Best software to review own games

PeterPeter · #1

I currently use Panda to review my games as sgf files. This works well enough at a basic level, but all it is doing is presenting the raw information from the sgf: a simple board, a move tree, and the comments. Is there any free software available that would enable more computer-generated insight, such as:

- The suggested move from a computer Go engine
- Territory predictions
- Highlight dead or alive groups
- Anything else I don’t even know I want yet

Alguien · #2

PeterPeter wrote:

Is there any free software available that would enable more computer-generated insight, such as:

- The suggested move from a computer Go engine
- Territory predictions
- Highlight dead or alive groups
- Anything else I don’t even know I want yet

. Fuego can give you a suggested move (just press f5 to make him make the next move.

. The KGS app (which name I can't remember right now) has a score estimator.

. The SE will tell you what he thinks is alive or dead. Don't expect it to fall very close to reality, though.

. You may want to try to do those things yourself and write in the game file your thoughts. Something like "Here I read [A] as alive, and I think I'm ahead by about 5 points". Then post your game in the appropriate forum or upload it in the teaching ladder to verify your ideas.
I think with each of those "set your thought in writing" reviews you'd get much more strength than ten reviews with automated estimations.

The "make fuego play the other side for half a dozen moves" could help with fights, though.

PeterPeter · #3

Alguien wrote:

The "make fuego play the other side for half a dozen moves" could help with fights, though.

This is a good point. I like try to out ideas - candidate moves that I rejected during the game - and it is difficult to play against yourself, so the idea does not get a proper test.

p2501 · #4

PeterPeter wrote:

I currently use Panda to review my games as sgf files. This works well enough at a basic level, but all it is doing is presenting the raw information from the sgf: a simple board, a move tree, and the comments.

For all this I can recommend CGoban:
http://www.gokgs.com/download.jsp

PeterPeter wrote:

- The suggested move from a computer Go engine
- Territory predictions
- Highlight dead or alive groups

Those are bad things to want from a program to review games. There are programs how can do that, but you don't want to trust their judgement.

Other programs you might be interested in:
http://senseis.xmp.net/?GoPrograms

judicata · #5

p2501 wrote:

Those are bad things to want from a program to review games. There are programs how can do that, but you don't want to trust their judgement.

I respectfully disagree with this. I do agree that you need to be cautious with those suggestions, but I've found that move suggestions from stronger computer programs can help by giving you ideas you perhaps didn't consider. But you must look at the suggested move, and read through it, and make sure you understand why it is better than your move. If you can't understand why, then it may or may not be a better move. Granted, even if you think you understand, the program *could* be wrong, but if you're considering alternative moves and thinking them through, it is a good thing. In my own games, I find that Many Faces and Zen Go (hard to get, unless you're in Japan) suggest a few moves every game that give me an "ah ha" moment.

Territory estimates are also *okay* when reviewing, but count on your own first. Then look at why the computer's estimate is different (if it is). Finally, marking dead or alive groups can also be good--but again, confirm for yourself whether it is right.

The theme here is that, these things can be very helpful if you think independently.

FYI, I think SmartGo is good if you want to browse a database of games, and is decent at territory marking. But I wouldn't use it to suggeest moves--for that, I'd use Many Faces or Zen (there mayu be others, I just haven't used them).

daal · #6

One useful tool when reviewing your own games is to compare your moves to moves professionals might make in similar situations. Programs that do this in one way or another all use a database of professional games. None of the programs that I know of that provide a database are free, but they are also not particularly expensive. A few years ago, I tried to compile a comparison chart of the programs I was familiar with. Perhaps you find it useful - though be warned that this is entirely my subjective impression, partly based on downloadable demo versions. Here is a link to the chart:

https://www.dropbox.com/s/wnjcxzihkdluh ... 0Chart.doc

edit: Just looked at it again - drago is free, but doesn't have a database, and moyogo was free, but now costs $27. Kombilo and Drago have both been updated in the meantime, and are both better than they were then. GoGoD includes a slew of further interesting tools and information.

edit 2 changed link to a later version of the comparison chart.

PeterPeter · #7

Alguien wrote:

You may want to try to do those things yourself and write in the game file your thoughts. Something like "Here I read [A] as alive, and I think I'm ahead by about 5 points". Then post your game in the appropriate forum or upload it in the teaching ladder to verify your ideas.
I think with each of those "set your thought in writing" reviews you'd get much more strength than ten reviews with automated estimations.

I take your point, but this is mostly for fairly disposable, 9x9 games against igowin. I don't think they are worthy of a proper game review by another person, or is that the wrong way to look at it?

oren · #8

judicata wrote:

I respectfully disagree with this. I do agree that you need to be cautious with those suggestions, but I've found that move suggestions from stronger computer programs can help by giving you ideas you perhaps didn't consider. But you must look at the suggested move, and read through it, and make sure you understand why it is better than your move. If you can't understand why, then it may or may not be a better move. Granted, even if you think you understand, the program *could* be wrong, but if you're considering alternative moves and thinking them through, it is a good thing. In my own games, I find that Many Faces and Zen Go (hard to get, unless you're in Japan) suggest a few moves every game that give me an "ah ha" moment.

The problem I would have is knowing why the computer has played the way it has. Is the computer doing a tenuki because he thinks a group is alive without help, dead, or could live with another move but it's not worth it? If a computer can't explain it's reasoning for a way it plays, I don't know how you can understand what it's suggesting in a meaningful way.

PeterPeter · #9

daal wrote:

One useful tool when reviewing your own games is to compare your moves to moves professionals might make in similar situations. Programs that do this in one way or another all use a database of professional games.

I don't understand how that would work, when there is such a vast number of different possible positions on the board.

Nice comparison table - I will check out the options.

alejo · **#10**

I like the way SmartGo helps me during the reviews... and hate it at the same time.
Pros: you can load a huge database from professional games and study fuseki and joseki. The score estimator is quite good and you can analyse the whole game "scoring" each position automatically. This may give you an idea of which moves led you to win/lose

Cons:
1.It's not free.
2.The tree variation system is awful. Sincerely, it's way below kgs' or any other tree style. Although there is an option for autosolving tsumegos, it doesn't work most of the time.

Depending on your level, having an AI may help you by suggesting moves.

jts · **#11**

judicata wrote:

p2501 wrote:

Those are bad things to want from a program to review games. There are programs how can do that, but you don't want to trust their judgement.

Territory estimates are also *okay* when reviewing, but count on your own first. Then look at why the computer's estimate is different (if it is). Finally, marking dead or alive groups can also be good--but again, confirm for yourself whether it is right.

The real problem with territory estimate, imho, is that seeing all of those spaces with black or white spots on them fuels the popular illusion that certain areas of the board "belong to" one side or the other, and that scraping away your opponents' "territory" is a self-evident success, while having your own scraped away is a self-evident failure.

A few areas of the board really are secure profit in this way, and sometimes a well-timed invasion really stings; but normally you can be ambivalent about an area you can control - a chance to block it off and put cash in the bank will eventually be good, but a living enemy invasion which reflects your power towards the center will be good too.

PeterPeter wrote:

Alguien wrote:

You may want to try to do those things yourself and write in the game file your thoughts. Something like "Here I read [A] as alive, and I think I'm ahead by about 5 points". Then post your game in the appropriate forum or upload it in the teaching ladder to verify your ideas.
I think with each of those "set your thought in writing" reviews you'd get much more strength than ten reviews with automated estimations.

I take your point, but this is mostly for fairly disposable, 9x9 games against igowin. I don't think they are worthy of a proper game review by another person, or is that the wrong way to look at it?

9x9 games are excellent material for review at your level, because you probably get into situations where tactical advice would be helpful within a few moves. Games that don't involve serious thought on your part, or don't involve an opponent you respect, may be less suitable for reviews (either from other people, or the sort of in-depth computer analysis you're contemplating)... but you could post it anyway. I know there are several people on the forum who are fans of reviewing all sorts of games - blitz games, lopsided victories, you name it.

EdLee · **#12**

My favorite Go editor/reviewer is MultiGo, completely freeware:

http://www.ruijiang.com/multigo/

judicata · **#13**

oren wrote:

judicata wrote:

The problem I would have is knowing why the computer has played the way it has. Is the computer doing a tenuki because he thinks a group is alive without help, dead, or could live with another move but it's not worth it? If a computer can't explain it's reasoning for a way it plays, I don't know how you can understand what it's suggesting in a meaningful way.

I share the concern, and definitely think people should be aware of that. But having tried some programs myself now, they really can help. For example, if I come to a move I wasn't sure about in a review, and then I see the suggestion, I'm often reminded of a principle/joseki/shape that I had been taught (this is more often the case when a suggested move hadn't even crossed my mind...though that sometimes means the suggested move is ridiculous, which is why you still have to think it through). As another example, if a suggested move is one I had considered and rejected, I look at it more closely and realize that my reason for dismissing it was wrong (either I read the sequence incorrectly, or I realize the result was better for me than I realized).

I want to stress that programs are in no way a replacement for live (including online) instruction. Not at all. Really I view programs as a supplement--to give you ideas while reviewing your own games. I'd say a player 4 stones stronger than you, teaching you live is far better than a computer program 10 stones stronger than you.

ProtoJazz · **#14**

Many Faces has a few interesting features like that, it will give some move sudjestions, as well as tell you the status of a group (Such as needs to fight to live, alive, dead, and I think it has a few more like probably will live, or will stuggle and not likely live) though there may be better software for that, I am interested in what other people recomend in this thread

lleig · **#15**

I'd suggest Kombilo:
http://www.u-go.net/kombilo/

My favourite feature is the "Guess" mode, very useful when watching pro games.

daal · **#16**

Did anyone look at this link from my post above?

https://www.dropbox.com/s/wnjcxzihkdluh ... 0Chart.doc

PeterPeter · **#17**

This is probably a long shot, but is there any software that shows a graph of a game, with each move along the x-axis and the predicted final territory up the y-axis? This would be really useful for seeing the flow of the game, and picking out the good and bad moves.

Rémi · **#18**

PeterPeter wrote:

This is probably a long shot, but is there any software that shows a graph of a game, with each move along the x-axis and the predicted final territory up the y-axis? This would be really useful for seeing the flow of the game, and picking out the good and bad moves.

Crazy Stone can produce an analysis that looks like this (click on move for detailed move analysis):
http://www.grappa.univ-lille3.fr/~coulo ... /Uberdude/

Or that (weaker game):
http://www.grappa.univ-lille3.fr/~coulo ... /bofinken/

Rémi

PeterPeter · **#19**

That looks almost perfect!

But I'm not sure it justifies that much expense at this early stage of my Go career...

PeterHB · **#20**

PeterPeter wrote:

This is probably a long shot, but is there any software that shows a graph of a game, with each move along the x-axis and the predicted final territory up the y-axis? This would be really useful for seeing the flow of the game, and picking out the good and bad moves.

SmartGo (£30) does this.
I think it is the closest analogue to Chessbase and the thing I happen to have. A direct match (which is what you are angling at) doesn't exist. This is to do with the way Chess engines can trim a lot of future lines easily, due to the losing player being down a piece with no compensation. The equivalent in Go, ( being down a large dead group ) isn't clear to computer engines. That is just the current state of the art. Wait 20 years and the answer will be different. But SmartGo does have the predicted territory graph, engine, database with updated pro games, integrated feel.

With the caveat that the following weak/stronger notes are my personal belief, not rigorously tested, I think the programs to compare SmartGo against are:
Manyfaces of Go (£56). Stronger engine , no database with updated pro games, but many other instructional tools, e.g. Joseki tutor.
Go++ Deluxe (£56). Stronger engine, no database with updated pro games, shape feature.
Drago (£0) No database with updated pro games. Plug in the weaker GNU Go (£0) engine.
GoGoD (£20) database with updated pro games and historical reference material. Not a program itself. A reference.

Your best bet though is to search for a real live player™ at a club.

Best software to review own games

Who is online