Looking at the two snapshots again (Aug 2014 and May 2015) there were 3098 players lacking a FIDE id that could be "easily" found.

(Easily = exact or near-exact match on name for only one FIDE player, i.e. unique 1-to-1)

Of those players, 33 disappeared from the near snapshot (presumably correctly merged).

Of the remaining, about 2/3's look to have been FIDE-identified by the near snapshot, since only 899 EZ's remain.

The EZ identification is not without it's problems, of course - being a simple name lookup errors can be introduced.

Of the ~2k matchups by <CG> about 14 disagree with my FIDE id. Let's assume my lookup is wrong. That means the simple, automatic lookup is wrong about 14/2166 = 14/2166 = 0.65% < 1%

I think I can flag the ones needing inspection even better.

But the early numbers readily demonstrate (imo) the feasibility of automating much of this process, with <CG> itself providing an estimate of the error rate.

Not at all. The whole goal is to amass a huge number of chess games by opening the floodgates. We might even create an API so that blitz games played around the net get automatically pumped in.

<Would I be able to organize the games into tournaments, with tournament pages, standings, etc?>

Eventually, that would be an ideal feature. However it won't be available when the Sandbox is first released. In fact I'm not sure the first demo will even allow you to put them in game collections. However, we definitely want to have kibitzing on Sandbox games.

Maybe eventually the ability to organize Sandbox games in collections/tournaments/etc. will be the premium feature.

<would I be able to control the users that can "see" the games in the sandbox?>

That's an interesting request that hadn't crossed my mind.

If you don't mind me asking, what is the reason you want that? Is it that you don't want your chess opponents to have online access to your secret preparation? Or is it just that you don't want people making fun of you for hanging a rook? Or something I haven't thought of?

What does this mean?

<chessgames.com: <AylerKupp> The "small fee for posting" idea is not in the spirit of the site, and it's not a novel idea either. In the few cases where other sites have attempted that it's ended up in disaster. It would definitely "fix" the problem, at the cost of turning the page into a virtual ghost town.>

I agree with <AylerKupp> and think you're wrong referring to this particular extreme case.

More members would definitely participate and post, including the GM whose name appears on top of that particular page, and perhaps other GMs and rated masters if you rid the site of trolls, especially in the case of previously banned "sock puppets" which are not that hard to discern.

Also would clean up many pages in chessgames.com!

I submitted three back in early February (two by Blackburne and one by Tal) and notice they're still not yet in the database.

It's not the incessant clock times that's the problem (we can strip that out) — the problem is that 95% of the games are just blank. For example:

<I submitted three back in early February (two by Blackburne and one by Tal) and notice they're still not yet in the database.> I'm sorry, I'll slip those in. Yes, there is a ridiculous backlog, and that's one of the motivations of the Sandbox.

The Sandbox sounds like a good idea.

Hanging a rook... I wish it were only that!.. :D

No, the reason I was wondering about the level of privacy was that in my ideal world, our local chess club would have its own 'Chessgames' environment, where we can view and kibitz our own games. If I'm not mistaken, I believe I'm allowed to decide who posts in my personal forum, no? So it's sort of that idea...

Biographer Bistro (kibitz #10932)

And if so, what's the best, most useful, way to transfer the results?

I looked at only the first of your examples, Richard Biolek, which you presume software could automatically assign a FIDE ID # to. But read the bio: <Both Richard Biolek and Richard Biolek Jr. are IMs. The first was born on the 29th of September 1970, the second on the 26th of March, 1990.> In other words, "This player record is invalid and merges two players together. We need to fix that some day."

That rating is supposed to come only from FIDE (modulo ICCF, etc).

The obligation should be to produce the FIDE id before you have the right to cite a rating (imo).

If you have a rating, you should have an id, that's my view of it.

* * * * *

Now there are cases where the FIDE id has been retired, because the player is deceased. So, no FIDE id. But then, <CG> should have a dob.

So, having such a list can also be used to check for players who bio needs updating. Are there any other schematic approaches for updating such information? Consistency with FIDE provides one such.

You'll notice several recent cases I found like that in the last day... in particular of

Tudev Ujtumen

* * * * *

Also, I find it a little odd that <CG> is citing detailed bio info, like dob and rating, without FIDE id.

I have to wonder how this can happen?

I have found several <NIC> hits for players on <CG>, where <NIC> included the FIDE id but <CG> didn't. Perhaps a biographer felt that the FIDE id wasn't important, or wasn't aware it was the responsibility of the biographer to update it?

* * * * *

Also, what about the ~300+ players where I matched both name and dob?

You'll notice that I conveniently, and intentionally, split the population.

I trust there is little if any debate about automatically updating those?

* * * * *

Now, about Biolek:

< <Both Richard Biolek and Richard Biolek Jr. are IMs. The first was born on the 29th of September 1970, the second on the 26th of March, 1990.> In other words, "This player record is invalid and merges two players together. We need to fix that some day.">

Give the FIDE id to the guy born in 1970. That's not "my" problem, that's <CG>'s.

When you find Jr., add an entry with his name, and I'll have a FIDE id for him too! (Of course I saw that during my explorations).

And <About Schmidt>:

http://www.imdb.com/title/tt0257360/

He has a rating. There is only one <Theo Schmidt> in the FIDE database.

<CG> cites a rating and no dod (i.e. Schmidt is still alive). Who else could it be?

Unless it's the ICCF guy here:

https://www.iccf.com/player?id=81015

*and* he didn't play any OTB games.

But then there shouldn't really be any quote of a rating, should there?

Anyways, your argument could be applied to any proposed match. There will still be hundreds in the phonebook. But, surprisingly few on FIDE. And interestingly, the one with the right name also has the right federation/nationality. And the dob looks about right too!

* * * * *

Anyways - how can this work be anything but a tremendous help to whoever adds the FIDE id's to the bios.

Let's hear from them!

Who are they? Who did the bulk of the work over the last year?

How did they find the players in need of updating?

Be creative and inventive...

The list can be edited before being run. Or can't serve as the basis of a loop with editor input for each addition.

What tools does <CG> offer the biographer for this task currently?

* * * * *

I have to recant to some degree.

I think I might have misconstrued your request and actually there is something that can be done right in line with what you are asking. For most of CG's existence, the term "rating" in the database really meant "highest rating" and nothing more. But now we don't just have one rating, we have a "highest rating" and a "FIDE rating".

So while it should be no surprise to see a rating on a player without a FIDE number, it's certainly legitimate to ask: "How can you have what you clearly label <last FIDE rating> when you don't have a FIDE #?"

As always, the answer has to do with some piece of software. There is code that I run at the start of every month to process the new FIDE data. It's a complicated little program that uses funny logic. Although its named "update_ratings", these days it's about far more than ratings. It helps normalize names, it resets "highest ratings", it can set the birth year of a player, and it does some guesswork that you could call a slapdash attempt at artificial intelligence.

The comments to this code are funny and frank. The top of the main loop starts with this explanation:

<
# here's where the rubber meets the road.
# WHICH PLAYER IS THIS, REALLY?
# our #1 test is the fide number.
# if it has a match, we are golden.
# if not, we resort to trickery and witchcraft.>

All the doctors (and there be plenty) are Dr. or Dr.

And what's a Lord without a Lady?

Lady B.

We even have one Elder in the lot: