I am looking for a Go training program with the following characteristics:

1) Computer based, either on-line or downloadable

2) Full, explicit curriculum. The trainer would have an explicit sequence of lessons, starting from 99 kyu, and leading eventually to 1 dan. It would be reasonably inclusive of all major Go concepts.

3) Each lesson within the curriculum sequence would have a lesson plan, which would consist of an automated instructional segment and visual demonstration of the concepts to be covered in that lesson. After each instructional segment, there would be a series of exercises; problems for the student to solve interactively, specifically addressing the concepts being taught in that segment.

4) When the student performs satisfactorily on the exercises for a given lesson, the student would be automatically advanced to the next lesson

5) At each stage of the process, the “flow” would be entirely automated. The instructional segment would run at the pace the student sets, and when the instructional segment is finished, the student will automatically be presented with the exercises/problems for that segment. When those problems have been satisfactorily mastered, the student will automatically be presented with the next instructional segment. The “flow control” will be entirely automated, without requiring the student to jump back and forth from instructions to problems, or to some master menu of the program

6) During the exercises, if the student makes an incorrect move, the program will point out that it’s the wrong move, and explain why, making reference to the underlying concepts taught in the current or previous lessons, supported by demonstrations on the automated board. The student will be able to control the pace of the explanations

7) The program should be able to play against the student at some reasonable level (10 – 5 kyu, initially), and record those games in a standard format. The student can then send those games to a live tutor/instructor for comment, if they have one.

I have done a moderate level search (more than a surface scan, less than a detailed search), and as far as I can determine, no such program exists. There are many programs that will teach the very basics in roughly this way, but they all stop at a very low level. There are many programs with good problem sets, and good sets of recorded and commented games. There are several different programs with good AIs.

But as far as I can find, there is no comprehensive, curriculum-based program for teaching Go up to even the 5 kyu level, never mind the 1 dan level, which is my objective. I have an extensive background managing software development projects.

Are any of you aware of such a program?

Thanks,

TCS

Hmm, I'm afraid such program doesn't exist.
As you mention, there are such programs to teach basic concept.
Even as books, there is no such learning method which would bring you from 30k to 1d in a step by step approach. I would say it is critical to associate practice to theory to make progress and in that aspect, playing against human players cannot be replaced.

I think I'd like such a program.

_________________
“The only difference between me and a madman is that I’m not mad.” — Salvador Dali ★ Play a slooooow correspondence game with me on OGS?

Have you seen Bruce Wilcox's Go Dojo? It's pretty much the closest thing I've heard of to what you're looking for (but, of course, the caveats we all mentioned in your first thread still apply - no one is going to vouch that what you're looking for is that important, Bruce Wilcox seems to be a pretty idiosyncratic fellow, etc.)

The closest thing I know of is Bruce Wilcox's Go Dojo It's not exactly what you're looking for, but it's pretty close. It's been around for quite some time, and although it doesn't have all of the features you suggest, it manages to do quite a lot of teaching in an interactive way. It teaches principles first for beginners, and then on up to shodan level. I bought contact fights, and liked it quite a lot - though it didn't propel me to shodan unfortunately. Nonetheless, I'd recommend it. There are some reviews on SL, if you're interested.

ninja'd.

_________________
Patience, grasshopper.

I think it'd be cool if Myongji University provided an online course suited for this through coursera or something...

Actorios:

It is not my intention to delete human interaction from the equation! However, modern education theory is coming around (kicking and screaming, in some cases) to the understanding that the vast majority of "basic" and "rote" education can be done in an automated fashion, and that live human teachers are best utilized to deal with the specific problems that individual students have during that learning process. The more the machines can do, the more the teachers can focus on specific student needs. I expect someone will progress enormously faster using Go For It! (I may as well call it something...) in conjunction with a human teacher, when compared to someone using the program alone. I expect, certainly with early versions, that there will be cases where the student gets stuck in a way that the program can't solve. That's fine. I don't expect real tutors to vanish anytime soon!


Bonobo:

I know I'd like it too. The old saying is that the best way to learn something, is to teach it. So, if I can work with a team to teach a machine to teach Go to the 1 dan level, then perhaps I'll get there myself in the process!




jts:

Thanks for the reminder! I bought Bruce Wilcox's Go Dojo years ago, and put it on a machine I no longer have. For some reason, I didn't like it, but it was long enough ago that I don't remember why. I should probably buy the current version just to check it out. Maybe it's a lot better now. I know it's not (or at least, wasn't) what I'm looking for, but if nothing else, I want to remember why I didn't like it!


Araban:

If a university wanted to develop and/or offer something like this on-line, that would be great. It's a combination education project and software/game development project. Many human educators resist a decline in the requirement for human educators for some reason, but if a university was able to get past that, it might work. My last assignment was conducting a study for the Commissioner of Higher Education for the State of Utah in the US, and we talked a lot about educational theory, including the likely increasing role of automation in education.


Thank you all for your interest! If any of you can provide pointers to existing Go curricula, even if incomplete, that would be a good start for me. I'd be more concerned with breadth (covering the whole range of topics) than depth (how those topics are taught) initially.

For a program to teach all major go concepts as such, many years of further research and go book writing in the directions of Thomas Wolf and myself are necessary. I guess that we will be faster than Myongji and KABA. Currently, by far too little explicit, methodical knowledge is described to allow an expert system program to teach all concepts well. So far, mostly trial and error and mostly for life and death exists, but not explanation of concepts by the programs.

Your wish is roughly 30 years ahead of time.

I completely disagree! Or, perhaps, we have a definitional disconnect.

In any software project where the number of possible variations is huge, you have to completely reject the notion of getting it completely "right" the first time. You plan out a series of releases, knowing in advance that all of the early ones are going to be "wrong", at least in the sense of being incomplete.

Q: How do you eat an elephant?
A: One bite at a time!



That's why the first step isn't to start writing code. The first step is to get a first version of the curriculum baselined. And, the first version will be "wrong", at least in the sense I define above. And that's fine. By making the curriculum explicit, you reveal the holes, which then become available for repair.

I'm probably one of the weaker Go players on this forum. I haven't actually written computer software code in over 20 years. But I've been managing complex technology projects for over 30 years, with the last 20 or so years including software development efforts.

Will it take 30 years to get it "done"? Dunno. I'm not sure I even know what "done" means in this context, or if it's even possible. But that's not the point, or the intent, so it doesn't matter.

Fully funded, I could have a working, usable first version done in a year. Probably less. It will be "wrong", and the next version will be better. As a volunteer project, which is how this is likely to happen, it will probably be a couple of years. But that's fine.

In either case, the first part of the project is to get version 1.0 of the curriculum baselined.

Thx!

Yes.

I have been working on it and its contents for years and started with my books' contents. To reach, what you want, contents up to 1d level expressed ready for an expert system,...



...it will take another 30 years. You don't know, as a beginner you can't. I do know (roughly) because I know the amount of contents and the contents needed, and because I can compare that to what has already been written down.

It will be almost trivial to write the program once the contents will be complete and available. 29 years for research and writing contents, 1 year for the program.

Robert--

I'm not worried about getting it perfect on the first pass...or the second...or the third. My intent is to get something baselined, and then continuously improved over time.

So, it appears we will be pursuing different development paths for this effort. Good luck!

Daal--

I went to the link you posted for Bruce Wilcox's Go Dojo, but it wasn't at all obvious how to download/buy the program. The links to the individual chapters point to a convoluted way to buy those, but it's not obvious that those include the basic program.

Is there any convenient way to just buy it, and be done with it?

Thx,

TCS

They all include the program, yeah. When in doubt, try emailing the guy. It's not a huge corporation with a support center; it's just him. I was able to negotiate a slight discount when I bought both his "dojo" programs/chapters and his ebook together as a bundle, for instance.

_________________
1k KGS

Looks very interesting, but sadly, it’s Windows only :-/

_________________
“The only difference between me and a madman is that I’m not mad.” — Salvador Dali ★ Play a slooooow correspondence game with me on OGS?

Numsgil--

I did, and he already answered. We'll work it out!

Bonobo--

Sorry about that! "GFI!" will start out as Windows only also. For the moment, Windows owns the world...or at least the US market. What are you running? The last I heard, all reasonably modern Mac's can run Windows, albeit slowly. Most of my Unix buddies also run a Windows partition on their machines.

Thx!

I run a Mac, and I also have a virtual machine with Windows Vista (which I’ll have to use for my upcoming job), but I’d really really prefer anything running natively OS X.

_________________
“The only difference between me and a madman is that I’m not mad.” — Salvador Dali ★ Play a slooooow correspondence game with me on OGS?

If it's go software, I'd say write it in Java. It would be the easiest thing to have work on multiple platforms.

This is not a bells and whistles program - it was written over ten years ago and I seriously doubt there would be any performance issues on any system capable of running it.

_________________
Patience, grasshopper.

Well, I have also used Windows since 3.11 (and MS-DOS and DR-DOS before that and CP/M before that) but it (still) gives me itches Nevertheless I’ll consider. Too bad there’s no demo of this …

_________________
“The only difference between me and a madman is that I’m not mad.” — Salvador Dali ★ Play a slooooow correspondence game with me on OGS?

Oren--

I'm a long way from worrying what language to use. It may well depend on the preference of the person I get to write the code! Also, there's a lot of Go software already out there, and while some of it's copyrighted, some of it isn't. If there's a large existing code base in one particular language that the authors are willing to let us use, that will bias the language selection in that direcction. For example, writing a program to display the board isn't rocket science. If somone who's written that well is willing to let us use it, why not? Given the nature of this development process it's pretty certain the code will be highly modularized.

Bonobo--

Sounds like a personal problem to me!



I just bought Bruce's program (again...he gave me a break for the previous purchase!), so I'll report back after I get a chance to play with it. What I'm finding as a problem in every program I've looked at so far...and this may have been the issue I had with GDJ in the past...is the lack of automated flow control. The user experience should be a seemless, automated progression from instruction, to exercises, to next phase of instruction.

I'll post a first draft version of the curriculum sometime in the not too distant future. More than a week, less than a month. It will be wrong, and I look forward to the corrections.