HexWiki is back

Some more advices that aren't yet found in hexwiki:

take a look at "Bridge templates" here: http://www.drking.org.uk/hexagons/hex/templates.html

broadly speaking, you've to try to make these appear in the board as you play a game.

Also, you should welcome these patterns: http://www.trmph.com/hex/board#19,i7s19g8s18h10r19o11s17n14r18e17q19h17

They're not yet connected but they're easy to connect (they're 1 move away from a connection).

I've updated the initial section of Strategy page on Hexwiki. Feedbacks welcomed.

I've also made significant edits on Rules, Swap rule, and the Home page. I hope I've made them less verbose and more interesting.

I hope you like the new Home page. The pictures are there because I had them at hand and they seem to me amusing and pleasant.

Checklist of what has to be done in my opinion:

1) Give an icon to the website!

2) We need to be able to hide irrelevant cells, because things like this are unacceptably ugly:

http://hexwiki.amecy.com/index.php/Advanced_(strategy_guide)#Template_Vb

3) We need hex boards to "float" on at the right of text, so that it's possible to read both text and look at the board.

I've forgot one:

4) Disable tool tip when mouse is over hex boards

lazyplayer, thank you for your advice!

Thanks lazyplayer, strategy page looks really clean now.

However, as opposed to the "checklist for beginners", I support reinventing the wheel for the sake of deep understanding!

@lazyplayer:  Put your list somewhere on the wiki.  I have some things to add.

Any ideas for the favicon?

Tommah, http://hexwiki.amecy.com/index.php/About_HexWiki

It's also linked from the home page, so people will see it :D

Arek, well, that advice is on how to win, and not how to manipulate people into working for us :D

Most important change in that section (was called "general advices", now "checklist for beginner") is that i hope I've restored symmetry between attack and defense. I think there is only one asymmetry, that attacker has one move less on the board. But once one learns to take into account for this (with the method specified here, by studying colors separately) then everything should be fine. Well, at least I hope so.

By "attacker" i mean, the one with the right to move. Let me re-state my point in clearer way: I think the only advantage of looking at the board as if the opponent wants to connect and we want to block is that, when it is our turn, the opponent has one additional stone on the board. If we take a "defensive" point of view like this, we do have to defend against ALL possible next move by opponent. But on the other hand, if we take an "offensive" point of view, then we do have to find a win for the player with fewer stones on the board, it is true, but we only need to find ONE of these. In my opinion, "offensive" point of view is at least as valid as "defensive" point of view.

While the symmetry between attack and defense is obvious from a mathematical point of view, I am not sure if the same is true from a learning point of view. I clearly remember that 'don't try too hard to connect, just try to hinder your opponent from connecting' did help me a lot as a total beginner. (By now I am thinking more symmetrically again. But I'm not a total beginner anymore I hope.) It's just a psychological thing I guess, and it may be very subjective of course.

richy, in fact in an early revision of the hexwiki advices i had tried to make a distinction between the start of the game, where one has to be careful to not confuse himself into thinking that one trivial win is at hand, and the later stages where indeed trivial wins are there to be found. i think in this sense defensive is good at the begin and it's probably due to this reason that it seems to make sense psychologically. But this is difficult to pin down without going into a circular argument! Because length of the game is defined according to perfect play, and one doesn't receive a message like "hey, game is about to end (under perfect play), please look for winning combination now"!

Ironically, the beauty of the current formulation ("Strengthen/[weaken] your/[his] potential [winning] combinations") is that it's vague on this point. It doesn't say if you have to strengthen just the one or two obvious winning paths or the millions of possible paths. And so it's always correct in every stage of the game... :)

From a pragmatic point of view, i think we can say this. This idea of playing "defensive" in some (unspecified) sense is already in the 3 guides (out of 3) that we've in Hexwiki (and for David Boll's FAQ, I'm the one that has copied/pasted it there). There is no urgent need to repeat this questionable advice at the top of the strategy page.

This weekend I will try to add a forum to the wiki, so that we don't have to monopolize LG's forum  :)

Also, I have registered hexwiki.net and have pointed it at the wiki.  Let's use that domain from now on; it is more dignified than hexwiki.amecy.com (but I'll keep this domain active too).

richy, i've tried to reinsert the concept of game length and "defensive" play in the checklist. i hope this is as clear as it can be.

naturally we would also need more figures and examples to really explain what we're trying to say there, but this would require another "guide"...

"Hex is not like a race toward a goal; it's like proving a theorem in small steps."

I hope you guys like this. I really can't find a way to explain the "defensive" point than without a silly analogy (and anti-analogy) like this!

Here is what I think.

In a defensive approach we look more at opposition stones.

In an offensive approach we look more at our own stones.

The objective of defensive approach is to avoid losing. If we succeed then all of a sudden we win.

The objective of offensive approach is to attain a win as quick as possible. In the meantime we might, all of a sudden, lose.

Now I see a big difference between the two. The attacking way one can repeat and make principles to play accordingly. In defence one will always be dependant on opposition play so principles are harder to define.

In my play I use offensive approach for 90% of the effort.

Arek, for me it's like 75% of effort is offensive and 25% defensive. But it also depends on how much respect I've for opponent and his play. When i think he is playing very poor, my brain simply refuses to look for his win, because i "know" there is none. On the other hand when I think he is totally winning, then it's more or less 50%/50%. And empirically it seems i usually play best when i think opponent is totally winning, hehe.

Anyway, i've re-edited it again 100 times. This time I've decided to replace the idea of stages of the game and not attempting to win too quickly into a more general and I think probably better "precautionary" principle. This should be a sound advice for all levels i think, even for you Arek! :-)

"Plan for the the very best play by opponent and weak and simplistic play by you. "

The idea is to assume weak play in future for our side and strong play for his side. This should be orthogonal with defensive/offensive. That is, it should be possible to look for wins for either sides while still assuming that one side will play weak and simplistic and the other will play strong and near perfect.

Naturally the goal of this is to allow us to make some errors in our planning and still fall by accident into a winning position.

lazyplayer, "looking for wins" leads to problems even when assuming our own weak play and opposition perfect play.

Probably for most players, me too, strategy of "looking for wins" starts at some stage of the game. In my case this stage is always played the worst by me, because even with all above assumptions, at some point I think I've found the winning way. Usually I'm wrong.

In earlier stages I don't look for wins, I just compare if a single move is visually superior to other options. The "early stages strategy" gradually took bigger part of my game and recently I've decided to try to play always like that, until the very end.

Concluding, my advice is: instead of assuming weak play in our analysis, just don't do analysis! (at least any deep analysis)

Arek, actually i agree "detailed analysis" (searching for the "principal variation" like computers) is completely the wrong way to go most of the times.

But it seems to me that it makes more sense to "optimize" opponent's moves than your own. Maybe this is why "defensive" play seems intuitively right for people?

In earlier revisions, i tried (two times) to make a distinction between stages of the game, but overall i failed and re-edited this way.

Or maybe I'm just wrong, and even in this case, there is no distinction between players. Indeed this is probably the case, it's just illusions...

Ok, i edited away the dubious points,  "don't reinvent wheel" part and the "plan simplistic moves for your own side".

We only want the more sure and proven advices there.

Arek, for explaining josekis, we can create many sub-pages, but first we've to agree on a naming convention. I say for acute corner josekis, let's name them by coordinates, and place them on top-left. For obtuse corner, i propose to use a silly coordinate system like "A-1" is the A column and minus one row, that is the last row! :D

I also propose the same system for naming edge templates. Current naming system doesn't make any sense to me (who remembers what is "Tempalte III-1-B" ???)

Ok, i've erased my checklist. Maybe same ideas will reappear in a "self-contained tutorial" one day, maybe not...

Explaining josekis is a huge task. By "explaining" I guess you mean showing all variations that are objectively suboptimal and describing why they are. For this there should be a Q&A system where we could create the database step by step.

I agree a naming convention would be helpful, although I don't see how to apply naming by coordinates. Obvious convention is already applied, that is "n-th row joseki" when the first move is on the n-th row. For further distinction I don't have any practical idea hence I used a random order.

Btw. there are only acute corner josekis. Obtuse corners give freedom to play whatever like center. @lazyplayer, I know what example you are gonna show: http://www.trmph.com/hex/board#13,b10c10, yes, we can see this pattern quite often, but it's not a joseki - almost always you can as well play tenuki there.

@lazyplayer: I have also thought about finding a better way to name the edge templates.  I think the easiest way for beginners would be to name them after the shape of the carrier.  I'll write my ideas up on a wiki page.

Tommah, yes, that is a reasonable idea and i support that too. I think both naming schemes would be good. The coordinate scheme has the advantage that it shows you the template in exactly the same place where it's likely to happen. It is a way to unify treatment of templates and "josekis".

Example:

http://www.trmph.com/hex/board#13,a4m13c6 vs http://www.trmph.com/hex/board#13,e4m13g6

These two are the same edge template, but naturally the name of this is "a4 c6", not "e4 g6". Because that is where one is likely to find it in practice.

@lazyplayer, we should generalize more, these are all the same thing:

http://www.trmph.com/hex/board#13,a4m13c6

http://www.trmph.com/hex/board#13,a3m13c5

http://www.trmph.com/hex/board#13,a3m13d6

http://www.trmph.com/hex/board#13,a4m13d7

http://www.trmph.com/hex/board#13,a5m13c7

etc.

Arek, and sure there are obtuse corner josekis too, like the one you pointed out, but the problem is they also depend on the *other* edge...

And this is why we should not play them at the start.

A good definition would be (I edited wikipedia's definition):

Joseki is a studied sequence of moves for which the result is considered _balanced_ for both black and white sides on an empty board.

These sequences only appear in acute corners.

In obtuse corners there are "patterns" that are playable for particular positions, but not on empty board so they cannot be objecively defined.

Example:

this is very often played http://www.trmph.com/hex/board#13,b10c10

but on an empty board black is happy to have this http://www.trmph.com/hex/board#13,b10c10b11c11b12

http://senseis.xmp.net/?Joseki

i think any pattern with fixed coordinates is should be considered "joseki".

Ehehe, their page on Joseki is very good, and so is their page on Strategy. It's very Wikipedia-style!

Take a look: http://senseis.xmp.net/?Strategy

I'm tempted to copy and paste from there to our hexwiki :D

I'll add something as a passable go player and a hex neophyte.

If you're keen to follow the go definition, then it's worth stressing that the result of a joseki need not be balanced, at least in the sense that you'd be willing to play it immediately after all four corners are occupied.

One reason is that you have to play a joseki in the context of the whole board. A result that's normally good can look ridiculous if there are other stones on the board that make it ineffective. But a more striking reason is that the result may always be uneven, but be considered acceptable. One reason is that there are joseki that start after a player (Black) has two moves in the corner. The result will not be "even", but it might be the best that White can get. Another reason is that many joseki leave one side with a group that is under attack. In some cases, it may be worth giving up a lot to avoid being attacked.

Considerations about attack and defense don't really apply to Hex, but I suspect some of these ideas might be relevant to how hex players use the term joseki.

hyperpape, in hex, the equivalent of a capture is to make a stone (or several of them) completely useless. What has no equivalent in hex is that you do need to maximize your score AND minimize score of opponent at same time. In hex, at least in theory, you can just focus on your own (or on your opponent's) connection.

Naturally, "balanced" should be relative to number of stones played in that corner. I think this was implicit in Arek's proposed definition. Anyawy I agree with the fact that his definition is excessively restrictive. We should use joseki like in Go i think: all reasonable sequences that tend to happen often in same position of the board.

Example, consider this potential "Joseki". Black can now play elsewhere, and eventually when play will reach that corner, white will probably not have enough uses of his stones there. So white group is in some sense "dead", but this is NOT obvious AT ALL at first look. It's just experience that taught me this.

Anyway you're invited to contribute to hexwiki too even if you're not strong at hex. If your contributions contain errors or dubious statements, we'll just edit it away and replace with something else. Everyone acting in good faith is welcomed! :)

Indeed even just adding questions or "TODO" would be useful. Someone will fill the blank one day...

@lazyplayer, your example is obviously a valid joseki, it is C1) here http://hexwiki.amecy.com/index.php/Joseki#4th_line_josekis

I agree my definition is very restrictive, but I think it's proper so that we separate them from other stuff. Corner fight is a very special thing and is very restrictive itself because one cannot play similiar to joseki! Playing in the acute corner else than any known joseki is just wrong and weakens ones position. Maybe in my definition I should have mention something about moves being almost forced.

@lazyplayer, I also made it restrictive on purpose so that you stop saying that there are obtuse corner josekis. In obtuse corner a fight is free in terms of choice:

http://www.trmph.com/hex/board#13,e9b12

http://www.trmph.com/hex/board#13,e9c11

http://www.trmph.com/hex/board#13,e9d10

All above are correct and none of them has any default answers nor they are forcing to play locally.

Arek, same thing on obtuse corner:

http://www.trmph.com/hex/board#13,b11 <- weakens your position

http://www.trmph.com/hex/board#13,d9 <- weakens your position

http://www.trmph.com/hex/board#13,e8 <- weakens your position

see? there is a pattern here

@lazyplayer, I don't understand how is that related to what I said

I added a forum to the wiki.  You can access it at http://www.hexwiki.net/index.php/Special:WikiForum or from the "Forum" link in the sidebar.

Please let me know if it works.  If it does work, let's continue this discussion on there.

Thanks Tommah, great job! I posted to the forum. Seems that it works fine.