know of stresses that the creative act is intuitive and nonverbal: mostly kinesthetic and iconic. We could think of that as "right brain" if we wanted. But people have been doing this kind of creativity for hundreds of thousands of years. What is more interesting is the last 500 years in which creativity in art, music, science, and even politics (in the American Constitution, for example), have been required to work in formal systems, to be BOTH creative AND arguable.

I think that mathematics and programming are primarily about "Seeing and Constructing in terms of new formal knowledge about how things are put together". The understanding and teaching of this is just what we need to improve greatly. The user interface and general approach should not primarily be about problem solving but about "situation illuminating".

Seymour has pointed out that most people learn by bricolage, by bits and pieces, here and there. At some point there has to be motivation and will to drill deep to principles or the bricolage remains at the common sense level as superstition, heuristics, and proverbs. I don't want to get rid of common sense,

and couldn't if I tried. I want to both reform common sense so that intuitions are done in terms of stronger models, that is to say "uncommon sense", and I want to have people learn processes that are outside that kind of projective thinking altogether.

I don't know how to do this, in spite of many years in the attempt. I can point to things I think are profound about programming, that have enormous reach into other areas of human thought. But I can't say that I or anyone else that I know has ever been able to reach more than the 5-10% that we would expect would be somewhat prewired to learn them easily. In other words, pedagogy is so much more for the "80%" than for the "5%". When the "5%" learn quickly, we shouldn't congratulate ourselves overly much.

See Also