<Waiataka's, deep-breath's, and daibertz's forums all branch from 19b'th move, so why not merge them into a single forum, and reserve the "freed up resources" to host later variations that descend from 19b.>

I'm trying to understand what you are saying here (excuse me, I'm a little fatigued right now). All these forums (plus a couple of others) are descendant from <Deep Breath>'s forum on the previous move, while the previous move's forums were also descendant from his forum from the move before last (see <kutztown>'s bio for the history of the last 4 ply).

Aren't we already branching each move's descendants as they're made?

I wonder whether combining the three 19b forums wouldn't overload the combined forum?

<Let each forum be at a different depth>

This is what I don't understand: if each forum is focusing vertically rather than horizontally, how do we stop them from overlapping? Analysis will often produce deep lines that will descend all the way through descendant forums sometimes to 30 or 40 ply beyond.

Is someone going to edit these analyses into the tree, locating them at the branch points from which they stem, and then continue to brachiate them as more lines are produced?

Our current system seperates engineAnalysis from the general human-input. My method integrates the two.

Yes we're unofficially mimicking a data-tree. Why not make it official so that we can capture the 2-3 extra benefits.

<I wonder whether combining the three 19b forums wouldn't overload the combined forum?>

One forum would not host all them. Instead, one forum would host the "top 1/3 of content" (vertically) and the others get middle 1/3 and bottom 1/3.

Whenever a forum feels close to getting full, the tree "sends you" to someone else's forum depending on which branch of the tree you depart on. A forum using data-tree design can adequately handle 3 or 4 full alternative positions that stem from the same parent-position. Obviously the best 1-2 of these 4 positions will re-link almost immediately to someone else's forum, exactly like <15. cxd6 Nxc6 16. Bf4> on my demo did. Since it is the main line and packed full of ideas/analysis/highest-priority of the four.

This way the data-tree will continuously prioritize itself by resolving positions that can be easily resolved in less than a half page. And the highest priority branch of every forum will be the de-facto main line, and will always be linked to another forum where it is the root-node (which the whole forum is devoted to)

Just like the opening explorer, there are more games behind 1.e4...c5 than any other, and it is the highest priority etc... and eventually when you explore the sicilian deep enough, you find branches that only represent 1-single-game. This can symbolize a "resolved branch" and the data-tree will continuously explore high-priority branches and resolve those which aren't linked out (because at that moment the team decides no whole forum should be devoted to it).

<This is what I don't understand: if each forum is focusing vertically rather than horizontally, how do we stop them from overlapping? Analysis will often produce deep lines that will descend all the way through descendant forums sometimes to 30 or 40 ply beyond.> Yes there will obviously be some overlap as there is now, and will be dealt with as best we can. Overlap is not but a minor problem, and having visual chessboards of the FEN diagram will make users say "wait, i saw this diagram on another forum--and i'll alert both forum hosts of the transposition" and this example would lead to 2 different forums hosting a common branch, would both link to a 3rd-forum for that move if it merits a whole forum.

Kutztown assigns forums for the same-depth-only.

Let forum hosts model my demo and host the 3-4 most likely replies.

Once one of those replies start to become so highly analyzed and become way out of proportion compared to the other moves, then it gets a whole forum dedicated to it.

The other 2-3 positions can probably be resolved and adequately covered as just a single section on a forum.

For example, in my demo, I hosted 15. cxd6. My forum was dedicated to the move 15. cxd6 but I made individual Sections representing replies to it.

Section V and VII were weak replies and these moves were fully discussed and concensus was reached at these moves and we considered them adequately resolved for the moment.

Section VI is black's best reply and has the most work to be done. Upon examing only 1 depth of replies, we immediately created Tybate's forum to host a 16th move which will itself generate black's replies (the Sections of Tybates forum) and white's replies (the subsection aka subdescendants). All the branches on Tybates forum will branch one or more times and will ultimately die on that forum, or will link to another forum, which analyzes black's possible replies (as Sections on the forum devoted entirely to a subdescendant branch of Tybates forum which is a subdescendant branch off my forum). Thus that newly created forum is deeper than Tybates forum and Tybates forum is deeper than my forum.

This isn't to say that I won't have branches that might be deeper than some of Tybates branches, but they will always start shallower.

If one branch on my forum branched to Tybates, and another branch on my forum merited a whole forum and branched to ZeroOne, then Tybates and ZeroOne would indeed have "horizontally the same" forums. But since they don't cover the same move (at least not the same root move) and they both travel in the same direction, then they are actually parallel lines both traveling down with occasional overlap due to transpositions.

Forums are "freed" when say my 15. cxd6 is parallel to a forum hosted by Stolok devoted to 15. Na4.

Since we will either do one or the other, the 15. Na4 will get freed once the vote is official, and all the forums that 15. Na4 linked to and all its descendants are freed/useless and are recylced to host branches off someone else's forum.

Under this system, new forum assignments are always a branch off another forum. A "living data-tree" because at each vote finalization, forums are made obsolete and are made new as branches that are the best places to bud from the remaining tree.

If Kutztown and the team give it the go ahead, the only implementation would be for every forum host to create between 2-4 additional FEN diagrams and Assign them a subordinate Section.

Simple as that, have each forum host create FEN diagrams for the highest priority/most important replies and give each one a Section-header like my forums.

I'll even design the layouts and copy/paste to the forum hosts and will have them ready for the team at a wait-time of about 30 minutes per forum and can do them as soon as tonite and finished before Sunday.

So the easiest way to adopt my proposal is to make NO CHANGES to the forum assignments, but only to make FEN-diagrams of the absolute best replies (2-3 is plenty).

That's it and the rest will unfold itself once people start disassociating 1 forum = 1 FEN diagram and start learning to accept 1 Primary FEN-diagram and subordinate sections each having 1 FEN diagram for a total of 3 or 4 FEN diagrams per forum. Just like my demo. It is 1 primary move 15. cxd6 and 3 replies each with a section-header.

Your energy and enthusiasm is amazing. You are obviously very intelligent and apparently you also have lots of time on your hands. I hope that we can find a way that you can make a strong contribution to the team.

So far, despite your impassioned attempts to rally the team around your ideas, only <twinlark> has responded to your appeals, and that with more questions. I think many on the team are interested in the concept of a data tree but do not have a clear understanding of how it would work. Regarding our current forum system, you say <I don't think the forum system is organized well enough to have my contributions worth any value to the team.> but I believe most team members think the existing system, while not not perfect, meets our needs. The system has evolved slowly over the course of three games. I don't think very many (including myself) are interested in a major change at this point, in the middle of a game.

So I do not plan to change the forum system in accordance with your wishes at this time. HOWEVER, I am hopeful that with the help of cg.com, we can work together to develop something better for future games. I am glad you abandoned the Googledocs idea. Many on the team feel a loyalty to cg.com and want to use tools provided to us. That does not mean we can't ask for better tools.

Someone suggested that a data tree can be modeled after Openings Explorer. I think this is a good idea. An enormous amount of details would need to be worked out, but my basic idea would be that the "database" would be analyzed lines rather than actual games. Team members could drill through the moves as we do on OE now. At any "node", perhaps a kibitzing page could be created, where team members can discuss that particular position. With a system like this, we may very well not need forums. The new tool would function as a data tree, index of variations and analysis forums all at the same time. It would also make it very easy to find previous analysis.

As far as I can tell the current situation is that the forum owners don't need to do anything. Kutztown just assigns the forums and people emerge into it and analyze the given line. And if kutztown sees it fit (if there is/are some forced lines after each move), he extends the forums to cover more plies.

After listening to other people's feedback, I scaled back my proposal to a very simple suggestion.

Making "1 deep" forums out of the current "flat forums". Have each person make 2-3 immediate replies to the move which their forum is dedicated to and letting the team see the usefulness of a extremeley scaled back data-tree.

Let each forum which currently is dedicated to one move, now have 2-3 subsections, and have a FEN diagram for each. That is all, nothing major, and I feel that is the most that I can persuade the team to adopt at this moment. Just a small minorchange. Add 3 more FEN's to each forum that hosts a single move.

So the revised summary is now as scaled back as I can make it, and that is for us to adopt a "1 deep" data-tree. Have the main move hosted (as it is already now i.e. the temp forums) plus a measly 2-3 replies with a labeled FEN diagram and section header (SectionI SectionII and SectionIII) is all I ask.

Abandoning googledocs reflects my team-skills. I have worked with teams heavily for the past 4-5 years, and I'm a concensus builder, a listener, and an encourager.

I want the team now to just add 2-3 descendant FEN diagrams per forum to the current system.

I wish I left a link to Artar's post which I left about 12 hours ago. Here it is, I have scaled back my proposal tremendously after listening to Twinlark's feedback that the complexity is a bit too much, so I have simplified it to its barest. A "1 deep" improvement over the current "flat" forum headings (I know the forums are deep, but the forum-headings stand to be improved).

<So 2-3 more FEN's per Temp-forum, and those 2-3 moves to immediately descend from the main position which the forum is dedicated to. I feel that this alone will finally quench my dire feeling to contribute to our team's organizational structure.>

Once we do this, I'll fire up my rusty 1.8Ghz machines and will start posting 25-ply lines (take 24-36 hours) four at a time. And I'll leave the forum system alone and will devote my energy to churning data and sliding into some deep lines.

Here's the post on Artar's forum that I left 12 hours ago, after reading Twinlark's feedback.

Artar1 chessforum

I'm a listener, the team spoke, and I have responded by scaling back my proposal to its bare-bones. Any incremental improvement towards the data-tree system at this time will be a profound success in my books. I've spent a month and a half on trying to encourage people to give my ideas a chance, but my failure was in not reminding myself that not everyone here at chessgames has a DIS computer background like myself, and while I could convince a team of my old classmates to adopt a data-tree system, its a different story here.

I like correspondence chess because of the technology, so naturally I try to make improvements there when a member on this team, but we're all here from various backgrounds and different countries even.

I hope this issue of adding 2-3 descendant FEN's per forum, for every forum that is dedicated to a specific move, will be resolved this weekend.

I am one of a thousand teammates, and if I can make this small contribution, then I am happy with it.

I am in the process of finding one forum host to try your latest proposal on a pilot basis. If it works out, we can migrate it to the other forums.

<sentriclecub> I'll need for you to look over the changes made to my forum so that you feel I am giving your proposal a fair demo. I'll take this opportunity to respond to some of the comments you made earlier. I can appreciate the attempts you have made to devise an efficient, self-organizing data tree. I can sympathize with your disappointment in trying to convey your message to a group where CS or IS backgrounds are not a given.

I for one *do* have a CS/IS background and recognize the similarity between your proposal and the algorithms used by chess engines in evaluating positions. The seeming lack of support from the Team I attribute to several other factors in addition to ones mentioned previously, which I won't rehash:

The method is complicated - logical and efficient, yes, but also complicated and requiring a computer-like discipline we do not have. It's hard enough getting Team members to post to the right forums or forum hosts to summarize their moves. The data-tree method (as I understand it) requires self-organizing forums to open, branch or close as forum hosts see fit.

The data-tree requires that members be able to recognize positions they have analyzed previously or in other forums (like an engine hash table). I might just be getting too old, but I can't do that. I cannot instantly recognize a position I have run across in another forum and alert the host that lines have transposed and a duplicate forum can be closed. The methodology requires mental effort in addition to the primary objective of finding the best chess lines.

I think the toughest obstacle is that while you had a clear concept of your data-tree proposal, the implementation details were constantly evolving: from googledocs, to wiki-based, to forum-based. Without a concrete and specific plan, the Team simply carried on with the game while perhaps waiting to see what you could actually show.

and Thank you Zanshin!!!

I would say the Zanshin experiment is already a success, as it has motivated me to fire up 2 of my 4 computers for contributing analysis to Zanshin's forum since it has receptor sites for my data output. Pardon my hybrid ComputerScience premed jargon.

I have all this free time because I have finished my undergrad (double major) and am taking 2 years off to study for the MCAT (gotta score a 41 which is 98th percentile). I only sleep 6 hours a nite thanks to meditation I do through self-induced hypnosis that I learned/stole from my ex-classmate/friend who was Buddhist.

I used to play chess on FICS and construct variations of the Sicilian which I can call my own, and I made elaborate data-trees and have made over 70 branches for my invented line of the Sicilian. However, I reached diminishing marginal returs as I really reached the point to where I'd have to triple my node-count of positions I've studied, just to keep my opponent 2-more ply deeper in my pre-memorized replies before they get me out of opening book. Chessgames (the Yury game mostly) started grabbing my attention because team correspondence chess "the world vs 1 person" should really always be won by the world, or at least drawn, because the resources held by a team working together are astronomical when they follow a data-tree format. Imagine RV getting depths of 30 on his Rybka's, that's almost the whole game.

Zanshin informed me tonite that he has a computer science/Decision Information science background and he could tell you, that if 100 people like he & I armed with 2.0 Ghz computers and creating an organic/living and self-updating data-tree would be unstoppable and could never lose.

I was a spectator in the Kasparov-WOrld game when it was played when I was 15. I remember not being able to understand why there's an advantage for having more time per move (I also didn't know chess and my elo back then was probably 400). Everyone who was active in that game would undoubtedly remember Irina Krush's role and her rise to becoming the world team leader because she understood data-tree-cognivity and did a wonderful job at marketing it to defeat the inertia of the skeptics but mostly that of the too-passive participants

Here's a quote from Irina Krush's entry on wikipedia

<In addition to discovering a new move, Krush displayed great patience, tact, and generosity while participating in the bulletin board discussion of the game. The analysis tree she maintained gained widespread acceptance, and the recommendations therein were chosen by the team for more than forty consecutive moves. Even grandmasters rated considerably higher than she began making primary reference to her analysis tree.

Thanks to Krush's efforts, the internet team worked in a more coordinated fashion. Unfortunately, on the 58th move, her recommendation was inexplicably delayed from appearing on the MSN website, and the team voters chose an objectively weaker move.>

I've even done a school report on her both in High School and 3 in college. Also I feel like I am getting smarter by getting involved in correspondence chess. My elo has grown 100 points since the Yury game (but credit also to the beginner's curve) and I believe that any OTB player who hasn't played correspondence chess can gain a similar leap once they "think data tree".

I know its a long post, but I wanted to tell you more about myself, so you know that you helped a real person tonite get his ideas into a correspondence chessgame between two amazing opponents.

I agree, that's why I know your job is 50x harder than it seems. You have to invest analysis time in people's forum before they even do.

*** <Analysis of 16. Bf4 Nxc3 17. bxc3 Nb5 18. Rxb5 cxb5 19. Bxd6 Be6> User: Deep Breath

*** <Analysis of 16. Bf4 Nxc3 17. bxc3 Nb5 18. Rxb5 cxb5 19. Bxd6 Bc6> User: dalbertz

*** <Analysis of 16. Bf4 Nxc3 17. bxc3 Nb5 18. Rxb5 cxb5 19. Bxd6 Re8> User: Waitaka

*** <Analysis of 16. Bf4 d5 17. Be5> User: mckmck

*** <Analysis of 16. Bf4 d5 17. Rbc1> User: zanshin

*** <Analysis of 16. Be3> User: Red October

*** <Analysis of 16. Rxd6> User: capafan

*** <Analysis of 16…Nxc3 17. bxc3 Nb5 18. Rxb5 cxb5 19. Bxd6 Be6> User: Deep Breath

*** <Analysis of 16…Nxc3 17. bxc3 Nb5 18. Rxb5 cxb5 19. Bxd6 Bc6> User: dalbertz

*** <Analysis of 16…Nxc3 17. bxc3 Nb5 18. Rxb5 cxb5 19. Bxd6 Re8> User: Waitaka

*** <Analysis of 16…d5 17. Be5> User: mckmck

*** <Analysis of 16…d5 17. Rbc1> User: zanshin

If by some chance GMT plays <16...d5>, could I request you designate a forum looking at

<16...d5 17.Be5 Re8 18.Bxg7+ Kxg7 19.e3>

or even

<16...d5 17.Be5 Re8 18.Bxg7+ Kxg7 19.e3 Qe7 20.Qxe7+ Rxe7>?

<I have a truckload of analysis and the post-queen exchange situation in this variation is very complicated and could IMO do with looking at as soon as possible.>

The above is simply to show up in the Chessforum Activity window.

I activated the Hide Chessforum Activity function in my profile. I had to investigate - didn't even know about that one.

*** <Analysis of 17. <Be5> Re8 18. Bxg7+ Kxg7 19. <e3>> User: mckmck

*** <Analysis of 17. <Be5> Re8 18. Bxg7+ Kxg7 19. <Rbc1>> User: dalbertz

*** <Analysis of 17. <Be5> Re8 18. Bxg7+ Kxg7 19. <Nd4> Qe7> User: Red October

*** <Analysis of 17. <Be5> Re8 18. Bxg7+ Kxg7 19. <Na4>> User: Waitaka

*** <Analysis of 17. <Rbc1> Re8> User: zanshin

*** <Analysis of 17. <Be3>> User: Deep Breath

*** <Analysis of 17. <Ne5>> User: capafan

*** <Analysis of 17. <Nxe4> fxe4 18. Ne5> User: not yet a patzer

<After 16...d5 17. Be5 we would have extra forums if we want to subdivide one or more of the 19th move forums.>

I see your point, and a good one it is.

<BTW, I am not yet personally convinced that 19. e3 is our best choice.>

I don't know if anyone besides myself believes that it is the best choice at this point. It was only with reluctance that I actually came to that conclusion myself.

I respect your impartiality and the assiduousness with which you construct and allocate the variations for analysis.