Latest Updates:
Page Index Toggle Pages: [1] 2 
Topic Tools
Hot Topic (More than 10 Replies) Pgn sucks. (Read 8693 times)
Jay
Full Member
***
Offline



Posts: 212
Location: USA
Joined: 04/18/09
Gender: Male
Re: Pgn sucks.
Reply #19 - 02/03/11 at 14:48:04
Post Tools
Papageno wrote on 01/28/11 at 21:24:47:
You are right, that the author(s) of PGN didn't have game analysis in mind. That's true. They probably did not have chess positions, chess problems and Chess960 in mind either, admittedly. But otherwise PGN helped a lot for information exchance so far.


I know I have run into the PGN 960 issue.  At gameknot I tried to annotate a 960 game and realized it was not possible there.  I personally think chess 960 is essential to growth as a chess player.
Markovich wrote on 01/28/11 at 22:08:38:
Jay wrote on 01/28/11 at 16:45:20:
Markovich wrote on 01/27/11 at 19:06:06:
Beyond the year, does it really matter to anyone upon which date a certain game was played between Steinitz and Chigorin?
  Removing the date would certainly make finding a certain game between players that play often against each other in a year very difficult.


I'm not saying that it should be forbidden; I'm saying that it shouldn't be mandatory.  Site-Year combination has been enough hisorically, but if someone wants site-event-date, that's perfectly OK.  But not all contexts need an exact date, or both a site and an event (which is very rare in historic game citations) so why require this?

Further as I've been saying, not every piece of chess notation is a game.

I actually think that year and date are redundant.  If year is important, and the format is largely for electronic exchange, why should year have to be entered at all?  Seems to me that Year is derived from date rather easily.  Drop Year for date, but leave year in as a derived field that never needs to be entered.

Papageno wrote on 01/29/11 at 17:47:00:
It seems to me that everybody is having different expections to what PGN (or a new format) should be able to do. Guest comment by "daily dirt" Mig Greengard:
Quote:
(...) Nice of them to mention the draw offer, so allow me to refresh one of my habitual rants about how we discard all this information. Both players must record draw offers on their scoresheets, but this essential information vanishes into the database formats we all all rely on at home and in online viewers. The clock times, critical to understanding the flow and the key moments of the game in the minds of the players, and which are recorded automatically by the relay software every top tournament uses, also vanish. Moronic. It's like the script of a play without the stage direction. Organizers need to export the scores of top events with the clock times and draw offers and the clearinghouses like TWIC and ChessBase have to preserve them. If PGN is awkward for this, improve it or use something else. Make it so! (...)

Source: http://www.chessninja.com/dailydirt/2011/01/nakamura-leads-tata-with-two-rounds-...

I actually agree with this.  It may not make much difference to me why some top player played a poor move, but it certainly would make some difference to chess writers whose books I am likely to purchase.

MNb wrote on 01/30/11 at 02:59:36:
When playing through games I hardly care about draw offers. Games that have clock times each move I find unreadable.
So I am with Markovich.

See this is partly a valid issue.  Those of us who work with computers are perfectly fine with having the clock times in the available information set.  When displayed on a computer chess reader, the clock time is hardly intrusive, but in print that is another matter!  Perhaps that ought to be a shorthand version of any XML/PGN standard for print publications?  Of course if this happens, the real standards may fall into disuse.  I think there should be a universal XML standard with DRM built in.  I would love to be able to put Chessbase on an iPad and play through a book or annotation.  Currently it is hard to find ebook versions of books except for Everyman which many criticise.  This is because it is incredibly easy to share those works via torrents or other peer networks.  Add DRM, encryption, and portability?  Well, we would probably see chess stars and quality chess enter the digital revolution!  Let's face it.  Many tournament players would love Fritz/Chessbase + ebooks on an iPad.
  
Back to top
 
IP Logged
 
MNb
God Member
*****
Offline


Rudolf Spielmann forever

Posts: 10486
Location: Moengo
Joined: 01/05/04
Gender: Male
Re: Pgn sucks.
Reply #18 - 01/30/11 at 02:59:36
Post Tools
When playing through games I hardly care about draw offers. Games that have clock times each move I find unreadable.
So I am with Markovich.
  

The book had the effect good books usually have: it made the stupids more stupid, the intelligent more intelligent and the other thousands of readers remained unchanged.
GC Lichtenberg
Back to top
 
IP Logged
 
Markovich
God Member
*****
Offline



Posts: 6099
Location: Columbus, Ohio
Joined: 09/17/04
Re: Pgn sucks.
Reply #17 - 01/30/11 at 02:39:10
Post Tools
Papageno wrote on 01/29/11 at 17:47:00:
It seems to me that everybody is having different expections to what PGN (or a new format) should be able to do. Guest comment by "daily dirt" Mig Greengard:
Quote:
(...) Nice of them to mention the draw offer, so allow me to refresh one of my habitual rants about how we discard all this information. Both players must record draw offers on their scoresheets, but this essential information vanishes into the database formats we all all rely on at home and in online viewers. The clock times, critical to understanding the flow and the key moments of the game in the minds of the players, and which are recorded automatically by the relay software every top tournament uses, also vanish. Moronic. It's like the script of a play without the stage direction. Organizers need to export the scores of top events with the clock times and draw offers and the clearinghouses like TWIC and ChessBase have to preserve them. If PGN is awkward for this, improve it or use something else. Make it so! (...)

Source: http://www.chessninja.com/dailydirt/2011/01/nakamura-leads-tata-with-two-rounds-...


I disagree with Mig.  We never had clock times before and chess never languished for it.  This info is merely nice to know, imho.  It's information overload nowadays!  Info is so cheap that its marginal value falls, via market forces, to zero.  Actually for many of us, falls to less than zero (it's like spam, you see?).  The hoi polloi might want to know what round it was, what exact day of June it was, what hour, minute and second sunrise was, what the weather was, the latitude and longitude, and the time zone; I don't!!  And having all that crap in the game score gets in my way.

What exactly is anyone's interest in whether someone was short of time at some point?  He had to play a chess move, didn't he?  And he played one.

Or else let's post info about whether the guy's girlfriend was waiting for him in the bar, fuming because he hadn't finished his game yet!  The one is just as relevant as the other. 

But I take all that back!  What we should really do is put heart monitors on the players and record their pulse rates at each point in the score.  Now that would be truly interesting to the interpretation of the score.
  

The Great Oz has spoken!
Back to top
 
IP Logged
 
Papageno
Senior Member
****
Offline


FM

Posts: 299
Location: Germany
Joined: 06/12/08
Gender: Male
Re: Pgn sucks.
Reply #16 - 01/29/11 at 17:47:00
Post Tools
It seems to me that everybody is having different expections to what PGN (or a new format) should be able to do. Guest comment by "daily dirt" Mig Greengard:
Quote:
(...) Nice of them to mention the draw offer, so allow me to refresh one of my habitual rants about how we discard all this information. Both players must record draw offers on their scoresheets, but this essential information vanishes into the database formats we all all rely on at home and in online viewers. The clock times, critical to understanding the flow and the key moments of the game in the minds of the players, and which are recorded automatically by the relay software every top tournament uses, also vanish. Moronic. It's like the script of a play without the stage direction. Organizers need to export the scores of top events with the clock times and draw offers and the clearinghouses like TWIC and ChessBase have to preserve them. If PGN is awkward for this, improve it or use something else. Make it so! (...)

Source: http://www.chessninja.com/dailydirt/2011/01/nakamura-leads-tata-with-two-rounds-...
  
Back to top
 
IP Logged
 
Markovich
God Member
*****
Offline



Posts: 6099
Location: Columbus, Ohio
Joined: 09/17/04
Re: Pgn sucks.
Reply #15 - 01/28/11 at 22:08:38
Post Tools
Jay wrote on 01/28/11 at 16:45:20:
Markovich wrote on 01/27/11 at 19:06:06:
Beyond the year, does it really matter to anyone upon which date a certain game was played between Steinitz and Chigorin?
  Removing the date would certainly make finding a certain game between players that play often against each other in a year very difficult.


I'm not saying that it should be forbidden; I'm saying that it shouldn't be mandatory.  Site-Year combination has been enough hisorically, but if someone wants site-event-date, that's perfectly OK.  But not all contexts need an exact date, or both a site and an event (which is very rare in historic game citations) so why require this?

Further as I've been saying, not every piece of chess notation is a game.
  

The Great Oz has spoken!
Back to top
 
IP Logged
 
Papageno
Senior Member
****
Offline


FM

Posts: 299
Location: Germany
Joined: 06/12/08
Gender: Male
Re: Pgn sucks.
Reply #14 - 01/28/11 at 21:24:47
Post Tools
Just for your information, here is the link to an XML project by the chess problem community.
http://problem-xml.sourceforge.net/
It's XML for chess problems. But we can easily think of something similar for chess games, analysis, opening repertoires or solving tactics.

This XML looks quite useful for the information exchange between different software tools, but certainly does not intend to immediately make it readable for the human eye. (Especially if you follow the link "documentation by example" at the end of the article I linked.) If I get their overall intentions right, human readibility is only one "representation" (or say: possible output) of an XML definition and requires a tool for it.


When it comes to some points I mentioned in my reply #4 yesterday (different ratings are put into the [WhiteElo ] tag and name spelling problems due to missing unique identifier for players), I have to correct myself a bit. Both of these things are not really points where we have to critisize the PGN standard. The criticism must rather address those programs used for generating PGN output, that they do not handle more and additional header fields. To be more specific: ChessBase cannot handle multiple ratings of one player. And when I get PGN output from ICC, they put their incredible high server blitz ratings right into the same [WhiteElo ] tag where other tournaments store FIDE Elo etc. So the problem is not PGN but the institutions that write rubbish into it.
(Same goes for the way the TWIC magazine writes the players names with abbreviated Christian names)


Markovich, in your original post you raised the Q, why is it mandatory that "Event" be supplied? Well, just because this is the first tag which a parser needs to find to identify the start of the next game in a PGN file with multiple games. So nothing dramatic I'd say. Any set of other starting/closing tags like <game> and </game> would be the modern solution (and this takes just about the same space).

You are right, that the author(s) of PGN didn't have game analysis in mind. That's true. They probably did not have chess positions, chess problems and Chess960 in mind either, admittedly. But otherwise PGN helped a lot for information exchance so far.


For the future, we might want to improve on the information exchange standards, say with XML. But we also have to do some more work on a better data quality that is to be collected then.
  
Back to top
 
IP Logged
 
Devilman
YaBB Newbies
*
Offline



Posts: 28
Joined: 07/09/09
Gender: Male
Re: Pgn sucks.
Reply #13 - 01/28/11 at 18:14:27
Post Tools
i'm waiting for an example because i don't understand your dense formatting
  
Back to top
 
IP Logged
 
Jay
Full Member
***
Offline



Posts: 212
Location: USA
Joined: 04/18/09
Gender: Male
Re: Pgn sucks.
Reply #12 - 01/28/11 at 17:21:51
Post Tools
Jay wrote on 01/28/11 at 16:58:45:
Papageno wrote on 01/27/11 at 22:15:06:
Other things you mention can be simply added to PGN notation. For instance if you wanted to add different ratings to PGN, why not extend the format of header
  [WhiteElo "2065"]
by header fields like
  [WhiteICCFElo "2066"]
  [WhiteUSCFElo "2067"]
  [White_ICC_Elo "2068"]
etc.?

Keys are a definite improvement, but they can end up being arbitraty.

Okay, you all got me thinking.  Seems to me that an html or xml inspired chess notation should be possible.  Imagine the following:
<game type = "chess" language = "English" elo = "2350" ELO.Organization = "FIDE" /* etc */ />


It might not work out to be much more readible with <game><game/> tags, but any standard reader should be cheap and easy to produce.  There would be freeware versions.

I think any computer/database friendly annotation standard is not going to be user-friendly when written without formatting which a magazine on chess might do if they are presenting a massive amount of games.
  
Back to top
 
IP Logged
 
Jay
Full Member
***
Offline



Posts: 212
Location: USA
Joined: 04/18/09
Gender: Male
Re: Pgn sucks.
Reply #11 - 01/28/11 at 16:58:45
Post Tools
Papageno wrote on 01/27/11 at 22:15:06:
Other things you mention can be simply added to PGN notation. For instance if you wanted to add different ratings to PGN, why not extend the format of header
  [WhiteElo "2065"]
by header fields like
  [WhiteICCFElo "2066"]
  [WhiteUSCFElo "2067"]
  [White_ICC_Elo "2068"]
etc.?

Keys are a definite improvement, but they can end up being arbitraty.

Okay, you all got me thinking.  Seems to me that an html or xml inspired chess notation should be possible.  Imagine the following:
<game type = "chess" language = "English" elo = "2350" ELO.Organization = "FIDE" /* etc */ />
  
Back to top
 
IP Logged
 
Jay
Full Member
***
Offline



Posts: 212
Location: USA
Joined: 04/18/09
Gender: Male
Re: Pgn sucks.
Reply #10 - 01/28/11 at 16:45:20
Post Tools
Markovich wrote on 01/27/11 at 19:06:06:
Beyond the year, does it really matter to anyone upon which date a certain game was played between Steinitz and Chigorin?
  Removing the date would certainly make finding a certain game between players that play often against each other in a year very difficult.
  
Back to top
 
IP Logged
 
Willempie
God Member
*****
Offline


I love ChessPublishing
.com!

Posts: 4312
Location: Holland
Joined: 01/07/05
Re: Pgn sucks.
Reply #9 - 01/28/11 at 15:59:10
Post Tools
Papageno wrote on 01/27/11 at 22:15:06:
Markovich, I think you have raised some quite interesting points to which I fully agree. E.g. it is a serious shortcoming that PGN or any chess databases don't identify players by unique keys. That is elementary knowledge for anyone in computer sciences: databases, relational or not, work best with unique keys. Then, thinks like spelling of names and transcription and representation of foreign languages' names would not be such an issue and could easily be displayed for the user in their favourite language.

Yep, but that is due to the huge amount of different game providers. Just check what kind of fun there is around with names like Korchnoi and Petrosian. Still something like a FIDE number or Dutch chess federation number (as I dont trust the FIDE to arrange anything properly except for bribing) would be nice: Ie NL123456 for dutch players.

Still an XML format would have been preferable as PGN is very similar to standards that have evolved in the same way (eg early HTML). If you can make that work for chemistry, chess shouldnt be a problem.
  

If nothing else works, a total pig-headed unwillingness to look facts in the face will see us through.
Back to top
 
IP Logged
 
dfan
God Member
*****
Offline


"When you see a bad move,
look for a better one"

Posts: 725
Location: Boston
Joined: 10/04/05
Re: Pgn sucks.
Reply #8 - 01/28/11 at 15:24:40
Post Tools
Markovich wrote on 01/28/11 at 01:43:40:
Well what do you do when you read Informant or NIC Yearbook?  Have you ever tried to transcribe data from these sources, or even just to follow along with them and a chessboard?  What you find in these sources is that the torturous complexity of PGN "recursive annotation variations" is replicated literally.  The upshot is that it takes forever to trace through a given variation.  Your eyeballs freeze up looking for where the current variation, once departed from, rejoins itself.  Please answer my point here.

Were Informants noticeably easier to read before PGN was invented? (Serious question; I have no idea.)
  
Back to top
 
IP Logged
 
Markovich
God Member
*****
Offline



Posts: 6099
Location: Columbus, Ohio
Joined: 09/17/04
Re: Pgn sucks.
Reply #7 - 01/28/11 at 01:43:40
Post Tools
Paddy wrote on 01/27/11 at 22:56:49:
Goodness gracious Markovich, you actually try to READ pgn?

I thought it was just designed as a "lingua franca" for exchanging game data between different programmes, (e.g. Chessbase to Chess Assistant).

If I want to produce readable output, e.g. from Chessbase, I choose a more congenial format from the available options, such as RTF or HTML.


Well what do you do when you read Informant or NIC Yearbook?  Have you ever tried to transcribe data from these sources, or even just to follow along with them and a chessboard?  What you find in these sources is that the torturous complexity of PGN "recursive annotation variations" is replicated literally.  The upshot is that it takes forever to trace through a given variation.  Your eyeballs freeze up looking for where the current variation, once departed from, rejoins itself.  Please answer my point here.

Your post is rich with irony, in that you think it ridiculous that anyone should be able to make sense of text data!  I understand that that's what you've grown used to, but I say you shouldn't need CB or CA or whatever PGN reader you have, to be able to read a text game score, plus comments, or someone's tree of analyzed variations, and make sense of it (given a chess board, of course).  If that were true, we might as well have our raw data in binary form.  That truly is not subject to immediate inspection.

As Eric Raymond said in The Art of Unix Programming:

Text streams are a valuable universal format because they're easy for human beings to read, write, and edit without specialized tools. These formats are (or can be designed to be) transparent.  (http://www.faqs.org/docs/artu/)

Without specialized tools, I emphasize, particularly proprietary ones that cost any given user hundreds of dollars.  But with chess it goes beyond that, because publishers mimic the raw data structure in their publicized text (or perhaps more correctly, the authors of specialized tools like CB find it convenient to issue reports stuctured like the raw data is, and publishers borrow these).   And so if the raw data isn't transparent, the published text isn't either.  Witness the NIC Yearbooks.  And I would go so far as to say that pgn, combined with the almost universal dependence upon ChessBase, is what explains the modern prevalence of the "examplary game" method of presenting chess theory.  A much more naturual and thoroughgoing presentation of theory is in tree structure a la Bilguer, MCO and ECO, but pgn isn't adapted to that!  Thanks to this, our whole modern way of thinking about chess data is around a set of collected games.  But that isn't what theory is.

Raymond's concern is the robustness of the structure of information, and that's what I'm talking about.

The only excuse for a binary data format or, by extension, an unintelligible text format, is that it speeds processing.  But it's possible to organize textual chess data in ways other than pgn that are just as fast to process, if not faster, and which are more or less transparent to direct inspection.  I would be happy to demonstrate this if someone would post a pgn example of sufficient complexity.  What we have today is unintelligible data thanks to nothing more than the stupidity of pgn's author, and everyone else's need for a any standard format at all, ready to hand.
  

The Great Oz has spoken!
Back to top
 
IP Logged
 
Paddy
God Member
*****
Offline


The truth will out!

Posts: 915
Location: Manchester
Joined: 01/10/03
Gender: Male
Re: Pgn sucks.
Reply #6 - 01/27/11 at 22:56:49
Post Tools
Goodness gracious Markovich, you actually try to READ pgn?

I thought it was just designed as a "lingua franca" for exchanging game data between different programmes, (e.g. Chessbase to Chess Assistant).

If I want to produce readable output, e.g. from Chessbase, I choose a more congenial format from the available options, such as RTF or HTML.
  
Back to top
 
IP Logged
 
Markovich
God Member
*****
Offline



Posts: 6099
Location: Columbus, Ohio
Joined: 09/17/04
Re: Pgn sucks.
Reply #5 - 01/27/11 at 22:19:13
Post Tools
motörhead wrote on 01/27/11 at 22:05:43:
You are absolutely right, Markovich.
I often run analysis with White: "Weiß" and Black: "Schwarz" or else. No possibility to call it an analysis. You have to give it names.
And really an annoyance: It is not possible to structure comments. If you have a lot of subvariations then your are lost in space. A1ac1a2b...
Without any paragraph, the coloumns are packed and you will never, never find a way throuh it.
The only possibility is to paiste and copy it right away and without any loss oft time give a structure to it. Do not miss a day or a week you will not get an entrance. And do not print pgn directly to store it on paper. It will be useless unless you are a mole to dig through...


Well I just think that raw data in text form should be more or less intelligible.  That's why text data is so nice.  Unintelligible text data might as well be in binary.  Chess data could be intelligible if not for pgn.  Right now we see almost raw pgn in Informant and in the NIC Yearbooks, and it's tortuous to read, or to transcribe into one's files.  This makes these journals quite difficult to use in paper form, in my view.

If someone will post a pgn of a very a densely commented game in pgn format, like Kasparov's comments on Piket - Kasparov, Tilburg 1989 (the famous KID game) or perhaps something even more complicated, I'll post in reply a reorganized text file to demonstrate what I mean.  For my own amusement, I've invented something I call "structured chess notation."  I'm well aware that there's no use trying to get it accepted, but I thought I'd demonstrate it here anyway.

@Papageno:  If someone sends me a dense pgn example, you'll see what I mean.  I agree that text commentary is better, but the option of good structured commentary, such as you invented, should be available.  For one thing it's international (though it might not be too nice for Serbs or Russians to have to read Latin letters); for another, it's capable of being processed by a machine.  But you're right, it's really a separate issue.  I mentioned it here because the idiotic "numeric glyphs" specified in pgn are virtually useless.

As for Elo, I think if we had something like [Aegis: ICCF] then it would be understood that both ratings were according to the ICCF rating tables.  Also then, [WhiteID: 511471] would be understood as an ICCF ID, and could be checked against the ICCF lists.  Likewise FIDE, RSF, etc., etc.. 
  

The Great Oz has spoken!
Back to top
 
IP Logged
 
Page Index Toggle Pages: [1] 2 
Topic Tools
Bookmarks: del.icio.us Digg Facebook Google Google+ Linked in reddit StumbleUpon Twitter Yahoo