The issue I find with your article is that although the topic is highly entertaining, I couldnt really understand anything but caro and french unsimilarity for the chess style of most players. For example I as a najdorf-gruenfeld player of 2200, who is looking forward to enlarging of a repertoire, what openings should I try? what do people with a repertoir similar to mine usually play except benoni and kings-indian? I would love to know the answers but could not grasp through the graphics.
I may be misunderstanding the graph but I would expect to see things like the French defense and the Dutch to be related since Dutch players often start with e6 to avoid certain variations which allows white to choose to play the French and so the black player must be able to play the French. It looks like the French and Dutch are on opposite sides of the graph.
> bicm?
Perhaps, make a blog just about the background paper, and opening a discussion for others to also read it or at least its figures and legends (and most important paragraphs to understand what goes in each figure, and I mean what is assumption, and what is conclusion, seems important when talking to a mixed chess interest audience).
Specially here, as it seems that the thesis is about population wisdom, as suggesting hypotheses of "relatedness".
what is defined or assumed before the data processing and graph representations (slices and projection). so that we would know for further reasoning the assumptions going in, so we can judge without being ourselves data analysts how to interpret the diagrams. This is not a critic at all. The subject is neither easy from the methodological viewpoint or the chess knowledge proposal. Taking the time to explain enough for autonomous critical reading of the figures might be needed. Good job on finding that paper and proposing it. thanks for that.
There are a few excepts from the paper that might help. I deleted a few posts of mine here.
Perhaps, make a blog just about the background paper, and opening a discussion for others to also read it or at least its figures and legends (and most important paragraphs to understand what goes in each figure, and I mean what is assumption, and what is conclusion, seems important when talking to a mixed chess interest audience).
Specially here, as it seems that the thesis is about population wisdom, as suggesting hypotheses of "relatedness".
what is defined or assumed before the data processing and graph representations (slices and projection). so that we would know for further reasoning the assumptions going in, so we can judge without being ourselves data analysts how to interpret the diagrams. This is not a critic at all. The subject is neither easy from the methodological viewpoint or the chess knowledge proposal. Taking the time to explain enough for autonomous critical reading of the figures might be needed. Good job on finding that paper and proposing it. thanks for that.
There are a few excepts from the paper that might help. I deleted a few posts of mine here.
<Comment deleted by user>
<Comment deleted by user>
Is there any analysis out there which relates openings by their transpositional possibilities?
I'm thinking that Caro-Kann Exchange is related to QGD lines etc.
This might be much more two-dimensional, but would help people understand the themes and strategic implications of different openings.
I'm thinking that Caro-Kann Exchange is related to QGD lines etc.
This might be much more two-dimensional, but would help people understand the themes and strategic implications of different openings.
I don't understand this graph at all, what do red, blue and black colors mean?
Also, how do I find openings "related" to what I play in this graph?
Also, how do I find openings "related" to what I play in this graph?
You did pin point the difficulty of interpretation about what is "relatedness". It might be that the given assumption, is that people will somehow tend to steer the game toward familiar openings, hence defining relatedness that way, and using clustering method based on how often 2 "openings" creating what, a node?
I guess just saying what a node represents, versus the links, might help us follow. What is being validated? The paper figure legend is not telling that much more. You have done the graph, you could tell us more about your experience in that direction.
It may seem going slow.. But spending time at the framework behind the figure, with aim to explain, not just report, might do us a lot of good. My idea of presenting the paper first (not too late) for only the parts that you want to apply here, might be easier done without having your work to report on top. Then we can follow you on how this can be reproduced here. If I got the objective of your experiment right.
I guess just saying what a node represents, versus the links, might help us follow. What is being validated? The paper figure legend is not telling that much more. You have done the graph, you could tell us more about your experience in that direction.
It may seem going slow.. But spending time at the framework behind the figure, with aim to explain, not just report, might do us a lot of good. My idea of presenting the paper first (not too late) for only the parts that you want to apply here, might be easier done without having your work to report on top. Then we can follow you on how this can be reproduced here. If I got the objective of your experiment right.
@Benny-Frandsen said in #8:
> I don't understand this graph at all, what do red, blue and black colors mean?
>
> Also, how do I find openings "related" to what I play in this graph?
In the blog there is opening names as in ECO, and then repertoire, and also opening type. It might be where we might lose some bearing in "reading" the figure.
The paper talks first about a bigger graph called a bi-partite graph, on one side (partite) all the players, and the other all the ECO (last named opening variation the game wiggled through, with 2 colors each, as the opening is attributed to both players).
That mother not display graph would have each ECO name (twice, one for B one for W, but not paired in any way, just 2 of them) on one side being nodes that each player node on the other side would link to.
I guess that the validation AND projection to the figures that we do see is some kind of reduction of that bipartite graph based on counting how many of players keep linking to same openings..
Now to think of it, (and I am guessing, what would I do with those givens so far) there might already be an assumptoin of repertoire-ness, and that might be my difficulty. I could think that validation and projection (meaning we only see the opening ID side, perhaps each node is of the ECO W or ECO B name side. and the linking is some threshold applied on counting pairs of ECO names (W or B included) the more 2 ECO names have players doing both, the more likely they would get a validated link in the projection. That link being some kind of integration of the more refined not show bi-partite graph.
I top here. maybe the op could validate the above for mis-conceptions. Short of us reading the details of the paper.
my priority in reading such papers, is extract the flow of information, what is assumed (it is not bad to assume, it is not good to assume and not being aware of doing so, in that the conclusions might not be the one stated, but that is how science goes, we are not all knowing, we make mistakes, and the many of us, are there to filter.. where we are official peer experts, or just rational being able to consider what is presented with own logic.. I think chessland even newbies should be rich in those capable persons. But to extirpate the knowledge contribution (be it in working hypothesis to further study with other means), not getting lost in the method is a first step. I think blogs are limited medias to fully understand together such new findings or papers. I know that also in chessland, discussion and expertise might not always be propicious to many voices putting their shoulder to the truth seeking wheel. So let's do mistakes and give up on expertise inhibitions. In the spirit of the paper. population wisdom (it does not say mob rule, or hive mind).
> I don't understand this graph at all, what do red, blue and black colors mean?
>
> Also, how do I find openings "related" to what I play in this graph?
In the blog there is opening names as in ECO, and then repertoire, and also opening type. It might be where we might lose some bearing in "reading" the figure.
The paper talks first about a bigger graph called a bi-partite graph, on one side (partite) all the players, and the other all the ECO (last named opening variation the game wiggled through, with 2 colors each, as the opening is attributed to both players).
That mother not display graph would have each ECO name (twice, one for B one for W, but not paired in any way, just 2 of them) on one side being nodes that each player node on the other side would link to.
I guess that the validation AND projection to the figures that we do see is some kind of reduction of that bipartite graph based on counting how many of players keep linking to same openings..
Now to think of it, (and I am guessing, what would I do with those givens so far) there might already be an assumptoin of repertoire-ness, and that might be my difficulty. I could think that validation and projection (meaning we only see the opening ID side, perhaps each node is of the ECO W or ECO B name side. and the linking is some threshold applied on counting pairs of ECO names (W or B included) the more 2 ECO names have players doing both, the more likely they would get a validated link in the projection. That link being some kind of integration of the more refined not show bi-partite graph.
I top here. maybe the op could validate the above for mis-conceptions. Short of us reading the details of the paper.
my priority in reading such papers, is extract the flow of information, what is assumed (it is not bad to assume, it is not good to assume and not being aware of doing so, in that the conclusions might not be the one stated, but that is how science goes, we are not all knowing, we make mistakes, and the many of us, are there to filter.. where we are official peer experts, or just rational being able to consider what is presented with own logic.. I think chessland even newbies should be rich in those capable persons. But to extirpate the knowledge contribution (be it in working hypothesis to further study with other means), not getting lost in the method is a first step. I think blogs are limited medias to fully understand together such new findings or papers. I know that also in chessland, discussion and expertise might not always be propicious to many voices putting their shoulder to the truth seeking wheel. So let's do mistakes and give up on expertise inhibitions. In the spirit of the paper. population wisdom (it does not say mob rule, or hive mind).