chessgames.com
Members · Prefs · Collections · Openings · Endgames · Sacrifices · History · Search Kibitzing · Kibitzer's Café · Chessforums · Tournament Index · Players · Kibitzing

search collections   |   new tournaments   |   vote on collections   |   editor help
Kibitzer's Cafe
< Earlier Kibitzing  · PAGE 360 OF 360 ·  Later Kibitzing>
Jan-29-15
Premium Chessgames Member
  hoodrobin: <Phony Benoni: ...sometimes you'll be wanting him to shut up already and come to the point...> You perfectly caught the point, <PB>.
Jan-29-15
Premium Chessgames Member
  Tabanus: The page http://lv.wikipedia.org/wiki/Tenis_... has Tenis playing in Kemeri, but the link goes (wrongly by 365Chess) to Emilis. At thttp://lv.wikipedia.org/wiki/1938._..., the Kemeri 1939 table shows Tenis.
Jan-29-15
Premium Chessgames Member
  zanzibar: Somebody once promised me that promoting a collection to a tournament would properly normalize it.

I hope I wasn't in error to assume that this meant that all the tournaments in the <Tournament Index> were properly normalized.

But, oh so sadly, perhaps this belief is mistaken.

<Looking at the actual PGN for all of <CG>'s tournaments shows a wretched mess, at least for the Event/Site headers>

This isn't hyperbole - it really is all over the map.

Like illegal PGN, I thought this wasn't supposed to happen.

Jan-29-15
Premium Chessgames Member
  Tabanus: <z> Event and Site will be what we write in these fields when nominating. I haven't seen any deviation from that yet. In a few cases, upon request/correction slip, some games have got another Site after voted in.
Jan-29-15
Premium Chessgames Member
  zanzibar: <Tab> Have you looked at the actual PGN from the games in the tournaments?

Maybe the inconsistencies predate the voting series, I don't know. Is there a list of tournaments which shows when they were voted in and by who?

Let me take just one tournament:

Tal - Botvinnik World Championship Match (1960)

This is a <creme de la creme> WCC match, fully vetted and sparkling <CG> tournament right? Yet look at the various Event/Site headings the actual games contain:

<
Date Players Games Site:Event

1960.03.15 2 1 Match, Moscow (1): Russia

1960.03.17 2 1 Ch World , Moscow (Russia) (2): match

1960.03.19 2 1 Match, Moscow (3): Russia

1960.03.22 2 1 Match, Moscow (4): Russia

1960.03.24 2 1 Match, Moscow (5): Russia

1960.03.26 2 1 Match, Moscow (6): Russia

1960.03.29 2 1 Match, Moscow (7): Russia

1960.03.31 2 1 Match, Moscow (8): Russia

1960.04.02 2 1 Match, Moscow: Russia

1960.04.05 2 1 Match, Moscow (10): Russia

1960.04.07 2 1 Match, Moscow (11): Russia

1960.04.12 2 1 Match, Moscow (12): Russia

1960.04.14 2 1 Match, Moscow (13): Russia

1960.04.19 2 1 Match, Moscow (16): Russia

1960.04.21 2 2 Match, Moscow (15): Russia

1960.04.23 2 1 Match, Moscow (14): Russia

1960.04.28 2 1 18: Moscow-Wch

1960.05.03 2 1 Match, Moscow (19): Russia

1960.05.05 2 1 Match, Moscow (20): Russia

1960.05.07 2 1 Match, Moscow (21): Russia >

The match of 21 games maps into 20 "tournaments" (i.e. unique Event/Site headers)!

<Russia> is fairly consistent as the Event name (which is wrong of course), except for one game whose Event tag is <Moscow-Wch> (just for a change of pace).

The main reason we have 20 different tournaments is because the Site tag changes with each Round, as if their are different locations within Moscow, or should I say, within <Match, Moscow>.

Actually, the Round number is copied into the Site field, which is both redundant, and wrong.

And it's not even consistent, as one game omits it. And two games have Round 15 in the Site tag.

And then there's the issue of the Date correspondence with the Round, which is out of synch maybe? That is, if the Round number in the Site header can be trusted.

Still with me?

That's just for one tournament, we have 1800+ more to go.

Jan-29-15
Premium Chessgames Member
  Tabanus: <z> That match predates the voting series. I think we have the same problem with a few hundred matches and tournaments "only". For the WC matches, I believe these are planned republished (ask WCCEP about the process) from new and improved game collections such as (the unfinished) Game Collection: WCC: Botvinnik-Tal 1960.
Jan-29-15
Premium Chessgames Member
  crawfb5: <z> All the world championship matches predate the voting process because they were already established as pages by CG prior to development of the current system. The ongoing process of reworking them is a rewrite of the introductions only, and does not involve any changes to the game PGN tags.

As for non-WC events, the most likely sources of site/event inconsistencies would be events with multiple venues because these could not be automatically normalized upon being voted in.

Volunteer editor powers have been slowly increased; the next likely expansion would be the ability to correct event and site tags. Until then we'll have to rely on correction slips or direct appeals to admins.

Jan-29-15
Premium Chessgames Member
  zanzibar: First, is there a record of tournaments which were voted in?

Basically, I'd like to know what tournaments are supposed to meet the <CG> "gold-standard".

Secondly, I think the process, like any good process, should be documented and publicly available. I don't know where it is.

Jan-29-15
Premium Chessgames Member
  zanzibar: <Tab> writes <<z> Event and Site will be what we write in these fields when nominating. I haven't seen any deviation from that yet. In a few cases, upon request/correction slip, some games have got another Site after voted in.>

This isn't acceptable, if I understand you correctly.

Once a game is incorporated into a tournament the Event/Site tags cannot be changed unless the entire tournament tags are changed.

And to be explicit, the full qualification for a tournament are a unique combination of Event/Site/EventDate tags (did I miss any?).

Again, what tournaments are fully vetted?

And also - after a tournament is voted in, does anybody download the PGN and doublecheck the published xtab (in the intro) against that reconstructed by a database using the PGN?

That should be the last step in a checklist before the tournament gets fully promoted to "gold-standard" status.

Jan-29-15
Premium Chessgames Member
  zanzibar: <crawfb5:> writes <<z> All the world championship matches predate the voting process because they were already established as pages by CG prior to development of the current system. The ongoing process of reworking them is a rewrite of the introductions only, and does not involve any changes to the game PGN tags.>

Thanks <crawfb5> (Larry?) for that info. It's quite a different situation from what I had been imagining.

The WCC PGN is actually in the worst shape of all - ouch!

This situation needs to be remedied.

<As for non-WC events, the most likely sources of site/event inconsistencies would be events with multiple venues because these could not be automatically normalized upon being voted in.>

I'm sorry, but my early explorations suggest this is not the case. The headers are just simply inconsistent.

Of course the multiple venues do pose a problem, but one that was anticipated in the early PGN standard documentation.

The PGN standard recommended using a universal compound location for the Site location in this situation. So if a tournament took place in cities A, B, C the tag should look like:

<[Site "A / B / C"]>

This is somewhat akin to the <EventDate> and <Date> labels where <Site> is analogous to <EventDate>.

I'm a little unsure what the tag analog is for <Date>, i.e. how to specify the actual specific location of a game in the PGN.

<Volunteer editor powers have been slowly increased; the next likely expansion would be the ability to correct event and site tags. Until then we'll have to rely on correction slips or direct appeals to admins.>

OK, I agree we need more help.

But the idea is that the games have already been grouped into tournament clumps. <CG> should (must) process the game to normalize the headers.

This shouldn't require too much work, and is mostly mechanical, and also automatic. Plus the Bistro is always there to help if any bugaboos arise.

The goal is 0th order consistency - as a necessary starting point.

Jan-29-15
Premium Chessgames Member
  zanzibar: <Tab><WCC> - Is there a reason the rewrite of

Game Collection: WCC: Botvinnik-Tal 1960

goes to 24 games? Do you know something I don't know?

(He asked with a wry smile and a wink)

Jan-29-15
Premium Chessgames Member
  zanzibar: The idea is to download a tournament PGN into SCID and be able to produce a valid xtab that matches the xtab in the intro.

Consider this tournament: Showalter - Albin (1894)

SCID breaks it down into two tournaments:

<Date Players Games Site:Event

1894.12.26 2 2 New York, NY USA: Showalter - Albin

1894.09.24 2 23 New York, NY USA: Showalter - Albin>

This seems to be a bug in SCID, from what I can tell.

That is disconcerting. Does anybody know if ChessBase has the equivalent of SCID's <Tournament Finder> window?

Jan-29-15
Premium Chessgames Member
  zanzibar: <RE: Tournament Finder>

I guess ChessBase's full blown version has it, but ChessBase Light doesn't.

Is that correct?

Jan-29-15
Premium Chessgames Member
  zanzibar: <RE: SCID Tournament Finder Window>

I was wrong, SCID doesn't use the EventDate as it really should (provided it's there). Instead it uses a 3-month window:

<The Tournament Finder lets you find tournaments in the current database. It scans all the database games and collates data about the tournaments found. Note that two games are considered to be in the same tournament if they have the same Event tag, same Site tag and were played within three months of each other.>

http://scid.sourceforge.net/help/Tm...

So, the Showalter--Albin match stretched over by a couple of days, so SCID just up and decides to break it into two tournaments.

It's funny though, both tournaments show the same 25-game xtab.

This stuff is driving me to drink!

Jan-29-15
Premium Chessgames Member
  crawfb5: <zanzibar: <Tab><WCC> - Is there a reason the rewrite of

Game Collection: WCC: Botvinnik-Tal 1960

goes to 24 games? Do you know something I don't know?

(He asked with a wry smile and a wink)>

Tal - Botvinnik World Championship Match (1960) hasn't been rewritten yet. Game Collection: WCC: Botvinnik-Tal 1960 is what Jess has been calling a "mirror," which is a place to park material that might be part of an eventual rewrite. Why she left space for 24 games is anybody's guess, unless she had a template she used for other WC matches as well. In any event, it's completely unofficial.

Jan-29-15
Premium Chessgames Member
  crawfb5: <zanzibar: <crawfb5:> writes <<z> All the world championship matches predate the voting process because they were already established as pages by CG prior to development of the current system. The ongoing process of reworking them is a rewrite of the introductions only, and does not involve any changes to the game PGN tags.>

Thanks <crawfb5> (Larry?) for that info. It's quite a different situation from what I had been imagining.

The WCC PGN is actually in the worst shape of all - ouch!>

Of course. The games come in from different sources and weren't normalized at the time, so here we are. Be glad you weren't here for the mixed bag of New York games misidentified as <New Orleans Wch> we were chasing down a few years ago. :-)

As things currently sit, only an admin can fix them. In fact, even to change the intros Jess sends me the plain text, I add HTML tags and put it in the format expected by the WC pages, then send it to Daniel, who has to do the actual replacement.

<This situation needs to be remedied.>

Agreed.

<<As for non-WC events, the most likely sources of site/event inconsistencies would be events with multiple venues because these could not be automatically normalized upon being voted in.>

I'm sorry, but my early explorations suggest this is not the case. The headers are just simply inconsistent.>

It would help to see an example or two of a non-WC event with this problem. Not that an editor could currently fix it, but it might give a hint how it happened.

<Of course the multiple venues do pose a problem, but one that was anticipated in the early PGN standard documentation.

The PGN standard recommended using a universal compound location for the Site location in this situation. So if a tournament took place in cities A, B, C the tag should look like:

<[Site "A / B / C"]>

This is somewhat akin to the <EventDate> and <Date> labels where <Site> is analogous to <EventDate>.>

Here's where you and I part ways. In a multi-venue event, I'd like to know which games were played at what venue. [Site "A / B / C"] isn't going to tell me that.

Jan-30-15
Premium Chessgames Member
  zanzibar: <crawfb5> a couple of comments.

(Skip to below the * * * to get right to the point)

<It would help to see an example or two of a non-WC event with this problem. Not that an editor could currently fix it, but it might give a hint how it happened.>

Fair enough. I'll try to post an example or two in a moment.

You'll have to give me some allowance, as this is the first time I've actually looked closely at tournament data (vs generic game data when building a collection).

<<This is somewhat akin to the <EventDate> and <Date> labels where <Site> is analogous to <EventDate>.>

Here's where you and I part ways. In a multi-venue event, I'd like to know which games were played at what venue. [Site "A / B / C"] isn't going to tell me that.>

This is an important point. The goal is to have the data in such good shape that an accurate xtab can be reproduced for each tournament, just from the PGN.

At the moment I'll yield on the stub games (temporarily!), lets just assume the <CG> tournaments are complete (i.e. have all the games).

Still we live in a PGN world, certain concessions must be made for the "stupidity" of computers. We don't have complete freedom to use the PGN tags as we might like.

For example, ask yourself how one can group generic PGN games into a tournament cluster, like a program must. What algorithm?

You can't just use the Event tag, as many locations around the globe will have tournaments with the same name, e.g. <Alekhine Memorial> or <Capablanca Memorial> etc.

So the location must be used too, e.g. the Site tag.

And the date must be included as well, since some locations have annual versions of the same tournament.

(Well not always every year, but the idea is the same. Myself, I also like to include the version of the tournament in the Event name, but not everybody is as formal.)

OK, we have to make a concession here, as the date of a tournament needs to be the same over all the games. Yet we also want the date of each game. How to do this?

Consider, if only the PGN <Date> tag is used. We already saw SCID's tournament finder used an algorithm with an arbitrary window of 3 months. Which failed to recognized the Showwalter--Albin games all belonged to the same tournament.

The PGN standard recognized the weakness of that approach, and so introduced the <EventDate> tag to compliment the game <Date> tag.

Unfortunately, they didn't standard the analogous problem with a Tournament's Site vs the actual game locations.

* * * * *

So then, what to do about the Site tag for tournaments that have multi-venuews?

For example the recent Wijk aan Zee tournament took place in three locations. What should be the Site tag for the recent Wijk aan Zee tournament? It was held in three different locations for Group A (Masters); R5 in the Rotterdam and R10 in the Hague.

http://www.tatasteelchess.com/tourn...

I don't think we actually disagree, as you indicated. We both think each game's PGN should somehow indicate the actual location of play.

The difference then is how to do so?

That's where we might disagree. I would like to follow the PGN standard, and list the Site as:

<Wijk aan Zee / Rotterdam / The Hague>

Of course I'd accept <Wijk aan Zee> if pressed. Just so long as the <Site> tag is the same for all the tournament games, so as to not break SCID (or ChessBase, etc.).

You, instead, might like the <Site> tag to reflect the actual city the game was played in. Which is quite reasonable and intuitive. Unfortunately, it's also problematic for software.

(As an aside, I'll mention that if a game was started in one location, adjourned and resumed in a different location, you would also have to consider using compound locations.)

We agree we should record the actual location of where a game was played. So we have two reasonable approaches. The question is how to reconcile them?

Since the PGN standard failed to accommodate us, and since extra PGN tags are hidden, I think a good compromise is to use a compound location for the SITE tag, and to record the actual location of the game in a comment before the first move.

That way, the location data is readily viewed. If the location data is missing the location would be the first location in the compound Site field. That might be a good convention and save a few bytes while we're at it.

This is already too long a post, so I'll stop here.

Jan-30-15
Premium Chessgames Member
  Tabanus: <zanzibar: First, is there a record of tournaments which were voted in?> This all we have I believe: New Tournaments (link on bottom of TI page).

<Once a game is incorporated into a tournament the Event/Site tags cannot be changed unless the entire tournament tags are changed.> I don't know the technical consequences, but in for example Portisch - Spassky Candidates Quarterfinal (1980), Site was changed (from Mexico City to Xalapa) for games 2 & 3 after voting.

Jan-30-15
Premium Chessgames Member
  zanzibar: So what is the actual status of the recent added tournament

<Tata Steel (2015)>?

Let's look at four versions of the tournament - from <Tata Steel>, <ChessBase>, <TWIC> and from <CG>:

<Tata Steel>
http://www.tatasteelchess.com/tourn...

http://www.tatasteelchess.com/year/... (masters+candidates round2, ..., round13)

<ChessBase>
http://en.chessbase.com/post/carlse...

http://en.chessbase.com/Portals/4/f... (masters only)

<TWIC>
http://www.theweekinchess.com/chess...

http://theweekinchess.com/assets/fi... (masters only)

<CG>
Tata Steel (2015)

http://www.chessgames.com/perl/pgnd... (masters only)

I would say the PGN in the best shape is TWIC's.

But the only PGN which mentions either The Hague or Rotterdam is from Tata Steel itself. And they use the <Site> tag, which causes ChessBase Light to drop rounds 5 and 10 from the xtab (for the masters).

Jan-30-15
Premium Chessgames Member
  crawfb5: <zanzibar: The goal is to have the data in such good shape that an accurate xtab can be reproduced for each tournament, just from the PGN.

Still we live in a PGN world, certain concessions must be made for the "stupidity" of computers. We don't have complete freedom to use the PGN tags as we might like.

For example, ask yourself how one can group generic PGN games into a tournament cluster, like a program must. What algorithm?

You can't just use the Event tag, as many locations around the globe will have tournaments with the same name, e.g. <Alekhine Memorial> or <Capablanca Memorial> etc.

So the location must be used too, e.g. the Site tag.>

I'm starting to see why you want to do the compound tags. As an aside, it's probably rare that two different tournaments held in the same year would have the same event name. I can't think of any, but who knows, there might be one or two odd cases. Otherwise, the SITE tag doesn't add any resolving power.

<And the date must be included as well, since some locations have annual versions of the same tournament.

OK, we have to make a concession here, as the date of a tournament needs to be the same over all the games. Yet we also want the date of each game. How to do this?

Consider, if only the PGN <Date> tag is used. We already saw SCID's tournament finder used an algorithm with an arbitrary window of 3 months. Which failed to recognized the Showwalter--Albin games all belonged to the same tournament.

The PGN standard recognized the weakness of that approach, and so introduced the <EventDate> tag to compliment the game <Date> tag.>

As long as you're using the EVENT DATE tag and have it set to the start of the tournament, that should work. Otherwise you have to handle all those events that split years, starting in December and ending in January. Hastings is the most famous, but there are plenty of others.

<I don't think we actually disagree, as you indicated. We both think each game's PGN should somehow indicate the actual location of play.

The difference then is how to do so?>

Yes, we agree on general goals, but differ on preferred solutions.

<That's where we might disagree. I would like to follow the PGN standard, and list the Site as:

<Wijk aan Zee / Rotterdam / The Hague>

Of course I'd accept <Wijk aan Zee> if pressed. Just so long as the <Site> tag is the same for all the tournament games, so as to not break SCID (or ChessBase, etc.).

You, instead, might like the <Site> tag to reflect the actual city the game was played in. Which is quite reasonable and intuitive. Unfortunately, it's also problematic for software.>

I'd say the compound tag is a little problematic for the wetware. :-)

<(As an aside, I'll mention that if a game was started in one location, adjourned and resumed in a different location, you would also have to consider using compound locations.)>

I wouldn't. Just as <game date> refers to <start> and not <completion> date (remember adjournments were common back in the day), I would stick with start location for specific games.

<We agree we should record the actual location of where a game was played. So we have two reasonable approaches. The question is how to reconcile them?

Since the PGN standard failed to accommodate us, and since extra PGN tags are hidden, I think a good compromise is to use a compound location for the SITE tag, and to record the actual location of the game in a comment before the first move.>

Not my preferred solution, but I could live with it as a compromise.

<That way, the location data is readily viewed. If the location data is missing the location would be the first location in the compound Site field. That might be a good convention and save a few bytes while we're at it.>

I think that is asking a bit too much from the average user. Better to spend a bit of memory in the interest of clarity.

Jan-30-15
Premium Chessgames Member
  zanzibar: <<crawfb5>

<I'd say the compound tag is a little problematic for the wetware.

:-)>>

As emissary for the machines, never let it be said that a peaceful solution with humans was not initially sought.

http://www.bbc.com/news/technology-...

http://www.washingtonpost.com/blogs...

http://www.smithsonianmag.com/innov...

Jan-30-15
Premium Chessgames Member
  zanzibar: <Tab <z> That match predates the voting series. I think we have the same problem with a few hundred matches and tournaments "only".>

No, this isn't really true. Every new tournament that is pre-approved without vote by the <CG> staff falls into this category.

<Tata Steel (2015)> is one of these. From the link you gave:

New Tournaments

I estimate that of the 200 new tournaments listed, 120 are from the <CG> staff.

The staff picks are almost always guaranteed to be the most bountiful, as these are the tournaments which are "hot off the press", i.e. the most recent, just finished, tournament. The very ones attracting the most attention.

So, a big part of what I'm trying to do is to help suggest/establish a process which is both convenient, practical, and accurate.

We don't have such a process in place, afaik, for the new tournaments. We need it.

Aside - the <New Tournaments> page is great, and a handy resource. Thanks for the link. But it's insufficient wrt process. We need a more authoritative page which records the dates the tournaments were added, the number of games and players, and the vote (and perhaps who voted) if applicable.

I would also like a link to an automatic xtab, computer generated, for each tournament to be on this summary page.

Using the xtab to ensure correctness of a tournament is invaluable, and I believe a manual inspection of such an xtab should be the last step in promoting a collection into a tournament.

This raises the issue of the automatic promotion of live or recent tournaments to <CG> tournaments. There are some issues there, ready for discussion some other time.

Jan-30-15
Premium Chessgames Member
  zanzibar: Thanks <gauer> for that link. It's perhaps the best formatted write-up I've seen on the PGN-standard on the web.

I haven't time to fully read all of it (got about 1/2 through). Am I mistaken to say it looks essentially identical to this version:

http://www6.chessclub.com/help/PGN-...

which is what I believe I mostly have been using (via Wiki/Google Search).

I've bookmarked your link - as I said, it's certainly the most readable.

http://www.saremba.de/chessgml/stan...

PS- Did you delete your post (again)?

Jan-30-15
Premium Chessgames Member
  zanzibar: <TOURNAMENTS ADDED BY MEMBER NOMINATION>

OK, putting aside <Tata Steel (2015)> for a moment, let's assume the member nominated tournaments are <CG> "gold-standard".

Case in point: Rio de Janeiro Interzonal (1979)

http://pakchess.org/wp-content/uplo...

Certainly the introduction has the fingerprints of <Tab> all over it, and seems to be the best writeup on the net.

(Though I would suggest adding a link to this set of photographs:

http://commons.wikimedia.org/wiki/C...

Also, it's too bad we can't somehow denote, visually, the qualifying players in the xtab)

* * * * *

OK, with those preliminaries aside, let's look at the data, eh, I mean the xtab:

https://zanchess.wordpress.com/2015...

There are two missing games from SCID's xtab:

Timman vs Velimirovic, 1979

and

Sax vs Ivkov, 1979

Why are these games missing from the xtab?

Because the PGN headers are not normalized:

<
[Event "8, Rio de Janeiro izt"]
[Site "8, Rio de Janeiro izt"]

[Event "12"]
[Site "Rio de Janeiro izt"&rback;
>

Ironically, this may be because of the corrections <Tab> and I found for these two games (slips send in by <Tab>).

Remember, this is the very best data on <CG>.

So, the bottom line is that nothing is safe, and the <CG> normalization process is out of control (because it has to be 99.999% reliable, not just 98%).

A mild rebuke, if I may...

Which is the more important to concentrate on? ...

Cleaning up the PGN on <CG>, or the game commentary?

Jan-30-15
Premium Chessgames Member
  zanzibar: My working notes on all the tournament games with illegal PGN on <CG>:

https://zanchess.wordpress.com/2015...

There should be enough information for <Sargon> or anyone else to correct the games.

Jump to page #   (enter # from 1 to 360)
< Earlier Kibitzing  · PAGE 360 OF 360 ·  Later Kibitzing>

Advertise on Chessgames.com
NOTE: You need to pick a username and password to post a reply. Getting your account takes less than a minute, totally anonymous, and 100% free--plus, it entitles you to features otherwise unavailable. Pick your username now and join the chessgames community!
If you already have an account, you should login now.
Please observe our posting guidelines:
  1. No obscene, racist, sexist, or profane language.
  2. No spamming, advertising, or duplicating posts.
  3. No personal attacks against other users.
  4. Nothing in violation of United States law.
Blow the Whistle See something that violates our rules? Blow the whistle and inform an administrator.


Messages posted by Chessgames members do not necessarily represent the views of Chessgames.com, its employees, or sponsors.


home | about | login | logout | F.A.Q. | your profile | preferences | Premium Membership | Kibitzer's Café | Biographer's Bistro | new kibitzing | chessforums | Tournament Index | Player Directory | World Chess Championships | Opening Explorer | Guess the Move | Game Collections | ChessBookie Game | Chessgames Challenge | Store | privacy notice | advertising | contact us
Copyright 2001-2014, Chessgames Services LLC
Web design & database development by 20/20 Technologies