Talk:Marvin Minsky essays: Difference between revisions

From OLPC
Jump to navigation Jump to search
Line 71: Line 71:


[[User:Gregorio|Gregorio]] 09:51, 2 May 2008 (EDT)
[[User:Gregorio|Gregorio]] 09:51, 2 May 2008 (EDT)

I just read an Alan Kay article that says essentially the same thing including making the analogy between learning and Topology! <br>
http://www.vpri.org/pdf/human_condition.pdf

As it happens I'm reading the book "Poincaré's Conjecture" about topology. <br>
http://www.amazon.com/Poincare-Conjecture-Search-Shape-Universe/dp/0802716547/ref=pd_bbs_sr_1?ie=UTF8&s=books&qid=1209742233&sr=8-1

Going way out on a limb here's a new conjecture. <br>
Consider a child's understanding of an idea as a "learning manifold". The child's cognitive maps are "charts" of her own learning manifold on to some other manifold which is being learned (call it the knowledge manifold). All the charts making up the atlas of the child's learning manifold represent the child's full understanding of the knowledge manifold. My conjecture is that the manifolds of all children are homeomorphic to each other.

If we could find that homeomorphism between children's learning manifolds we would have a deeper understanding of what learning means.

For knowledge manifolds, if we could find a simply connected compact manifold without boundary to which they are all homeomorphic, we would have a deeper understanding of what knowledge means!

That obviously needs a lot more detail starting with a definition of learning concepts that can be used mathematically (e.g. what and how many are the dimensions of the manifolds?).

Also, I think a manifold is defined by its relationship to Euclidean space. So the problem of defining the "knowledge manifold" for an idea may be the same as finding the Euclidean version of the idea (like its Platonic ideal). That's not very satisfying as I believe all cognitive maps are meaningful and there is no ideal one. Please correct me if I'm wrong about the relationship between the concept of "manifold" and Euclidean space. At a minimum the relationships seems obscure in dimensions greater than 3.

Just thought I would share a Friday afternoon idea generated by all those neurons and axons which the article brought back to life :-)

[[User:Gregorio|Gregorio]] 13:58, 2 May 2008 (EDT)

Revision as of 17:58, 2 May 2008

What makes mathematics difficult to learn?

Yes, but...

These are all great ideas, and powerful ones. However, if the OLPC project is to succeed on a general level, it has to have something to offer the average teacher who suddenly gets a shipment of these. Thus, it needs to have some response to the following challenges:

  • Why should I change what I'm doing? Anecdotes and disparagement of traditional practices are not enough. The best way to convince people is to get them as close as possible to actually participating. Watching or reading a concrete and detailed account of how it works are acceptable substitutes.
  • How can I apply this in my classroom? Many of these ideas are much easier to apply in one-on-one situations than in a classroom of 30 or more students
  • How can I do this step-by-step? OLPC has a philosophy of leapfrogging some educational hurdles, as developing countries leapfrog wired infrastructure by jumping to wireless technology. But in some cases, this leads to faddish pedagogy; a few overambitious failures can discredit an idea, even if the failure can be traced to lack of planning or some other extraneous factor. Wise educational administrators thus have a suspicion of ideas which are pitched as being so revolutionary that they cannot be implemented in an evolutionary manner.
  • How can I overcome resistance to these ideas? Many teachers and, yes, even students will initially resist change. Generally speaking, the younger you start and the smaller the group, the easier it is. Still, the main answer to this question is a bit of patience: do not expect instant results, especially when applying these improvements to larger and/or older groups. A countrywide implementation definitely counts as "larger".

None of these questions are easy to answer: they all involve sustained effort. Thus, another question arises:

  • How do we, as a project, motivate and sustain the necessary effort? Open-source principles are great, and definitely have a lower critical mass to maintain progress than many other business models. However, the hardest thing to do in an open-source fashion is integration, and the questions above demand integrated answers. I think that in order to gain enough real-world users to have a winning critical mass, OLPC cannot disdain traditional educational models. That means finding/developing integrated TEXTBOOKS and even some minimal support for drillware/quizware. Don't abandon your principles - but don't be so perfectionist that you abandon your principals, either (excuse the pun).
Homunq 15:37, 28 February 2008 (EST)
I don't think that OLPC (or Marvin) is disdaining of traditional education. (In Peru, we are working closely with the ministry of education to complement and augment traditional methods with the laptops. That said, it doesn't mean we shouldn't try to inspire the teachers (and children) to do more. What is missing in my mind is not so much integration as much as a means to share the successes (and failures). --Walter 15:43, 28 February 2008 (EST)

Trust that "open-source" is *the* right way for education. It can solve every one of these problems as soon as people see that it can && decide to do something productive about it.

Minsky's page-less open questions:

  • Ways to connect Students to Mentors
  • Making Intellectual Communities
  • Many Classics are better than Textbooks
  • Developing systems for Simulations

I'll paint a picture. A reverse-engineered social-networking code-base with P2P torrent swarming capacity, each client (including mobile phone && main-frame) simultaneously capable of serving to any other "social" participants. Those who truly appreciate this generous productive style are shaping this zeitgeist into what will someday be recognized as an "Intellectual Community" where Students && Mentors mutually seek each other. Both Classics && Textbooks will be shared communally within our info-sphere (even if that must remain as artificially scarce physical "property" for a while). Our collaborative Free-Software ("intellectual" && not merely "open-source") Community will have the collective computational resources to simulate anything worthwhile. The individuals will be the social arbiters of what warrants bandwidth as we progressively command more. We're accepting applications so-to-speak. It is an organic egalitarian meritocracy where value need not die due to under-marketing or cut-throat opportunism. If one person values data, they will propagate echoing mirrors (with change-logs) to all those nearest them according to abundance of available storage media && shared interests. This thinking cooperative society relentlessly plans && builds unbreakable systems, in spite of grossly inflated "market constraints". The young are inculcated to be hyper-social, through all their modern instant electronic messaging mechanisms, humiliations via online video, collaboration or competition in online games, etc.. It's just kept hard to perceive the ultimate reward for anything from software to curricula being developed for quality ahead of marketability, because there is a reinforcing community that truly seeks quality && can increasingly depend on those of similar inclination to decide what should be "marketed" to me ahead of any but the best executive of a marketing department. Maybe it's all too elaborate for you to see yet, even if nearly inevitable.

Homunq's technical progress issues:

  • Why should I change what I'm doing?

Indeed, anecdotes and disparagement of traditional practices *alone* are not enough, but they *can* lead to "the best way to convince people", at least when those practices are broken, eh? Getting them as close as possible to actually participating, watching or reading concrete && detailed working accounts, or other acceptable substitutes might be missing the forest for the trees when thinking primarily in terms of a particular laptop's marketability (regardless of its lofty intentions... to make billions? ;) ).

  • How can I apply this in my classroom?

Many of these ideas are much easier to apply in one-on-one situations than in a classroom of 30 or more students... && these aren't the *only* ideas that is true for!

  • How can I do this step-by-step?

Value raw sensitivity ahead of brute strength, since it takes some finesse, at least a few precision instruments, && a courage to explore in order to learn anything (including any step-by-step) thoroughly. Make the rest of the steps up as you go (i.e., make your own plans if none are provided or if those available prove inadequate). Feel free to decide for yourself whether or not that's revolutionary in an evolutionary manner.

  • How can I overcome resistance to these ideas?

Yep, change is resisted, generally less with youth because it has been more forcibly immersed in it, but many counter-examples exist. Patience alone will not overcome resistance to ideas. That would need to be accompanied by ideas that are ultimately more persuasive than those mounting resistance. Sure, don't expect instant results, but don't rule them out either. Instant change is possible, even when extremely difficult or otherwise unlikely (especially for widely varying values of "instant"). Universal should count as "larger" still.

Just because particular question answers involve sustained effort, that does not necessarily (or even predominantly) make them "hard", opposite "easy", since that effort need not be sustained personally && could instead be taken on communally or could often be contributed as automation software. We each decide what we consider worthy of sustained effort.

  • How do we, as a project, motivate and sustain the necessary effort?

I'm skeptical you are appropriately qualified to correctly identify "the hardest thing to do in an open-source fashion", especially since "integration" would certainly *not* be it! We integrate languages, platforms, protocols, hierarchies, organisms, societies, etc. in a blink. We cause all kinds of different hardware && software to work well, even in absence of substantial economic interest, because we simply want things to continue to work well or improve over time. Your OLPC questions aren't the only ones to "demand integrated answers". Free-Software facilitates this internet, cross-platform C compilers, && this dynamic environment where ideas can be refined. Go on with your "real-world" this or "critical-mass" that. If you care about money, stick to traditional marketing && educational models. If education is your priority (whether of yourself or others), recognize that non-traditional can be vastly superior when building quality in the open (of course including open textbook-like compilations *integrated* with quiz-ware). Sometimes things should be abandoned (including punny perfectionists).

Walter: Sharing successes (&& failures) is happening here && now. It's integral to Wiki. -PipStuart 84T7YpD (Tue Apr 29 15:34:51:13 2008 UTC)

Pip - this is true! but not everyone has access to this Wiki; so getting more people access, and showing them where to find these discussions to share their successes, failures, and dreams, is important. --Sj talk 05:27, 2 May 2008 (EDT)

Great essay!

I love it! I felt parts of my brain coming to life after years of dormancy just reading through it once! We still have to figure out what this means in terms of what software is built for the XO and how teachers and students use it.

I believe that praxis of software development and of education has to come from a close relationship between the teachers, students and developers. The hardest part so far seems to be finding a meaningful way to communicate between those groups. Maybe a linguist (Chomsky?) or someone versed in educational training (preferably Freierian style) can help find ways to put these ideas in to software and practice in schools.

In any case, we're strongly engaged in the deployment of XOs and building relationships with teachers We'll keep these ideas in mind when we get to discussing the ways teachers work and students learn on a daily basis.

I can't wait to read the other essays (right after I get through my next couple of hours of work drudgery :-)

Thanks,

Gregorio 09:51, 2 May 2008 (EDT)

I just read an Alan Kay article that says essentially the same thing including making the analogy between learning and Topology!
http://www.vpri.org/pdf/human_condition.pdf

As it happens I'm reading the book "Poincaré's Conjecture" about topology.
http://www.amazon.com/Poincare-Conjecture-Search-Shape-Universe/dp/0802716547/ref=pd_bbs_sr_1?ie=UTF8&s=books&qid=1209742233&sr=8-1

Going way out on a limb here's a new conjecture.
Consider a child's understanding of an idea as a "learning manifold". The child's cognitive maps are "charts" of her own learning manifold on to some other manifold which is being learned (call it the knowledge manifold). All the charts making up the atlas of the child's learning manifold represent the child's full understanding of the knowledge manifold. My conjecture is that the manifolds of all children are homeomorphic to each other.

If we could find that homeomorphism between children's learning manifolds we would have a deeper understanding of what learning means.

For knowledge manifolds, if we could find a simply connected compact manifold without boundary to which they are all homeomorphic, we would have a deeper understanding of what knowledge means!

That obviously needs a lot more detail starting with a definition of learning concepts that can be used mathematically (e.g. what and how many are the dimensions of the manifolds?).

Also, I think a manifold is defined by its relationship to Euclidean space. So the problem of defining the "knowledge manifold" for an idea may be the same as finding the Euclidean version of the idea (like its Platonic ideal). That's not very satisfying as I believe all cognitive maps are meaningful and there is no ideal one. Please correct me if I'm wrong about the relationship between the concept of "manifold" and Euclidean space. At a minimum the relationships seems obscure in dimensions greater than 3.

Just thought I would share a Friday afternoon idea generated by all those neurons and axons which the article brought back to life :-)

Gregorio 13:58, 2 May 2008 (EDT)