365chess.com isn't great at all. Not to brag but I have updated many thousands of player bios (do you know how many FIDE ID cards I've entered?), submitted thousands of games (how many?) and submitted thousands (?) of correction slips (how many?).

I don't know if you remember this but some years ago I complained a number of times about CG's database being a complete mess. The best I could do was either to quit using CG or become a biographer. I chose the latter. I have used many databases to help me and I found a lot of mistakes in all of them. And when I say a lot, I mean *a lot*. Whether it is NicBase, 365chess.com, ChessTempo, chess-db.com etc., they all contain many errors. The best of them is ChessTempo BTW.

As a matter of fact your own CG is becoming the best database, thanks to the volunteers and yourself and last but certainly not least, the Librarian. She's doing a really good job at it, you guys are very lucky to have her around. It will take years before this becomes a really good database but we will get there.

I don't do long bios BTW. English isn't my native language and besides I don't feel up to it; I'm chronically ill and I have my good days and my bad days, but most of them are bad. That's also the reason I've closed my forum, it's too much for me to keep up with that as well.

It's work in progress like so many sites that are organising chess history.

I found 365chess extremely useful as there are many tournament details that are clearly complete and accurate, and which is information that cannot be found elsewhere. The ones that aren't are not exactly mistakes in my opinion, simply incomplete. It's impossible for the PGN of every game to be preserved, and sometimes it's next to impossible to even find out the structural details of many tournaments in sufficient detail. Stuff gets lost.

Like the process that is underway here, many of it is painstaking reconstruction, frequently many decades after the event. Rusbase is another example of this, and it is plain where their data is complete and where it is not, where their cross tables are complete and where they are not.

They are constantly updating their website.

One reason why IMO this site may have one of the better databases is the many loyal volunteers putting in enormous amounts of their time and energy (and money!) into improving it. Not sure how the other websites compare in this respect, but it's a fair guess to assume that they are not as fortunate, if that is the right word, as CG in regard to its many dedicated volunteers such as yourself, and the worthies at the Biographer Bistro, WCC Editing Project and others.

I seriously don't think these other databases are recklessly mistaken let alone willfully so, merely <works in progress> assembling the information they have as a basic foundation and superstructure for future data to complete.

Just about any site on the Internet contains errors, they vary in degree according to the contributors.

This site has a decent record because a lot of work done by so many people. Personally, I am sure I have submitted hundreds of games, <to this site>; and there are people here who have submiited probably tens of thousands of games. And there are people like <The Focus> who have so many game collections ... that I have never been able to view them all!

If 365 is not as good as this one it is only because the contributors have not worked as hard or as long.

No need to be acrimonious about it. Read it, believe it ... but always check it!!! (Maybe only Edward Winter is right 99% of the time when it comes to chess history.)

Can you please update the <GERMAN BUNDESLIGA 2013-2014>? They are playing three rounds this weekend.

<ROUND 7 PGN> -> http://bundesliga.liveschach.net/do...

MVL just broke into the Top 10 by beating McShane in Round 7.

Thanks.

Now about that upgrade to a newer version of <PGN4WEB> ... :-)

Can we please have a forum for the recently completed <2014 MOSCOW OPEN>?

Thanks

----

TWIC DOSSIER -> http://theweekinchess.com/chessnews...

However, I think it is also important to check how the system is working. In other words, are the games selected by this method as notable games actually among the most notable games played?

Specifically, there is a concern among many users that the notable games are out of date, and don't reflect which games are actually notable. There are several possible reasons for this, but I believe the primary one is that <members do not make as much use of game collections anymore>. This may be due to increased use of tournament pages/player pages/etc., but the end result is that games were added to far more collections in the past than they are at present, and so the notable games sections may <never> reflect more recent notable games.

This is especially problematic with the page of Magnus Carlsen, but it's far from unique to him. Here are the current top 10 players in the world with how many of their most "notable game" were played in the last five years (2009-present), or when the most recent one was played if none were:

Magnus Carlsen - 0 (2008)
Levon Aronian - 2
Vladimir Kramnik - 0 (2007)
Veselin Topalov - 0 (2008)
Fabiano Caruana - 7
Alexander Grischuk - 2
Hikaru Nakamura - 2
Viswanathan Anand - 2
Sergey Karjakin - 4
Peter Svidler - 1

Have the current top 10 players really played so few games worthy of attention lately, or is the game selection method just not quite working anymore? Of additional note, not a single game from any of the last three world chess championships is in anyone's game list.

So in my mind, it seems clear this system isn't quite working anymore. While I cannot speak for everyone, I believe this opinion is shared by most, and that the quality of player pages would increase if this were changed.

I propose that the notable games lists of active players be manually curated by those already in charge of the bios. It could be a 50-50 mix - four games chosen by game collection lists and four games chosen by the biographers. Or it could maybe even just be a dynamic weighting of game collections, weighing games added more recently more highly, as they are clearly of greater current interest. There are definitely lots of options available.

I know this was a really long post, so I'll summarize what I'm hoping for a response about in two questions. One, is the chessgames administration generally satisfied with how notable games are selected or does it also see lots of room for improvement? And two, are there any plans (even if only long-term) to change that system in any way?

I agree with almost everything you wrote, and this is a problem which has been brought to our attention last year. The problem never existed a few years ago, it's a new situation that is a consequence of the maturity of Chessgames. Let me explain:

When it comes to players like Fischer or Nimzowitsch, the notable games works amazingly well. The list for Fischer is especially convincing. The first three are what any Fischer aficionado would name off the top of their head as his most famous games, and in that order too.

However, when it comes to players like Carlsen there is a clear bias toward the older games, which is especially disappointing with the younger players who undoubtedly have created their true masterpieces only recently.

The reason I think is not that people don't use game collections, but that it takes time for games to accumulate a presence in collections. The games Carlsen played in 2008 (say) have been around now for 6 years giving plenty of people chances to add them to game collections. Some of the people who added those games to collections don't even come around any more, so those games have extra "votes" which will boost their notability forevermore. His newer important games are being placed in game collections at a furious pace, but even then they have miles to go before they can overtake some older game that have been given hundreds of votes over the years.

Where our views differ is that I am not convinced of the necessity nor soundness of your proposal: <I propose that the notable games lists of active players be manually curated by those already in charge of the bios.>

While that would certainly work, I think it would be better to rework the algorithm so that the active players (especially younger active players) are computed differently. Specifically, what if newer games were given a heavier weighting? For example, instead of just counting up the raw number of collections, we say that 2014 games count for 3 points per collection, 2013 games count for 2.5 points each, and so forth. This would automatically create a bias towards newer games, although it certainly wouldn't eliminate the older games from the list.

Even if that method failed us, there are other tricks. For instance, we can tell when somebody just views a game, so we could keep track of which games are being viewed the most and sprinkle that into the formula. We also could place more weight on game collections of users who have been seen in the past 90-days, to eliminate the problem of defunct accounts perpetually voting for old games. We could even use the number of kibitzes on the game as a metric, although that might tend to give unfair weight to any game that was a live broadcast.

The motivation for solving this problem algorithmically is twofold. First, it's easier. No need to create new CG Editor panels or obscure voting systems, just change a short block of code. Second, and more importantly, we've spent years explaining over and over that the "notable game feature" is *not* determined by the staff. This saves a lot of grief, for if we did it any other way we'd constantly be petitioned to change these lists. After all, no set of 10 games will make everybody happy.

Having said all that, maybe we try the algorithmic approach and still can't get it working to our satisfaction. At that point, the idea of passing it off to editors is always an option.