I've always been confused with the extra layer of tournaments that <CG> introduces - I really think it should be reexamined.

IMO- there should be a <tid> index that maps directly into the PGN triplet structure (i.e. Event/Location/EventDate).

The normal <CG> TI can then map into an overarching collection of <tid>'s, let's call this the <cg_cid>. This would basically map into all the games contained in a tournament book (tb).

That would include the main tournament games, the various playoff stage games, the nullified games, and perhaps some exhibition of offhand games played during the tournament.

The PGN download would contain all the games, but all properly normalized for easy exchange/compatibility.

There are some technical details of course, eg. if the TI collection = one PGN <tid> (the extra layer could be omitted, or cg_cid=tid).

Anyways, the current problems with <Sinquefield (2018)> shows that the entire system needs reexamination and reworking - even if your claim that the machinery is already there is true.

Not lovers , <big pawn>...it is just that decent and honest people have mutual respect for each other... :)

:)

<Tiggler> , thank you , sir :)

< It is possible to submit bulk PGN; and CG will automatically allocate the games in that PGN to one or more tournaments, creating new tournaments for that purpose if necessary; and CG will automatically match player names in the PGN to existing players in the database; and the new tournaments' results tables are computed automatically from the games.>

Are you sure about this?

I worked on bulk submission once-upon-a-time with <Daniel>, trying to set up a <CG> framework with <Altona (1870) (??)> as a prototype.

If all that machinery did exist, <Daniel> didn't let on at the time. We only worked on it for a week, or maybe two at the most. He basically dropped the project when we hit the first snafu. I was disappointed of course, but I didn't want to impose on him given all his other obligations on the site.

We never resumed the project, though in retrospect I perhaps should have tried again. To tell the truth I was hoping for more popular support from the Bistro crowd too.

(All the above deserves a more detailed writeup, I think - maybe a blog post)

<The only thing, in fact, that isn't regularly done is flipping the switch that makes the new automatically created tournaments part of the Tournament Index and open to editing and kibitzing. >

Why not? It appears it would be natural to promote them if everything is working so smoothly and automatically.

Also, of course, ongoing tournaments have more eyeballs on them to point out every problem and blame the administrators for them :)

--

The problem with creating historical tournaments by uploading bulk PGN is that quite often the Chessgames database already has some games from the tournament you want to upload, but with different Event or Date details that would cause them not to be recognized as part of the same tournament. There was a tournament in Altona in 1872; I don't know if that's the one you worked on, but I will use it as an example because it illustrates the problem beautifully.

Anderssen vs K Pitschel, 1872 was played in Altona, but the Event is given simply as "GER". A few games like K Pitschel vs E Schallopp, 1872 have the event as "Altona"... and a few like E Schallopp vs C T Goering, 1872 have "NSB-03.Kongress". The "Altona" ones might be recognized by the machinery as being from the same tournament as your uploads (supposing you used simply "Altona" as the tournament name), but the others would not.

It wouldn't help if your PGN included those games and had consistent data in the Event field, because your copies of those games would be rejected as duplicates of the existing games. (In theory it could help, and I wouldn't be surprised if that was something Daniel worked on, but as things stand now it would not help.)

You'd need to manually locate every Altona game already in the database and make sure the Event details were ultra-consistent (which only an editor could do). That would probably be enough in the case of Altona. But if the tournament name were something like "Moscow", you'd need to <also> make sure that other games played in Moscow around the same time would not be wrongly assigned to your tournament; the machinery isn't particularly smart when it comes to things like that (this is another thing Daniel would have needed to work on). So that would be a lot of extra work for an editor if he couldn't use a more specific name for his new tournament. Again, a non-editor couldn't do it at all.

--

If you look at random TIDs it should be very apparent why most of them aren't opened to kibitzing or made part of the Tournament Index.

For example, here's a sample of 10 consecutive TIDs (from 71652 to 71661)...

Donner Memorial Open (1994)
Jugoslavija (1981)
Munich GER (1923)
Troll Masters (1989)
21th Open (2004)
Texas Western Open (1997)
Asturias-chT40 Primera A (2000)
Trio 685 email (1999)
TG..0.12792 (1998)
Class N.056 IECG (1998)

One of them is empty and completely obsolete. The other nine all consist of a single game. There's negative value to opening kibitzing on them - if people noticed them at all, they'd probably say things like "what the @#$% is this?", "where are all the other games" or comments about the actual game that would properly belong on the game page. Or they might drown poor <Sargon> in well-intentioned correction slips... (That's to say nothing of what would happen if the games were reassigned to a different tournament.)

And even then, only if the admin decides to switch from one <Live Game> to another one, say if the original selected game ends early.

Let's return to this after some foundational discussion - but the take home lesson is to ignore the live game PGN, and just update the <CG> tournament round by round from the official tournament sources.

I do have to say that I think <Daniel> complicated the process by not building safeguards into the game upload stage for tournaments. Stashing the times in the PGN Event field is a hack I don't think I'll ever truly understand.

OK, let's take a moment to see what I'm talking about...

At the moment the 6th Sinquefield Cup (2018) has R8 completed, and all player have 8 games.

But right at this moment, the <Carlsen--Nakamura> game still points to the Live Game:

Carlsen vs Nakamura, 2018

(gid = 1926411)

But clicking it goes to the <Live Game> page:

Live Broadcast Page

That's how <Daniel> set it up. The tournament, while ongoing, would have one extra game every day, i.e. the upcoming <Live Game>.

That breaks the symmetry, i.e. all games in the round should be equal. I should be able to play over <Carlsen--Nakamura> just like all the other tournament games right at this moment!

The <Live Game> shouldn't really show up on the tournament page in the game list - because it's a special feature.

One could just put a link on the tournament page advertising the upcoming live game (similar to the one on the <CG> front page). But don't make it subject to special handling. All one really needs to do is to graft the comment stream onto the game after loading the round into the tournament.

<<Switch> ...

Most "tournaments" in the database are hidden and not open to kibitzing; the majority are partial tournaments with 95% of the games either missing or allocated to another hidden tournament with slightly different details, or pseudo-tournaments automatically created from games that simply happen to have similar date and event data, or even "tournaments" of single games; but there are also some real tournaments that just aren't part of the Index, like CG.com Masters vs. Machines Invitational (2007).>

OK, I then wonder what exactly is the functional point of the "hidden layer"?

I can understand wanting a grading system of sorts - both for quality of competition in the tournament itself, e.g. category:

https://en.wikipedia.org/wiki/Chess...

and also for completeness in <CG>'s reconstruction of it.

(E.g. Large scale Swiss tournaments could deliberately have many missing games on <CG>, like say, all games involving a player rated less than 2000 are not included. Of course, I would like stubs for such games when the opponent was a prize winner - so <CG> could say all top-10 players tournament record is complete. But we're getting into technical asides...)

The idea would be to have a main Tournament page, with gold-standard tournaments, while allowing users who wish the ability to see/explore/search all the tournaments.

The logic is, if the hidden layer is useful to an admin, it should be useful to a <CG> user - thus, no hidden layer should remain hidden.

<CG> definitely needs to normalize the PGN headers for all these games - be they in regular or hidden tournaments. I've looked at enough of these games' PGN headers to realize there's rarely, and I do mean rarely, any useful information in the unnormalized headers (once the normalized data is extracted).

If there was any concern, one could easily put the errant header info into new PGN tags, or embed them in a PGN comment.

The last example you gave:

CG.com Masters vs. Machines Invitational (2007)

showcases one of the problems <CG> has with this hidden layer structure.

Consistency.

Why should the 2007 version of the tournament be hidden, whereas the 2011 version isn't?

Tournament Index

CG.com Masters - Machines Invitational (2011)

(Let's restrict ourselves to brand new tournaments for the first phrase of implementation.)

<
Editors (and other vetted contributors) should be able to submit an entire tournament PGN (or exhibition or match PGN) which is automatically processed and promoted to a level-0 <CG> tournament.
>

Why?

Well, first of all, to free the <CG> administrator from the task, which really doesn't need supervision in most cases (** - see below).

This has the obvious advantage of offloading the work onto the shoulders of the editors (and software), but also allows <CG> to operate in the event of the absence of any administers.

Of course, there should safeguards built into the process, i.e. an audit trail to allow rollbacks, and quotas to prevent DOS and other bad behavior.

The software will have automatically produced a level-0 minimum intro with xtab, leaderboard and description based on PGN headers. Then, once accepted as level-0, the tournament could then be promoted by a group vote in the Bistro. Editors are free to elaborate before (or after) promotion.

Once promoted, the tournament shows up on the <CG> front page, as one might expect.

* * * * *

** Details - there is much to discuss if automatic bulk submission were to be done in general, but we can restrict ourselves to the obvious simple case to begin - i.e. a brand new tournament of well-known players. Thus, the player match-up should be straight-forward (e.g. it could be done via FIDE ID if provided, and by canonical name if not), and not involve any chances to the <CG> player db. And since every game is brand new, there is no need to check for duplicates.

In all seriousness. You people are cowards. I don't care if <<I>> had to do a month or two in the ceegee slammer, I just think there's people who deserve death row and you're twiddling your thumbs, same mistake Daniel made.

Nuno's got proof I killed Daniel Freeman, he said it <on this page!>. Check your email and I <<guarantee>> you Keyser Soze aka Nuno Lisboa has sent you over 50 emails - half the amount he sent Daniel 6 weeks before his passing - compared to my <none>! Check whose blown the whistle on over 50 posts <since Daniel passed away>, check BP for a pulse, check out England reaching a semi, check out how Beatlesque Oasis are, choose life, check if you put enough petrol in the motor, I don't care.. But check out the people who've acted disgracefully since Daniel died FIRST!?

PS. The criminal who's wanted by the English and Brazilian police for some malicious communications act over a 7 year period will post after me, so will <Tiny bishop>. Nuno will boil with hatred, that's what stalkers do - ask one of the most respected members of this site, she knows :)

Who do you want to kill now, with your endless stress and bs?

First you call <Daniel> a coward for not taking orders from you, and now you call the new admins cowards in the same way.

<Daniel> eventually told you to sod off and that's basically what the community is telling you now.

And in case anyone is unclear of exactl what kind of valuable contribution <winston> offers the chess community he "cares so much about", let me offer this tidbit from yesterday:

<Winston: Do you think that anal sex between a man and a woman is wrong? Two people in a relationship, I don't see the problem if it's consensual, do you? I've kicked a few back doors in in my time but I prefer the front door, what do you think homie? >

And the follow up:

<Winston: So you prefer the easy pink to the tight brown? #SnookerAnalogy. As an American Asian I thought you'd approve of a back door entrance as long as Trump greases you up first? "Build that wall" Ouch! "Space Force all the way?" BWAHAHAHAHA It tickles! "lock her up?". Don't stop, that's the spot Don Don 😂 #SIMPLETONS>

Bobsterman3000 chessforum (kibitz #232)

Nuff said.

Or this one <big pawn>, just today.

<Dr Winston OBoogie: <Keep up your good work, BP,but you'll never get your soul back.> Never a truer statement posted. BP is the incarnation of a liberal Beelzebub, I'll bet he's got tattoos of Hitler over his horn removal scars. He definitely got raped by a family member, all this "manly man" and "girly men" crap means he was either abused as a child, bullied at school or he's gay af. Quite possibly the 2nd weirdest individual I've ever encountered, <Koma Toze> aka <Two sugars Soze> being the weirdest. And BP I'm taking <Koma Toze> back, you stole my b--ch Lol>

?!? ON a chess site..

No comments.>

"I will make you miserable" - Soze, K.

"I will troll your arse" - Soze, K.

"Like you said its unsure he even knows how to play chess" FREEMAN, D on Soze, K.

<Winston: Do you think that anal sex between a man and a woman is wrong? Two people in a relationship, I don't see the problem if it's consensual, do you? I've kicked a few back doors in in my time but I prefer the front door, what do you think homie? >

<Winston: So you prefer the easy pink to the tight brown? #SnookerAnalogy. As an American Asian I thought you'd approve of a back door entrance as long as Trump greases you up first? ">

Poor <sargon> >

Indeed, <keyser>, indeed.

He wants <Sargon> to check out our posts, but didn't he forget about something?

<Winston: Do you think that anal sex between a man and a woman is wrong? Two people in a relationship, I don't see the problem if it's consensual, do you? I've kicked a few back doors in in my time but I prefer the front door, what do you think homie? >

<Winston: So you prefer the easy pink to the tight brown? #SnookerAnalogy. As an American Asian I thought you'd approve of a back door entrance as long as Trump greases you up first? ">

lols

OK, I could post an self-contained article, but let's work off the 2nd half of <Switch>'s 2nd post:

<The problem with creating historical tournaments by uploading bulk PGN is that quite often the Chessgames database already has some games from the tournament you want to upload, but with different Event or Date details that would cause them not to be recognized as part of the same tournament. There was a tournament in Altona in 1872; I don't know if that's the one you worked on, but I will use it as an example because it illustrates the problem beautifully.>

OK, this probably was the tournament we worked on (I'll check later - it's not really important).

Let me begin by agreeing - it's a good illustration either way. The issue is how to allow historians to submit en masse, and what <CG> should do.

The simplest approach is to allow duplicate games and just "stand out of the way". In other words, allow the historian to just add all the new games, properly normalized, and create the historical tournament, complete as submitted.

It's easy for <CG> to do, as no matching is needed - just depend on the historian to get it right and promote it.

Are there any problems with this approach?

Yes, there will be duplicate games, which will mess up the stats. Plus the new games might be missing valuable comments already present in the old game discussions. Lastly, old game collections will be pointing to old duplicates, which are likely to be deleted someday.

Still, as a zeroth step it's easy enough to do, and dovetails perfectly with my proposed new tournament submission machinery.

I'd really like to talk about the following, but will return to the topic later. It is technical - but not too hard. And almost nobody is utilizing the manual machinery we currently have - save for small matches (and the ones <Tab> decides to work on).

Let me just focus on one small issue:

<Anderssen vs K Pitschel, 1872 was played in Altona, but the Event is given simply as "GER". A few games like K Pitschel vs E Schallopp, 1872 have the event as "Altona"... and a few like E Schallopp vs C T Goering, 1872 have "NSB-03.Kongress". The "Altona" ones might be recognized by the machinery as being from the same tournament as your uploads (supposing you used simply "Altona" as the tournament name), but the others would not.

It wouldn't help if your PGN included those games and had consistent data in the Event field, because your copies of those games would be rejected as duplicates of the existing games. (In theory it could help, and I wouldn't be surprised if that was something Daniel worked on, but as things stand now it would not help.)

You'd need to manually locate every Altona game already in the database and make sure the Event details were ultra-consistent (which only an editor could do) ...>

Which only an editor could do, *currently*.

Let's consider a case where all of a tournament's games are already in the <CG> db, but completely unnormalized. Further, let's suppose that a historian submits a proper tournament, completely normalized.

Let's also suppose that the tournament bracket dates are part of the submission, and that games in the DB can be excluded if not matched. Let's further suppose the names can be matched too.

(Obviously each assumption is a technical discussion of its own)

That leaves the Event/Location fields and the movelists. Generally, the Event/Location fields on <CG> are practically useless - as <Switch> pointed out.

But we have a fairly good match via the White/Black/Date fields, and if the movelists also match, well, that's a pretty clear situation.

So the submission can be used to upload the <CG> games with the Date/EventDate/Round numbers from the submission, and use the Event/Location fields as well.

In other words, it's not too difficult a job to normalize and promote the tournament from the historian's submission with just a little bit of code.

There might be a game or two (or three...) which have different movelists, and so a new game gets introduced into the DB. But the tournament "trusts" the historian's version, as it should.

Generally, such cases need the addition of historians anyways, and even then there might now be a clear answer (e.g. <CT> keeps duplicates, with "near-duplicate" links in its DB).