< Earlier Kibitzing · PAGE 186 OF 186 ·
|Feb-12-18|| ||zborris8: I'd like to announce an invitational mini-tournament for ten other GameKnot players from CG, but I don't have any idea how to create one of those chess forums. Do I need to ask a moderator to open one?|
|Feb-12-18|| ||chessgames.com: <zborris8> That’s great. Just go to your Preferences Page and select the option “activate my chessforum”. Then you might want to visit the Chessforum Configuration Page.|
|Feb-13-18|| ||chessgames.com: Another Fischer Random game: Nakamura vs Carlsen, 2018.|
|Feb-13-18|| ||zborris8: Thanks for the help!|
|Feb-17-18|| ||MissScarlett: What are you doing about making the game submission process both more expeditious and transparent? As far as I can tell, only myself, Gypsy and PhonyBenoni are regularly submitting games of historical import. <FSR> just submits his own crappy Internet blitz games.|
|Feb-18-18|| ||thegoodanarchist: <<FSR> just submits his own crappy Internet blitz games.>|
Fifty lashes with a wet noodle for him!
|Feb-18-18|| ||chessgames.com: First of all, in defense of <FSR>, he uploads many excellent GM games, tirelessly filling in holes in our database. Just because you happen to notice his internet games doesn't mean that's all he's doing. Just for example, a recent batch contained:|
E Moskow vs Romanishin, 2006
W Shipman vs Kong Liang Deng, 2004
J Pallesen vs J Aagaard, 2002 (which Aagaard calls "his Mona Lisa")
Oh wait, and what's this ...
F Rhine vs NN, 2007
Anyhow, it was about 20:1 in favor of good games vs. internet shenanigans.
Having said that, your question is a good one, because I fully agree that the current system is untenable.
Here's the basic problem. We can't afford a staff to vet games from every member who wants to help, clean up their mistakes, check for duplicates/authenticity, etc. No other chess database works that way, and I suppose for a good reason. Nobody can afford it.
It pains me to see uploads with typos, because I know that means somebody tried to hand-write a PGN, probably transcribing from an old book or periodical, and put a lot of time into the effort. Sadly it will take even more effort to get it into the database.
Even some of our most helpful game submitters make some technical errors that have to be pored over by eye to prevent disaster. I'd like to give some specific examples without mentioning any names, hopefully to prevent this in the future.
For the special "Source" tag the proper syntax is, of course, like any other tag:
[Source "My System, Aaron Nimzowitsch, 1925, pp45-46."]
[Source: "My System, Aaron Nimzowitsch, 1925, pp45-46."]
and please don't forget a quote as in
[Source "My System, Aaron Nimzowitsch, 1925, pp45-46.]
All of those mistakes make Sargon work overtime.
Another common mistake is seeing a game end with something like "21.Qxh7#" and no 1-0 is provided. To a chess player, writing 1-0 after a mating move may seem redundant, but that 1-0 is absolutely necessary or it's not regarded as complete PGN.
What I'm getting at is this: due to simple human error, it's impossible to put the game importation on some kind of "autopilot", and at the same time it's financially unfeasible to pay people to correct human error (even though we currently are, inasmuch as time and budget allows.) You start to do a cost analysis and you find that the "helpful members" are, in some sense, costing CG quite a lot of money per day. This forced us to revert to a clumsy but, hopefully, temporary policy in which we process new games in a very capricious fashion.
OK end of Part 1. In Part 2, I will actually address your questions.
|Feb-19-18|| ||Tabanus: All good CG, except it should be pp. 45-46 or pp 45-46 (not pp45-46).|
|Feb-19-18|| ||chessgames.com: Yes, pp 45-46, but that's not the type of error I was trying to illustrate.|
Anyhow, now to address your actual question. How can we import quality games into the database without paying a staff of people to fix errors and/or do research with every submission? My only idea is to make it somewhat wikipedia in fashion. It would work kind of like this:
When you go to upload a game, it doesn't wait, it instantly gets inserted into a "holding pen." The uploader is taken to the game(s) submitted and can play through them just like a regular CG game page.
If the uploader provided broken PGN then the game won't play, and they will be able to see that. They will be able to edit their own PGN to fix whatever problem was introduced. (This is exactly the work that Sargon and I do now regularly, perhaps the most time consuming part.) In this way the submitters will learn more about the PGN standard, what works and what doesn't, and should improve their ability to provide working PGN in the future.
While the game is in the holding pen some edits can be made, for example if it misattributed a game to the wrong player, or if the event can be improved from "?" to something sensible, now's the time to do it.
Now that it's in a "holding pen" and has been cleaned up well enough, how do we get the game from there to main database? I think a voting mechanism not unlike the Bistro mechanism is in order, although perhaps it doesn't need to be as rigorous.
One implementation I've envisioned is a "thumbs up / thumbs down" indication that is available only to "trusted members" (CG editors). With enough thumbs-up (as few as two) it gets promoted to the main database. However, if somebody trusted spots an error that demands fixing, they should be able to hold up the process with "thumbs down".
I'm not sure if editing ability should be open only to the person who uploaded the game, or to any CG editor. Either way, once an edit is made to the game, all votes should be wiped out and votings begins anew.
If any editor tries to stymie the process by overusing the thumbs-down option we'll have to consider their motives and decide what to do. Undoubtedly there will be some who demand perfection in every way while others adopt a "better this PGN than nothing at all" attitude. We can hash that out when the situation arises.
This would not typically be employed for modern tournaments (although it could be) or even older tournaments if we can extract quality PGN from trusted sources. But for the types of submissions you're waiting on right now, this could put the power in the hands of the users and turn it into a joint effort.
Once such a system is in place, one of the first things done will be to take all of the countless games in the backlog and upload them to the holding pen. Games that have been waiting for over a year might be resuscitated and inserted.
|Feb-19-18|| ||offramp: <Chessgames.com> the holding pen idea sounds really good. It sounds a bit like what I do all the time at Wikipedia: tiny little amendments that are barely noiceable...|
BTW, I am sure people here are aware that <FSR> has won some very serious awards from Wikipedia, top awards. He writes long, major articles that need little or no editing. That's tough to do! I've seen the games he contributes here, they are either missing historical games or little games of his own that have some element of humour to them.
Anyway, I think I'd like the chance to parse a pgn or two. You've GOT to parse a pgn or two.
|Feb-19-18|| ||MissScarlett: <We can't afford a staff to vet games from every member who wants to help, clean up their mistakes, check for duplicates/authenticity, etc.>|
How many games, on average, are submitted weekly? I thought duplicate games were automatically rejected.
|Feb-19-18|| ||MissScarlett: < <Chessgames.com> the holding pen idea sounds really good>|
Well, it's not a bad idea, I suppose, although it would require its own forum to be of much use, not only to discuss issues regarding submitted games, but, more generally, to request help and share information.
|Feb-19-18|| ||chessgames.com: <How many games, on average, are submitted weekly?> I'm not sure, I just know it's more than we can cope with. If we improved the system I can imagine it would only increase from there.|
<I thought duplicate games were automatically rejected.> They are but the process is imperfect. For instance, suppose we get a submission for an Alekhine-Capablanca game that passes the filters and declared to be unique. We know it's unlikely we'd be missing a game like that, so with scrutiny we find that it's a truncated version of a game we already have with a move sequence different. We weigh the likelihood of the alternate version of the moves being the true ones, but more likely than not, the new submission gets deleted.
If it weren't for that safeguard the database would proliferate with dozens of versions of the Immortal Game etc.
<[It] would require its own forum to be of much use> I was rather thinking that each game submission would get its own forum, a Kibitzer's Corner of just like the "official" games, in which the main topic of discussion should be whether the game is correct and should be included, or if not, why not.
Once a game is "voted in", what happens to that discussion? I rather like the idea of carrying it over to the "official" version of the game, although you could make an argument that the discussion be deleted or at least segregated, as the focus would be of errata and not of the game per se.
A general forum regarding all game submissions in general is also a good idea. A place where people can say "Here's my PGN, why isn't it working?" and discussion of where to obtain the PGNs in the first place.
|Feb-19-18|| ||MissScarlett: How long do you envision these plans taking to implement? Or have you already something in the pipeline?|
In the meantime, you are instructed to give my submissions top priority. I think you'll find them flawless; just wave them through.
|Feb-19-18|| ||zanzibar: <CG> End part 1.|
A huge waste of time would be to encourage users to submit only games which have been loaded into SCID or Chessbase.
That would screen out all the stupid PGN errors you pointed out in your first post, including the missing result info.
What's so hard about insisting trusted submitters self-screen games first?
My goal from years ago was to submit bulk-submit entire tournaments - which would involve ensuring a good xtab as prerequisite.
So much duplicate work and effort could have been avoided if we had actually spent some effort on <Altonia (1872)>.
Water under the bridge.
|Feb-20-18|| ||Tabanus: To save some work: is it not possible to just reject single games with extra colon(s), missing quote(s) and missing game result?|
Something like "Sorry, there's a pgn error. Please try again!"
|Feb-20-18|| ||offramp: I knew that pgn was garbage years ago. Why is there a code for !, and why is that code $1, and not !?|
It reminds me of that pointless utf=8 code for &, which is "& amp ;".
How can the code for the thing that it's code for be in the code of the code?
|Feb-20-18|| ||crawfb5: For those that don't use SCID, ChessBase, or something similar to prepare PGN for submission, how about a utility that lets a member play out the game on a board, asks for information on tags and then produces a legal PGN. It could eliminate the problem of hand-written PGN submissions if it was easy to use and members would actually use it.|
|Feb-20-18|| ||morfishine: Hey <CG> why do you tolerate such gutter miscreants as <john barleycorn>? Besides being nothing but a provocative idiot, I know for a fact he's driven multiple premium members from this site|
IMO terminate this cancer
|Feb-20-18|| ||zanzibar: <crawfb5> makes a very useful suggestion.|
The path of evolution is to make <CG> more and more like SCID in capability. Engine analysis, and PGN entry are good starting points.
I'd like to strongly suggest being able to search on Site/Event tags as well... while I'm here.
Auto-generating xtabs with hyperlinks to the games would also be a nice feature...
|Feb-21-18|| ||FSR: <chessgames.com> On February 11, 2018 I asked you to add this note to Rhine-Shankar if you added it to the database: <15.e5!! Qc6! (only move) 16.exd6 [insert symbol for large advantage to White]>. You said you would. The game is now in the database: F Rhine vs G Shankar, 2017. Could you please add the note? Thanks.|
|Feb-21-18|| ||zanzibar: <<offramp> I knew that pgn was garbage years ago. >|
Well, it may be garbage, but it's the best garbage we got.
* * * * *
A holding pen where you require a user to play over the game to know if the PGN is good or bad!?!
<<CG> ... The uploader is taken to the game(s) submitted and can play through them just like a regular CG game page.
If the uploader provided broken PGN then the game won't play, and they will be able to see that. ...>
My suggestion would be to screen the PGN immediately upon submission, bouncing it back to the user as rejected right there and then - perhaps even pointing out the approximate location the PGN goes off the rails.
SCID does this already.
I don't get it. Why would you even consider requiring a user to play through games by hand?
Sorry, but consider the interzonal I did of ~400 games. You're going to require me to play through 400 games?
(If you want to play games over by hand, then do it like <crawfb5> recommended, and allow PGN input via a playthrough board - one game at a time)
Scaling - not just a good idea, but a requirement.
|Feb-23-18|| ||morfishine: Is it too early to open a 2018 Candidates Tournament page? Games start in 15 days. |
|Feb-23-18|| ||chessgames.com: No, it's not too early. In fact I think it's exactly the right time.|
|Feb-24-18|| ||morfishine: <CG> Great, I can hardly wait!|
< Earlier Kibitzing · PAGE 186 OF 186 ·