What is the value in filing out the checkboxes for Other Section Info? Isn’t it obvious from the dates already entered and the time control whether it is a One Day event and/or Blitz Rules Event? And what is the consequence if those are not accurately filled out? Thank you!
I don’t remember why those checkboxes were added. The Blitz Rules box may have predated the Blitz ratings system. It is possible that there have been events that weren’t Blitz rated that used Blitz rules, though why that’s useful in a crosstable is unclear. The ‘One Day Event’ box might have had something to with some kind of promotional campaign.
We’re in the early stages of defining what information we want the next version of the tournament system to capture, these fields might not make the cut.
1 Like
What kinds of fields are you thinking about adding?
NOTHING is finalized yet, and there hasn’t even been a full discussion among staff yet, so this is just a coredump of things that come to mind at 1AM.
I think we need to clean up how we handle whether or not an event had a TLA or other publicity through US Chess. (Is it still possible to have a TLA that is online only, not print?) The ratings system does not have access to the current TLA system, that should change.
There has been some discussion about changing what TD categories we use, or possibly making more of them open-ended text fields. TDCC will need to review these categories, since this affects TD requirements for advancing to higher certification levels.
I think I posted something about the ‘participant coding’ recently (scholastic, non-scholastic, etc) those need some rethinking. Those fields were added because we always get a lot of questions about how many ‘scholastic’ or ‘adult’ events are held, but defining those terms is vague at best.
The ‘organizer’ field is confusing, it was originally intended only for FIDE events.
Ideally, there should be a new upload format that maps to more of the fields in the online editing form if not all of them.
Additional results codes may be defined, to help differentiate between results as paired and results for ratings purposes.
Fixing the double RR option so it reports the result of each game, including color, rather than just a total score (0,0.5,1,1.5,2) would help with total wins in game statistics, since a 1.0 score could be a W-L, a L-W or a D-D. (This requires cooperation from pairing program authors.)
We need to decide if tiebreaks should be included. (Re-computing tiebreaks is messy and time-consuming, I probably lean towards just having a set of fields for TB1, TB2, etc. (Is 4 sufficient?) The section header could include information on what each TB field is in that section.
Some provision for uploading games in PGN format would be desirable. Personally, I’d like to see a complete prize list (cash and non-cash) but that issue always raises hackles.
Yes, Among other differences, online only TLAs are free.
I’d love to see some of these changes. Remove duplicate entry (like time controls and chief TD), remove/simplify superfluous/ambiguous data collection. I wouldn’t get rid of the online TLA either - it would just be helpful if the systems talked to each other. I should be able to upload results for the tournament I told you about through my TLA. Remove the wretched hidden checkbox!
From reading up on this forum it seems like these changes have always been just 1 year out of reach. They are the chess equivalent of the fully autonomous vehicle. I hope it gets prioritized.
I’m not sure why you think time controls and chief TD information is duplicated.
There is a chief TD for the tournament and each section can have a Section Chief TD. For small events this is often the same person, but for large events it is not.
Similarly, each section needs time control information, because it may differ from the time control of other sections.
And I don’t know what you mean about a ‘hidden checkbox’.
The dbf files have TDs at the event and section level, each section has a time control. Why do I have to re enter the event chief TD and time control when submitting the rating report?
When you release the event there is this hidden checkbox to verify you reviewed the information. It is some of the worst UX in addition to the numerous unnecessary recalculate totals.
1 Like
We still support the original upload file format, which doesn’t have time control information.
The fields in the upload form for chief TD and time control are not mandatory, they are just overrides for the information in the upload formats, which is helpful in case those fields are either not present or need changing. (In other words, these fields were added to save steps for TDs.)
What you call a ‘hidden checkbox’ is, I assume, the checkbox for the Chief TD or submitting TD to certify that rules were followed. This is a carryover from the paper format that we used before 2005, whether or not it is still necessary is a policy issue and thus is above my pay grade.
If by ‘recalculate totals’ you mean the fees computation, that’s necessary because there are fees, such as FIDE fees or Grand Prix fees, that can’t be computed until the event is validated.
These are also needed because the chief/submitting TD may not be the one who pays the fees, that could be done by the sponsoring affiliate. (The chief TD is responsible for the validity of the ratings data, hence the certification box, the affiliate is responsible for the fees.) That’s also the point at which information about money prizes above the current money prize floor threshold is currently entered. (If we had full data on prize monies, that would probably no longer be needed.)
Once the TD has certified that the event was conducted in compliance with the rules and handed the event off to the affiliate for payment, the pairing/results data is locked, any changes to those fields requires handing the event back to the chief/submitting TD.
I think you’ve been dealing mainly with small events, but the form needs to be able to handle everything from a quad or two player match to an event with 1000 or more players and several dozen sections.
One of the things that we’ve discussed with Leago is giving the chief/submitting TD (and possibly the affiliate) the ability to make post-rating corrections to the event, to deal with situations where after the event is rated it is discovered that a member ID is wrong or a result is mis-reported Currently these have to be sent to the ratings staff. The staff would still have to review those corrections and approve them before the event would be changed.
Please. Even if 99% of the data points are correctly entered, that means an error every hundred games…and in a 4 round scholastic with 100 players…
…I think we average a correction request every 2-3 months.
Even with a crosstable-like format for updating results, (which staff have access to) there are still several gotchas to deal with. Since each player has the result relative to that player in a separate field, updating the result for player 1 means updating the result for player 2 as well. The results need to be consistent or logged as inconsistent results for ratings purposes. And a game between two players might not be in the same round for them both due to cross-round pairings,
One of the reasons we did not do this before now is that we re-sort the crosstable into score group order before posting it on MSA. That means their ‘pairing’ number could change from the internal report (as provided by the TD) to the re-ordered pairing numbers shown on MSA. (In other words, internally we might show a game between two players as a win by pairing number 27 over pairing number 32 but after resorting the event into score group order it could appear as a win by pairing number 14 against pairing number 19.)
Even the ratings department staff, which was trained on the process of updating results, makes mistakes doing this.
The best way to fix this might be a redesigned data structure, which is why the change to Leago is a good time to be considering this.
Another issue is when a player reports that their member ID was used in an event they didn’t play in and that report is accepted as valid. The proper way to fix this is to contact the TD/organizer and have them supply the correct ID, but there are events in the database where this wasn’t done properly, such as by blanking out the member ID or inserting an invalid one. As you may expect, this messes with ratings.
We have, on occasion, created a new ID for an otherwise unidentifiable player, because it is better to have that unknown player treated as an unrated player for ratings purposes than to use a wrong ID.
When I was considering the concept originally, I was imagining that the TD could simply re-open the existing event, then edit, validate, and submit as normal, making whatever corrections are deemed necessary or useful, and then have the edited event replace the prior one. I think that process would avoid the correction issues you brought up.
The pending event tables are organized differently from the rated event tables (the latter are based on a design that dates back to 1991) so this is more complicated than it sounds. (And we haven’t been storing the pending event tables indefinitely because that slows down the online editing process.)
There are also change made to events after they’re rated other than those that affect player IDs or results (such as changing the event name or updating the list of TDs) that would need to be kept.
There was also some legitimate concern among staff that some TDs might use a correction process to manipulate ratings, we’ve had ‘correction reports’ that seemed rather dubious. I think Leago’s suggestion of having an approval step (hopefully highlighting any changes) is a good one.
The only issue here is in storing the pending event table for some period of time for the corrections, it looks like, plus adding a review step. Really though if the event would have gone through originally with the edits I’m not sure what the need is for a correction review?
The initial validation/rating process includes steps designed to check for suspicious results, and we also do additional checking on events after they are rated. Both sets of checks have found events that resulted in complaints against TDs or affiliates.
Being able to tweak an event after it has been rated just seems to offer more opportunities for mischief if there aren’t supervisory controls. (My favorite example of mischief was a former TD who was using the ‘minor name change’ feature on the old membership system to change people’s middle name, then figured out if he changed a few letters at a time he could turn WILLIAM BROWN into SAMUEL GREEN.)
All post-rating event changes, regardless of who makes them or what is changed, should come through the same interface in the new system, right now that isn’t the case.
If I knew in 2004 what I know now, the ratings system programming would have had more than a few changes to the way data is organized, stored, maintained and processed. In discussions with Leago, they’ve understood a lot of the reasoning behind the current system and open to suggestions as to where and how to change things.
If you are saying the system is necessarily difficult to use because US Chess has been optimizing for data integrity I’d say you are optimizing for the wrong thing and also, given the quality of the data I’ve seen, Wow
I think ‘optimizing’ is difficult to define, some things are the way they are because of legacy issues, some because that’s how they were initially designed/programmed and changing them hasn’t been a high priority as long as they worked. And some things were clumsy because some things just are clumsy, or because organizational policies made it difficult to do them in a different/better way.
Moving to a totally new environment gives us an opportunity to re-envision things, and constructive criticism is always appreciated.
I’ve looked at tens of thousands of events over the last 20 years, and written many scripts looking for problems or data consistency issues. Sometimes what looks a bit suspicious turns out to be totally benign.
For a while we tried to look for events that looked like cherry-picking (not rating all the games in a section, a clear violation of US Chess policies), that turned up a lot of events, especially scholastic ones, that looked strange because they weren’t swiss or round robin events, there were some where over 50% of the potential games (rounds x players) were unplayed ones, those really do look suspicious online. (And we had more than a few TDs who tried to get around match rule restrictions, such as adding a game or two against a few other players to a match between two players.)
I don’t know the detailed timeline, but there should be ample opportunities in the next 6-9 months for TDs to test out the new system and offer suggestions.
You haven’t been around long, have you? The ratings submissions back in the day of the diskette and early on-line uploads were chock full of errors. An single ID typo makes a complete mess. (One player doesn’t get their rating updated, another (possibly in a different state) has a rating change without playing). Is it really that big a deal to be asked to double check your submission before it’s accepted?
I did a script to look for evidence of sandbagging (looking for highly improbable negative results). I went through 100’s of examples before finding one that could reasonably be interpreted as someone dumping games. The others were almost certainly ID typos. (Wrong state, wrong age…).
The one thing I did find was several players who did well in multiple Grand Prix events and not so well in multiple non-Grand Prix events, where presumably the prize money was less (or nonexistent.) But even that wasn’t at a high enough confidence level to take action, and there could have been multiple valid explanations. (If a high rated player takes a late round draw because he’s ready to go home and there isn’t much money on the line, is that sandbagging?)