http://www.365chess.com/players/Man...

(Scroll to bottom, for 1948 games before 1980 dob)

http://www.365chess.com/players/Man...

(Scroll to bottom, for 1948 games before Cuban player's 1980 dob)

And the real nail in the coffin:

http://www.pgnmentor.com/events/Mar...

which also uses Manuel Castillo.

Since I used PGNmentor as my source I would have had to have been fairly sharp to spot the error. By the time I found my two submitted games bounced - I had convinced myself <CG> was wrong.

My mistake was not using the vetted Di Felice crosstable on chesslibrary to double-check each and every player. Also, ARGbase's PGN for the tournament has the correct players - but I only just found that site.

I still feel a little bad - but not as bad as when I first thought I was asleep at the wheel. I was only half-asleep!

< Stem game: Vassily Ivanchuk vs Gata Kamsky (1994) 0-1

Opening: Sicilian, Dragon, Yugoslav Attack, 10.castle long (B78) Player profiles: Ivanchuk, Kamsky

Novelty of stem-game: 14. Bh6>

Novelty ?? Is this information generated at random?

I cannot find the following game among those listed by "find similar games" for the above game:

Mecking vs S Joksic, 1971

6.f3 Bg7 7.Be3 O-O 8.Bc4 Nc6 9.Qd2 Bd7 in Mecking vs S Joksic, 1971

Who mistakenly had <Manuel Castillo> at Mar del Plate (1948) instead of <Mariano Castillo Larenas> :

<365chess.com>, <redhostpawn.com>, http://www.rogerpaige.webspace.virg... and me!

Those who had <Mariano Castillo Larenas>:

<chesstempo.com>, ARGbase, <chesslibrary.com>, and eventually me.

It seems pretty clear to me that the misidentification is most likely due to the listing of the player as <M Castillo>.

Now, I understand that informally, with Spanish culture, <Mariano Castillo Larenas> should be referred to as Mr. Castillo (or Senor Castillo).

But identifying a chess player is not an informal activity. And despite the argument that "short" might mean "shortest possible", I think the opportunity for confusion and misidentification argues for the use of both surnames.

As a case in point: <Manuel Castillo> vs <Mariano Castillo Larenas> for <M Castillo>. Look at the chaos it has rent.

I think <CG> should be on the vanguard in the fight for accuracy, and to avoid confusion should use both surnames in the short form:

<M Castillo Larenas> (or <L Dominguez Perez>).

Of course <Mariano Castillo Larenas> should make prominent mention that his name is often shortened to <M Castillo>, both in contemporary and contemporaneous sources.

<M Castillo Larenas> (or <L Dominguez Perez>).>

I agree that we should be on the vanguard in the fight of accuracy, but I don't believe it necessary follows that the short form should contain both surnames. The short form was invented mostly for cosmetic purposes, not as a means to identify players. The short form should ideally be the name which chess players are most familiar with. As Switchy pointed out, we're not going to name Capablanca "Jose Raul Capablanca y Graupera" anytime soon. Info like that is best left in the biography.

In the crew list I would write Capablanca as:

CAPABLANCA Graupera, Jose Raul

I do think Capablanca isn't the best counter-example to use, because of the universal recognition of Capablanca as singularly referring to the master. (I like the Lenier Dominguez Perez example better.)

Still, reading the wiki Spanish Names article for the 5th time, I do wonder about the <y Graupera> part of his name. Why is the conjunction used?

http://en.wikipedia.org/wiki/Spanis...

I would right him down as:

DOMINGUEZ Perez, Lenier

Are there general rules applicable? You make it sound like its somewhat arbitrary.

The last line of the caption for the Norway Chess Tournament on the main page reads:

"Sergey Karjakin wins, winning 5 out of the last 6 games."

Actually he won 5 points in the last 6 rounds with 4 wins and 2 draws.

"Sergey Karjakin wins, scoring 5 points in the last 6 games."

Also, before OCF mentions it, I think <The last line of the caption for the Norway Chess Tournament on the main page reads> could use half a dozen of comma's.

An important tournament in chess history for several reasons.

Search <Game Collections> for <Buenos Aires 1939>:

Game Collection Search

Yields only a handful of subsets of the Men's Olympiad (e.g. Alekhine's games, or Petrov's games).

Looking at the <Tournament Index> yields only the two follow-up tournaments which were made possible by the Olympiad gathering:

Tournament Index

The entire collection of 928 games is available from the amazing <OlimpBase> site:

http://www.olimpbase.org/1939/1939i...

* * * * *

Now, a couple of questions for <CG>.

1) How many of these games are already on <CG>?

(And how do the administrators determine this number?)

Consider the Final A section:

http://www.olimpbase.org/1939/1939f...

Using my auto-pilot software I estimate <CG> has 169 / 396 games (~ 42%).

The Event/Site tags are, shall we say, rather dispersed.

2) What is the best way to migrate such a massive, but missing, collection of games to <CG>?

The working figure of 42% is likely to be optimistic for the less prestigious sections. That implies that <CG> has hundreds of games in place, but even more hundreds missing (say 400-500 or more).

Can I just submit <OlimpBase>'s zip file?

<zanzibar> This is a problem for which there is no easy answer. We sometimes get members trying to be helpful that do things like submit 50,000 games they extracted from Chessbase.

Had we wanted to duplicate Chessbase's data, we would have used them as an exclusive source many years ago. We decided not to go that route for several reasons, not the least of which is that it seems morally wrong, albeit legal.

Here's the problem with the eager member submitting 50,000 games from CB: suppose we processed the file. It would not be surprising to see a breakdown like this:

• 40,000 games that we needed processed and inserted (yay)

• 6,000 low-quality games which we don't really want in our database at all.

• 3,000 games we already had, with some minor changes (like opening moves, or year) inserted, thus creating duplicated entries.

• 1,000 new games with errors in them, accepted on faith.

So that should explain why we don't accept gigantic dumps of other databases. The 10,000 problematic games are not worth the 40,000 valid games. It would be easier to pore through 50,000 games and insert them one-by-one than it would be the clean up the mess described above.

However, you're not talking about 50,000 games, just 396. All the same problems can arise, but at least on a smaller scale it's a clean-up that we can handle. Still, it takes work.

<Can I just submit <OlimpBase>'s zip file?> I'd prefer you didn't, at least not until we have a better method of tackling the aforementioned problems. The solution may simply be that after the upload occurs, the "elves" pore over the new games to identify problems.

Is there some way to normalize the tournament, by taking the zip file known to be from a specific olympiad, and not only add the missing games but also assign the existing games to the proper tournament ID?

We are not set up to do that right now but it's an interesting idea.

Programmatically it's not a very hard task: once the duplicate is located it would be trivial to change the event field. However we must be careful that we're not tripping over false-positives. Ideally a logfile could be maintained explaining each event ID update so it could be at least scanned by eye for potential errors.